Search

Oil spills pose significant threats to marine and coastal ecosystems, biodiversity and local economies, necessitating efficient and accurate detection systems. Traditional detection methods, such as manual inspection and satellite imaging, are often resource-intensive and time consuming. This study addresses these challenges by developing a novel approach to enhance the quality and diversity of oil spill datasets. Several studies have mentioned that the quality and size of a dataset is crucial for training robust vision-based deep learning models. The proposed methodology combines advanced object extraction techniques with traditional data augmentation strategies to generate high quality and realistic oil spill images under various oceanic conditions. A key innovation in this work is the application of image blending techniques, which ensure seamless integration of target oil spill features into diverse environmental ocean contexts. To facilitate accessibility and usability, a Gradio-based web application was developed, featuring a user-friendly interface that allows users to input target and source images, customize augmentation parameters, and execute the augmentation process effectively. By enriching oil spill datasets with realistic and varied scenarios, this research aimed to improve the generalizability and accuracy of deep learning models for oil spill detection. For this, we proposed three key approaches, including oil spill dataset creation from an internet source, labeled oil spill regions extracted for blending with a background image, and the creation of a Gradio web application for simplifying the oil spill dataset generation process. [ABSTRACT FROM AUTHOR]

The detection of ship fires is a critical aspect of maritime safety and surveillance, demanding high accuracy in both identification and response mechanisms. However, the scarcity of ship fire images poses a significant challenge to the development and training of effective machine learning models. This research paper addresses this challenge by exploring advanced data augmentation techniques aimed at enhancing the training datasets for ship and ship fire detection. We have curated a dataset comprising ship images (both fire and non-fire) and various oceanic images, which serve as target and source images. By employing diverse image blending methods, we randomly integrate target images of ships with source images of oceanic environments under various conditions, such as windy, rainy, hazy, cloudy, or open-sky scenarios. This approach not only increases the quantity but also the diversity of the training data, thus improving the robustness and performance of machine learning models in detecting ship fires across different contexts. Furthermore, we developed a Gradio web interface application that facilitates selective augmentation of images. The key contribution of this work is related to object extraction-based blending. We propose basic and advanced data augmentation techniques while applying blending and selective randomness. Overall, we cover eight critical steps for dataset creation. We collected 9200 ship fire and 4100 ship non-fire images. From the images, we augmented 90 ship fire images with 13 background images and achieved 11,440 augmented images. To test the augmented dataset performance, we trained Yolo-v8 and Yolo-v10 models with "Fire" and "No-fire" augmented ship images. In the Yolo-v8 case, the precision-recall curve achieved 96.6% (Fire), 98.2% (No-fire), and 97.4% mAP score achievement in all classes at a 0.5 rate. In Yolo-v10 model training achievement, we got 90.3% (Fire), 93.7 (No-fire), and 92% mAP score achievement in all classes at 0.5 rate. In comparison, both trained models' performance is outperforming other Yolo-based SOTA ship fire detection models in overall and mAP scores. [ABSTRACT FROM AUTHOR]

Parallax processing and structure preservation have long been important and challenging tasks in image stitching. In this paper, an image stitching method based on sliding camera to eliminate perspective deformation and asymmetric optical flow to solve parallax is proposed. By maintaining the viewpoint of two input images in the mosaic non-overlapping area and creating a virtual camera by interpolation in the overlapping area, the viewpoint is gradually transformed from one to another so as to complete the smooth transition of the two image viewpoints and reduce perspective deformation. Two coarsely aligned warped images are generated with the help of a global projection plane. After that, the optical flow propagation and gradient descent method are used to quickly calculate the bidirectional asymmetric optical flow between the two warped images, and the optical-flow-based method is used to further align the two warped images to reduce parallax. In the image blending, the softmax function and registration error are used to adjust the width of the blending area, further eliminating ghosting and reducing parallax. Finally, by comparing our method with APAP, AANAP, SPHP, SPW, TFT, and REW, it has been proven that our method can not only effectively solve perspective deformation, but also gives more natural transitions between images. At the same time, our method can robustly reduce local misalignment in various scenarios, with higher structural similarity index. A scoring method combining subjective and objective evaluations of perspective deformation, local alignment and runtime is defined and used to rate all methods, where our method ranks first. [ABSTRACT FROM AUTHOR]

Image composition is a challenging image editing operation that targets the compositing of a new image by combining cropped regions from different images. A common application is background replacement of portrait images. Existing composition methods usually result in undesirable artifacts along the boundaries of pasted regions due to imprecise extraction. To address this problem, we propose a refined-mask guided multi-stream blending network (MGMB-Net), which consists of a multi-stream block (MSB), reconstructing the foreground boundary of composited image, mask-refine blocks (MRBs), providing better location information and refined-mask guided blocks (MGBs), guiding the reconstruction of composited image to eliminate boundary artifacts. MGMB-Net can generate more harmonious and realistic images by fusing multi-stream features and refining input coarse masks. We propose two data-generation methods to construct new datasets on image blending, and two evaluation metrics, PSNR (Peak signal-to-noise ratio)-Sobel and PSNR-Boundary, to decouple the performance of blending and harmonization. Compared with state-of-the-art methods, our method achieves the best performance on portrait images without additional refinements or prior information. Our code and model are available at https://github.com/Wekect/MGMB-Net. [ABSTRACT FROM AUTHOR]

In this paper, three approximate multiplier architectures are proposed: area-optimized approximate multiplier (AOM), power-optimized approximate multiplier (POM), and power- and area-optimized approximate multiplier (PAOM). These designs are implemented using speculative Han–Carlson adder and compressor-based multiplier blocks. Han–Carlson adder is used as the basic adder block in the final addition stage of all the three approximate multiplier designs. Different types of compressors (3:2, 4:2, 5:2, 6:2, 7:2, 8:2) are used for the implementation of the energy-efficient approximate multiplier blocks. All the simulations are performed on VIVADO design tool. Also, the designed multipliers are validated for image blending (an error-tolerant) application. The proposed power optimization approximate multiplier shows 0.86%, 10.54% PSNR improvement in comparison with area optimization approximate multiplier and power and area optimization approximate multiplier, respectively. [ABSTRACT FROM AUTHOR]

Background and objective: The availability of labeled data is crucial for training deep neural networks. However, in some cases, the available data is limited or unlabeled, which poses a significant obstacle in developing accurate models. Various approaches exist to address this issue, such as Image Augmentation, Transfer Learning, and GANs. However, these approaches often require a significant amount of training data or may not generate desired results. In this article, we present a novel method for generating synthetic images from very limited data using the ACGAN. Methods: We conducted experiments on a real dataset consisting of 198 ultrasound images of calcified and cystic thyroid gland nodules. We explored and improved different architectures and techniques in the Axillary Classifier Generative Adversarial Network (ACGAN) to generate high-quality synthetic images. To evaluate the generated images, we used the Fréchet Inception Distance (FID) test and human observation. Additionally, we developed an image blending method to generate larger images that simulate the output of an ultrasound device. To validate the accuracy of the merged images, a specialist doctor reviewed the generated data. Results: The modified ACGAN architecture successfully generated new synthetic images from limited data. The output images were assessed based on the image progress ratio with the FID test and human observation. Moreover, the Image blending method was successful in producing larger output images that mimic the nature of the ultrasound device output images. The final merged images were validated by a specialist doctor who confirmed their accuracy. Conclusions: Our method has significant implications for medical imaging, as it enables the generation of synthetic labeled data for training deep learning models, leading to better diagnostic accuracy and improved patient outcomes. This study provides a proof-of-concept for generating synthetic medical images from limited labeled data and can inspire future research in this area. [ABSTRACT FROM AUTHOR]

Image stitching involves combining multiple images of the same scene captured from different viewpoints into a single image with an expanded field of view. While this technique has various applications in computer vision, traditional methods rely on the successive stitching of image pairs taken from multiple cameras. While this approach is effective for organized camera arrays, it can pose challenges for unstructured ones, especially when handling scene overlaps. This paper presents a deep learning-based approach for stitching images from large unstructured camera sets covering complex scenes. Our method processes images concurrently by using the SandFall algorithm to transform data from multiple cameras into a reduced fixed array, thereby minimizing data loss. A customized convolutional neural network then processes these data to produce the final image. By stitching images simultaneously, our method avoids the potential cascading errors seen in sequential pairwise stitching while offering improved time efficiency. In addition, we detail an unsupervised training method for the network utilizing metrics from Generative Adversarial Networks supplemented with supervised learning. Our testing revealed that the proposed approach operates in roughly ∼1/7th the time of many traditional methods on both CPU and GPU platforms, achieving results consistent with established methods. [ABSTRACT FROM AUTHOR]

Constructing optical image time series for cropland monitoring requires a cloud removal method that accurately restores cloud regions and eliminates discontinuity around cloud boundaries. This paper describes a two-stage hybrid machine learning-based cloud removal method that combines Gaussian process regression (GPR)-based predictions with image blending for seamless optical image reconstruction. GPR is employed in the first stage to generate initial prediction results by quantifying temporal relationships between multi-temporal images. GPR predictive uncertainty is particularly combined with prediction values to utilize uncertainty-weighted predictions as the input for the next stage. In the second stage, Poisson blending is applied to eliminate discontinuity in GPR-based predictions. The benefits of this method are illustrated through cloud removal experiments using Sentinel-2 images with synthetic cloud masks over two cropland sites. The proposed method was able to maintain the structural features and quality of the underlying reflectance in cloud regions and outperformed two existing hybrid cloud removal methods for all spectral bands. Furthermore, it demonstrated the best performance in predicting several vegetation indices in cloud regions. These experimental results indicate the benefits of the proposed cloud removal method for reconstructing cloud-contaminated optical imagery. [ABSTRACT FROM AUTHOR]

Image blending is one of the alternative methods to fill temporal gaps in the monitoring of historical vegetation properties using continuous NDVI derived from Landsat 5 TM / 7 ETM+ and 8 OLI images. Frequent cloud occurrence in the tropical upstream catchment limits the use of image blending methods that allow to employment of a single pair base reference. This study aims to evaluate two image blending methods with nine input data configurations to select the most applicable one. Scatter plots and statistical indices such as ME, RMSE, model efficiency and structure similarity showed FSDAF outperforms STARFM in generating both synthetic Landsat 8 OLI NDVI and Landsat 5 TM / 7 ETM+ NDVI when employing unsupervised and supervised classification images, respectively, where both were applied along with MODIS NDVI 250 m v.005. When generating synthetic Landsat 5 TM / 7 ETM+ NDVI using AVHRR NDVI, both algorithms performed similarly. However, when considering the temporal over spatial variance ratio between base reference and predictor images, both algorithms performed almost similar when the value close to minimum. This study shows that selection of image blending algorithm with use single pair base reference image should consider input data configuration and temporal over spatial variance ratio.

In this paper, we consider the problem of single image marine snow denoising. Due to the complex underwater environment, the structure and statistical properties of marine snow noise are substantially different from those of the noises encountered in the atmosphere. To synthesize realistic noisy-clean image pairs for training, we propose an image blending-based noise synthesis method that can better simulate the marine snow without training deep networks. Specifically, the noise is first cropped from a real noisy image and pasted into a clean image to produce a composite image. Then, the Gaussian–Poisson Equation is employed to generate a well-blended image (synthetic noisy image). Furthermore, we introduce an attention-guided denoising network that leverages the location information of noise. The proposed network can detect the marine snow noise and then remove it guided by the estimated attention map. Extensive experiments on synthetic and real-world datasets demonstrate that our denoising network can effectively remove the marine snow noise, while preserving rich details of backgrounds. Other alternative marine snow synthesis approaches are also compared to show the superiority of our noise synthesis method in terms of visual quality and running time. [ABSTRACT FROM AUTHOR]

Nowadays image morphing has become one of the important techniques in applications that require a graphical representation of objects. Morphing tools have become very well known among users who work on multimedia applications such as art effects, virtual games, photo morphing, and social media, in addition to scientific and academic fields. There are many algorithms to apply morphing operations, including the basic and improved techniques, which share some essential stages, but vary in the algorithm details and the produced image qualities. Morphing techniques, in general, are based on image features and changing them through a warping process to produce another image or mixing two images to produce a new combined image. This paper provides an overview of different morphing techniques explaining how they work and discuss their features in some terms such as the morph visual quality, technical efficiency, and complexity, which can assist the researcher in the image morphing field to compare and identify morphing techniques that suit their working area. [ABSTRACT FROM AUTHOR]

The challenges of performance indices in highly complex computing have always been observed in terms of latency, throughput, number of gates, power, and yields. The use of CPU based vector processing and GPU parallel processing architecture has given ways to accelerate the computations but again lacks the scalability of parallel cores like Multiplier and Accumulator (MAC). With the evolution of reconfigurable hardware solutions and recent significant characterization in channel width (< 10 nm), power (zone-based partitions) and even cost of development, the FPGAs had come out of ages to be a most favorable choice not only for Proof of Concept (POC) or prototyping but as a product solution for many markets like medical, space and defense equipment. This work also proposes the FPGA based solution for hardware implementation of efficient computational techniques of de-convolution and image blending to cater the motion blurring problems in medical imaging. The active response time, taken during the real time image processing solutions, challenges the use case of many Image enhancement methods in the field. The proposed method and implementation scheme demonstrates the reduction on latency in motion de-blurring technique on hardware accelerated platform. [ABSTRACT FROM AUTHOR]

Virtual view synthesis, which generates novel views similar to the characteristics of actually acquired images, is an essential technical component for delivering an immersive video with realistic binocular disparity and smooth motion parallax. This is typically achieved in sequence by warping the given images to the designated viewing position, blending warped images, and filling the remaining holes. When considering 6DoF use cases with huge motion, the warping method in patch unit is more preferable than other conventional methods running in pixel unit. Regarding the prior case, the quality of synthesized image is highly relevant to the means of blending. Based on such aspect, we proposed a novel blending architecture that exploits the similarity of the directions of rays and the distribution of depth values. By further employing the proposed method, results showed that more enhanced view was synthesized compared with the well-designed synthesizers used within moving picture expert group (MPEG-I). Moreover, we explained the GPU-based implementation synthesizing and rendering views in the level of real time by considering the applicability for immersive video service.

Image compositing is the process of seamlessly inserting a portion of a source image into a target image to create a new desirable image. This work describes an image composition approach based on numerical differentiation utilizing the Laplacian operator. The suggested procedure uses the red-black strategy to speed up computations by using two separate relaxation factors for red and black nodes, as well as two accelerated parameters on a skewed grid. The Skewed Modified Two-Parameter Overrelaxation (SkMTOR) approach is a modification of the existing MTOR method. The SkMTOR has been used to solve numerous linear equations in the past, but its applicability in image processing has never been investigated. Several examples were used to test the suggested method in solving the Poisson equation for image composition. The results demonstrated that the image composition was successfully constructed using all six methods considered in this study. The six methods evaluated yielded identical images based on the similarity measurement results. In terms of computing speed, the skewed variants perform much quicker than their corresponding regular grid variants, with the SkMTOR showing the best performance.

Addition and multiplication are some of the most broadly adopted arithmetic operations in a wide range of applications. This paper proposes new structures of approximate multipliers to optimize the area, delay, and power without affecting the accuracy metrics. Multipliers and adders play a significant role in the functioning of any digital circuit or system. The overall performance of a processor highly depends on the speed of adders and the energy consumption. In this paper, two types of compact error-tolerant approximate adders are designed and used along with approximate 4:2 compressors to improvise the efficiency of the approximate multipliers. The proposed approximate multipliers show good results when compared to the existing structures in terms of area, delay, power, and accuracy. The approximate multipliers are applied to image sharpening and image multiplication applications. The error-tolerant adder's performance is evaluated in the practical domain using the image blending application. Peak signal-to-noise ratio (PSNR) performance and the structural similarity index metric (SSIM) are used to assess the modeled designs. The proposed approximate multipliers and adders exhibit better performance in terms of PSNR and SSIM and are found to be an optimized design to apply effectively in various error-tolerant image processing applications. [ABSTRACT FROM AUTHOR]

Increasing the efficiency of the quality control phase in industrial production lines through automation is a rapidly growing trend. In non-destructive testing, active thermography techniques are known for their suitability to allow rapid non-contact and full-field inspections. The robotic manipulation of the thermographic instrumentation enables the possibility of performing inspections of large components with complex geometries by collecting multiple thermographic images from optimal positions. The robotisation of the thermographic inspection is highly desirable to improve assessment speed and repeatability without compromising inspection accuracy. Although integrating a robotic setup for thermographic data capture is not challenging, the application of robotic thermography has not grown significantly to date due to the absence of a suitable approach for merging multiple thermographic images into a single presentation. Indeed, such an approach must guarantee accurate alignment and consistent pixel blending, which is crucial to facilitate defect detection and sizing. In this work, an innovative inspection platform was conceptualised and implemented, consisting of a pulsed thermography setup, a six-axis robotic manipulator and an algorithm for image alignment, correction and blending. The performance of the inspection platform is tested on a convex-shaped specimen with artificial defects, which highlights the potential of the new combined approach. This work bridges a technology gap, making thermographic inspections more deployable in industrial environments. The proposed fine image alignment approach can find applicability beyond thermographic non-destructive testing. [ABSTRACT FROM AUTHOR]

The hardware implementation of error-tolerant adders using the paradigm of approximate computing has considerably influenced the performance metrics, especially in applications that can compromise accuracy. The foundation for approximate processing is the inclusion of errors in the design to enhance the effectiveness and reduce the complexity. This work presents three base adders using the novel concept of error tolerance in digital VLSI design. The research is extended to construct nine variants of power and delay-efficient 16 and 32-bit error-tolerant carry select adders (CSLA). To attain optimization in power and delay, conventional CSLA is refined by substituting ripple carry adders (RCA) with the newly proposed selector unit to minimize the switching activity. The research work includes the power, area, and delay estimates of the design from synthesis using the gpdk-90 nm and gpdk-45 nm standard cell libraries. The proposed adders exhibit reduced delay, power dissipation, area, power delay product (PDP), energy delay product (EDP), and area delay product (ADP) compared to the existing approximate adders. The proposed adder is used in an image blending application. There is a significant improvement in the peak-signal-to-noise ratio (PSNR) in the blended image compared to the standard designs. [ABSTRACT FROM AUTHOR]

A numerical approach was implemented to precisely stitch together images from the same projector/camera that form a mosaic by regularly moving either the projector/camera or the scene/sample. Such an imaging approach is used, for example, in automated microscopy. The presence of optical distortions can lead to detrimental blurring artifacts in the overlaps. The present development identifies and corrects non-affine distortion functions using the gray-level conservation equation on reduced overlapping regions of adjacent images with sub-pixel accuracy. The present numerical development was first tested on synthetic images with known distortions to confirm that the algorithm is capable of detecting only non-affine distortions. Then, Digital Image Correlation (DIC) was applied to a pair of large laser scanning confocal microscope mosaics (121 images of 1024 × 1024 pixel 2 , i.e. , more than 100 MPixel) created using the proposed non-rigid stitching. The method aims to improve the quality of blended images after stitching using the sample pattern in the overlapped regions. This new numerical development significantly minimizes kinematic field artifacts due to lens distortion in overlapped regions. • A non-rigid stitching method with correction of optical distortions was proposed. • The algorithm estimates distortions using gray-levels in overlapping regions. • Minimal information were needed: small overlap, self-calibration, no extra images. • The blended image was improved reducing blurring in the overlaps. • The method outperforms classic stitching for sub-pixel kinematics analyses using DIC. [ABSTRACT FROM AUTHOR]

• A flexible orthoimage mosaicking method is developed considering the cloud coverage. • Dynamic variable patch is defined as the processing unit for orthoimage mosaicking. • This method can remove clouds and also relieve the impact of misalignments on the mosaic. • An area-weighted image blending method is presented to achieve smooth transition. Orthoimage mosaicking plays an important role in remote sensing applications, such as environment monitoring and protection. However, the cloud coverage in remote sensing images often results in substantial loss of information and degradation of data quality, bringing lots of challenges to orthoimage mosaicking. Therefore, this paper proposes a flexible multi-temporal orthoimage mosaicking method based on dynamic variable patches. Different from most existing pixel-based mosaicking methods, the proposed method defines patch as the minimal processing unit for orthoimage mosaicking, and the size and position of each patch are dynamic variable. Patches from different images are rearranged under the image texture consistency constraint to generate the mosaic image. In the proposed method, the initial positions of available patches of all images are determined first, then the initial size of each patch is determined by the cloud coverage and the effective area of image. Subsequently, considering the inevitable misalignments between different orthoimages, the texture similarity between patches is utilized to select the most suitable patch, and the size of the patch is also dynamically adjusted to make the misalignments within each patch controllable. Finally, considering the color inconsistency in the overlapping areas of neighboring images, an area-weighted image blending algorithm is also presented to achieve invisible seams and smooth transitions in the final mosaic. Not only does the proposed method produce mosaic images with minimal cloud coverage areas by excluding cloud pixels as much as possible, but it is also flexible for the mosaicking of images with varying degrees of misalignments, which is achieved by dynamically adjusting the size and position of each patch while maintaining texture consistency constraint. Both simulated experiments and real data experiments are carried out to verify the performance of the proposed method. The experimental results demonstrate that the proposed method can generate the mosaic image with higher spatial continuity and radiometric consistency. [ABSTRACT FROM AUTHOR]

Copyright of Frontiers of Information Technology & Electronic Engineering is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Background: The paucity of dark skin images in dermatological textbooks and atlases is a reflection of racial injustice in medicine. The underrepresentation of dark skin images makes diagnosing skin pathology in people of color challenging. For conditions such as skin cancer, in which early diagnosis makes a difference between life and death, people of color have worse prognoses and lower survival rates than people with lighter skin tones as a result of delayed or incorrect diagnoses. Recent advances in artificial intelligence, such as deep learning, offer a potential solution that can be achieved by diversifying the mostly light-skin image repositories through generating images for darker skin tones. Thus, facilitating the development of inclusive cancer early diagnosis systems that are trained and tested on diverse images that truly represent human skin tones. Objective: We aim to develop and evaluate an artificial intelligence--based skin cancer early detection system for all skin tones using clinical images. Methods: This study consists of four phases: (1) Publicly available skin image repositories will be analyzed to quantify the underrepresentation of darker skin tones, (2) Images will be generated for the underrepresented skin tones, (3) Generated images will be extensively evaluated for realism and disease presentation with quantitative image quality assessment as well as qualitative human expert and nonexpert ratings, and (4) The images will be utilized with available light-skin images to develop a robust skin cancer early detection model. Results: This study started in September 2020. The first phase of quantifying the underrepresentation of darker skin tones was completed in March 2021. The second phase of generating the images is in progress and will be completed by March 2022. The third phase is expected to be completed by May 2022, and the final phase is expected to be completed by September 2022. Conclusions: This work is the first step toward expanding skin tone diversity in existing image databases to address the current gap in the underrepresentation of darker skin tones. Once validated, the image bank will be a valuable resource that can potentially be utilized in physician education and in research applications. Furthermore, generated images are expected to improve the generalizability of skin cancer detection. When completed, the model will assist family physicians and general practitioners in evaluating skin lesion severity and in efficient triaging for referral to expert dermatologists. In addition, the model can assist dermatologists in diagnosing skin lesions. [ABSTRACT FROM AUTHOR]

Virtual view synthesis, which generates novel views similar to the characteristics of actually acquired images, is an essential technical component for delivering an immersive video with realistic binocular disparity and smooth motion parallax. This is typically achieved in sequence by warping the given images to the designated viewing position, blending warped images, and filling the remaining holes. When considering 6DoF use cases with huge motion, the warping method in patch unit is more preferable than other conventional methods running in pixel unit. Regarding the prior case, the quality of synthesized image is highly relevant to the means of blending. Based on such aspect, we proposed a novel blending architecture that exploits the similarity of the directions of rays and the distribution of depth values. By further employing the proposed method, results showed that more enhanced view was synthesized compared with the well‐designed synthesizers used within moving picture expert group (MPEG‐I). Moreover, we explained the GPU‐based implementation synthesizing and rendering views in the level of real time by considering the applicability for immersive video service. [ABSTRACT FROM AUTHOR]

Medical image-enhancing technology plays a significant role for processing and revealing discerning information from acquired images in many applications such as Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) which are frequently used for diagnosis and treatments in medical imaging. The clarity of these images become of great importance considering the details required to render diagnosis. The effects associated with blurred images in such applications is very challenging. The blurring effect is largely unavoidable due to the errors associated with capturing devices and natural motion in the human body. In this research, a method is proposed utilizing image blending approach to significantly reduce the effects of blur from an image through motion adaptive Point Spread Function (PSF). The proposed Efficient Image De-Blurring methods (EIDB) is realized using PSFs. And then get deblurred quality images from image de-blending image set in the alpha plane.

Image blending is one of the alternative methods to fill temporal gaps in the monitoring of historical vegetation properties using continuous NDVI derived from Landsat 5 TM / 7 ETM+ and 8 OLI images. Frequent cloud occurrence in the tropical upstream catchment limits the use of image blending methods that allow to employment of a single pair base reference. This study aims to evaluate two image blending methods with nine input data configurations to select the most applicable one. Scatter plots and statistical indices such as ME, RMSE, model efficiency and structure similarity showed FSDAF outperforms STARFM in generating both synthetic Landsat 8 OLI NDVI and Landsat 5 TM / 7 ETM+ NDVI when employing unsupervised and supervised classification images, respectively, where both were applied along with MODIS NDVI 250 m v.005. When generating synthetic Landsat 5 TM / 7 ETM+ NDVI using AVHRR NDVI, both algorithms performed similarly. However, when considering the temporal over spatial variance ratio between base reference and predictor images, both algorithms performed almost similar when the value close to minimum. This study shows that selection of image blending algorithm with use single pair base reference image should consider input data configuration and temporal over spatial variance ratio. [ABSTRACT FROM AUTHOR]

Approximate computing is gaining grip as a computing paradigm for computer vision, data analytics, and image/signal processing applications. In the era of real-time applications, approximate computing plays a significant role. In many computers including digital signal processors (DSP) and a microprocessor, adders are the main element for the implementation of signal processing applications and digital circuit design. The major problem for addition is the propagation delay in the carry chain. As the bit length of the input operand increases, the length of the carry chain increases. To address the carry propagation problem in digital systems, the most efficient adder architectures for VLSI implementation are classified as a parallel prefix adder (PPA) structure. In this paper, a novel methodology to implement and synthesize different adders (non-speculative and speculative) for any ASIC-based system is proposed. The proposed hybrid Han-Carlson and Kogge-stone speculative adders show improved performance (low power and delay) over the state-of-the-art approximate adders. If the approximation fails, then the proposed efficient error correction technique is activated. The proposed speculative H_C adder results in a 23.79% speed improvement over the proposed K_S adder, and 23.86% of energy is saved. The proposed architectures were synthesized for an operand bit length of 16 bits. Finally, the proposed adder is validated for an error-tolerant image processing application resulting in 41.2 dB PSNR. [ABSTRACT FROM AUTHOR]

Summary: Approximate computing‐based hardware is directed towards reduction in area and power consumption; it finds better suitability in various image processing applications. In order to achieve low complexity, approximate techniques are applied in the arithmetic units which provide promising results in enhancing the performance of DSP processors with optimal degradation on its accuracy. In this paper, two approximate full adder designs are proposed, namely, the first design as proposed approximate full adder 1 (PAA1) and the second design as proposed approximate full adder 2 (PAA2). In both the designs, the approximation on the Boolean expressions is carried out to reduce the hardware complexity with the significant tolerance on the error rate. Also, a multi‐bit approximate carry select adder is proposed with the utilization of PAA1 and PAA2, respectively. The performance of the proposed adders is analyzed in terms of parameters like gate count, structural synthesis, error, and also in terms of image metrics. The proposed and existing designs are synthesized using Synopsys Design Compiler with TSMC 65 nm technology for better comparison. Synthesis results indicate that the proposed 16‐bit approximate carry select adder (CSLA) shows a significant reduction of 58% in area delay product and 70% in power delay product, in comparison with the conventional CSLA. The image metrics results also validate that the proposed adder with highest peak signal‐to‐noise ratio is highly adoptable for image processing applications. [ABSTRACT FROM AUTHOR]

In the field of autonomous driving technology, both daytime and nighttime scenes are common. However, due to the poor illumination and difficulty in manual annotation of nighttime images, semantic segmentation of nighttime scenes is more challenging compared to daytime scenes. Therefore, achieving significant progress in nighttime semantic segmentation would greatly enhance the effectiveness of the application of autonomous driving scenarios. In this work, we investigate the problem of domain-adaptive nighttime semantic segmentation (DANSS). The problem aims to learn semantic segmentation in nighttime scenes by leveraging a labeled Cityscapes dataset and unlabeled but roughly aligned day-night image pairs. To address this, we propose an Align and Fusion Network (AAFnet), a network for adaptive domain nighttime semantic segmentation. AAFnet utilizes a novel DAFormer as the backbone to separately compute features for dynamic and static objects during the training process. It also employs methods such as small category sampling and image blending to improve learning effectiveness. Additionally, we propose utilizing local image patches to enhance the results during the testing process. Experimental results demonstrate a significant improvement of 10.4% compared to the previous method, DANNet. Extensive experiments conducted on the Dark Zurich dataset and Nightdriving dataset validate the effectiveness of the proposed approach. Our method outperforms previous backbones and achieves top-ranking results. • an image alignment module for the semantic segmentation network. • an alignment network that utilizes static loss to enhance the supervision capability of static objects in nighttime images. • an image blending method that focus on improving the segmentation performance of dynamic object categories. • combineing local and global image features during the testing phase, • Extensive experiments demonstrate the effectiveness of our framework [ABSTRACT FROM AUTHOR]

Summary: Imaging surface deformation of a coupon specimen in microtensile testing with an optical microscope presents challenges due to the narrow depth of field (DoF) of optical microscopes. Materials being heterogeneous at microscopic length scale, the sample surface deforms into a complex 3D surface texture, evolving continuously as the loading increases. Because of the narrow DoF, the region that is in focus within the field of view (FoV) decreases substantially in size with the increasing out‐of‐plane heterogeneous deformation. To address this challenge, a method based on image blending and stabilisation of the captured image frames is proposed. Image blending combines the partial regions that are in focus from a set of successive image frames captured at different working distances from the object surface plane to construct a single image that has a large part of the FoV in focus. The blended images are then obtained at different levels of macroscopic strains, that is the global homogeneous strain, in order to characterise the evolution of the heterogeneous deformation. The image stabilisation removes any misalignments of the blended images by spatially realigning them choosing a common feature as a reference point. The validation of the proposed method with conventionally and additively manufactured stainless steel 316L (SS 316L) specimens demonstrates excellent improvement in image quality. Almost 100% of the FoV is maintained in focus regardless of the amount of out‐of‐plane heterogeneous deformation caused during tensile testing, which is quite remarkable for optical microscopy imaging. Consequently, the blended and stabilised images enhanced the accuracy of digital image correlation (DIC). Time‐lapse videos of the deformation generated using these images captured the evolution of the slip bands and their transmission through twinning boundaries in the stainless steel microstructure. Overall, this study demonstrates the feasibility of using image‐processing techniques to advance optical microscopy to image complex 3D surfaces evolving with time. [ABSTRACT FROM AUTHOR]

The traditional visualization pipeline of cloud-motion wind(CMW) has considerable computational redundancy.This paper proposed a fast wind visualization pipeline,which could directly generate a vivid large-scale wind situation video for human eyes from any raw satellite image sequence.Guiding by sparse pixel block-matching for cloud motion estimation in satellite image sequence,the proposed algorithm warped and blended multi-frame noised textures by a warping mesh derived from the matching results.Without any complicated optical flow calculation and additional flow visualization post-processing,this compact visualization pipeline was more convenient in programming and less dependent on computing hardware.Experimental results show that the proposed straightforward algorithm can generate an easily identifiable and time-consistent dynamic visualization video of CMW directly from infrared,visible or water vapor satellite data.It is also more robust to the irregular dissipation and generation of cloud. [ABSTRACT FROM AUTHOR]

Face Recognition is a computer application that is capable of detecting, tracking, identifying or verifying human faces from an image or video captured using a digital camera. Although lot of progress has been made in domain of face detection and recognition for security, identification and attendance purpose, but still there are issues hindering the progress to reach or surpass human level accuracy. These issues are variations in human facial appearance such as; varying lighting condition, noise in face images, scale, pose etc. This research paper presents a new method using Local Binary Pattern (LBP) algorithm combined with advanced image processing techniques such as Contrast Adjustment, Bilateral Filter, Histogram Equalization and Image Blending to address some of the issues hampering face recognition accuracy so as to improve the LBP codes, thus improve the accuracy of the overall face recognition system. Our experiment results show that our method is very accurate, reliable and robust for face recognition system that can be practically implemented in real-life environment as an automatic attendance management system.

Forward-looking sonar (FLS) has been gaining attractions in the realm of near-bottom, close-range underwater inspections, owing to its high-resolution and rapid framerate capabilities. Although automatic target recognition (ATR) algorithms have been tentatively employed for object detection tasks, the necessity for human oversight remains vital, particularly in sensitive areas. A comprehensive FLS mosaic encapsulating all relevant information is highly desired to aid experts in managing an extensive array of perception data. Yet, prior research has often presumed that FLS operates under an ideal system configuration, assuming optimal sonar imaging setups and the availability of precise positioning data. Without these assumptions, intra-frame and inter-frame artifacts would emerge, deteriorating the quality of the resultant mosaic by obscuring essential information. In this study, we propose an innovative blending method specifically for FLS mosaicing aiming preserve important information. We first develop a long–short time sliding window (LST-SW) to rectify the local statistics of raw sonar images. These statistics then serve to create a global variance map (GVM). This step helps to emphasize the useful information contained in images in the blending phase by classifying the informative and featureless pixels, thereby enhancing the quality of final mosaic. This approach is substantiated through data collected in real-world scenarios. The results show that our method can preserve more details in FLS mosaics for human inspection purposes in practice. • Analyze the intra- and inter-frame artifacts of forward-looking sonar. • Develop a method to ameliorate the influence from these artifacts on the mosaic. • The method is validated through extensive field experiments. [ABSTRACT FROM AUTHOR]

Existing image dehazing algorithms typically rely on a two-stage procedure. The medium transmittance and lightness are estimated in the first stage, and the scene radiance is recovered in the second by applying the simplified Koschmieder model. However, this type of unconstrained dehazing is only applicable to hazy images, and leads to untoward artifacts in haze-free images. Moreover, no algorithm that can automatically detect the haze density and perform dehazing on an arbitrary image has been reported in the literature to date. Therefore, this paper presents an automated dehazing system capable of producing satisfactory results regardless of the presence of haze. In the proposed system, the input image simultaneously undergoes multiscale fusion-based dehazing and haze-density-estimating processes. A subsequent image blending step then judiciously combines the dehazed result with the original input based on the estimated haze density. Finally, tone remapping post-processes the blended result to satisfactorily restore the scene radiance quality. The self-calibration capability on haze conditions lies in using haze density estimate to jointly guide image blending and tone remapping processes. We performed extensive experiments to demonstrate the superiority of the proposed system over state-of-the-art benchmark methods.

Being able to cover a wide range of views, pan-tilt-zoom (PTZ) cameras have been widely deployed in visual surveillance systems. To achieve a global-view perception of a surveillance scene, it is necessary to generate its panoramic background image, which can be used for the subsequent applications such as road segmentation, active tracking, and so on. However, few works have been reported on this problem, partially due to the lack of benchmark dataset and the high complexity of panoramic image generation of PTZ cameras. In this paper, we build, for the first time to our best knowledge, a benchmark PTZ camera dataset with multiple views, and derive a complete set of panoramic transformation formulas for PTZ cameras. We further propose a fast multi-band blending method to address the efficiency issue in panoramic image fusion and mosaicing. Some related panoramic transformations are also developed, such as cylindrical and overlooking transformations. Our proposed approach exhibits impressive accuracy and efficiency in PTZ panorama generation as well as panoramic image mosaicing. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Image & Graphics is the property of Editorial Office of Journal of Image & Graphics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Foggy images suffer from low contrast and poor visibility problem along with little color information of the scene. It is imperative to remove fog from images as a pre-processing step in computer vision. The Dark Channel Prior (DCP) technique is a very promising defogging technique due to excellent restoring results for images containing no homogeneous region. However, having a large homogeneous region such as sky region, the restored images suffer from color distortion and block effects. Thus, to overcome the limitation of DCP method, we introduce a framework which is based on sky and non-sky region segmentation and restoring sky and non-sky parts separately. Here, isolation of the sky and non-sky part is done by using a binary mask formulated by floodfill algorithm. The foggy sky part is restored by using Contrast Limited Adaptive Histogram Equalization (CLAHE) and non-sky part by modified DCP. The restored parts are blended together for the resultant image. The proposed method is evaluated using both synthetic and real world foggy images against state of the art techniques. The experimental result shows that our proposed method provides better entropy value than other stated techniques along with have better natural visual effects while consuming much lower processing time.