Search

Background modeling is an important step for video processing applications. The varying illumination conditions deteriorate the visual data representation. The video image sequences with very existence of shadow effect deteriorates the overall visual quality. This existence if disregarded, it may cause major performance issues in crucial computer vision applications like segmentation, object detection and tracking. If this could be handled, the facilitation of computer vision tasks can be boosted. However, it is a challenging research problem to suppress shadows by computer. It is therefore crucial to suppress shadows and keep details of original image intact. The proposed work focuses on improved shadow suppression model to improve the subjective perception with enhancing image using Contrast Limited Adaptive Histogram Equalization (CLAHE) and image binarization using OTSU with YCbCr and HSV color space. The objective evaluation using entropy measure and subjective results have shown better result.

With an improved algorithm based on the Otsu method, this paper proposes solving the shadow region problem caused by non-uniform illumination. The recent image processing algorithms for handling the shaded area’s issue for the massive difference in the image grey value will not be practical. It is not easy to directly set the shaded area’s content to “black” based on this problem. This paper proposes improved threshold image segmentation for Non-uniform illumination based on the Otsu optimization Approach. It uses the local area threshold and window image pixel by sliding to enhance the image recognition rate of a shaded area. Experimental results verify that the proposed segmentation method has more robust adaptability to images with shadows. Compared with the classical global OTSU algorithm and genetic algorithm optimization, the image recognition rate is 96%.

The existence of shadows is difficult to avoid in images. Also, it will affect object recognition and image understanding. But on the other hand, shadow can provide information about the light source and object shape. Therefore, accurate shadow detection and removal can contribute to many computer vision tasks. However, even the same object, its shadow will vary greatly under different lighting conditions. Thus it is quite challenging to detect and remove shadows from images. Recent research always treated these two tasks independently, but they are closely related to each other actually. Therefore, we propose a multi-task adversarial generative networks (mtGAN) that can detect and remove shadows simultaneously. In order to enhance shadow detection and shadow removal mutually, a cross-stitch unit is proposed to learn the optimal ways to fuse and constrain features between multi-tasks. Also, the combination weight of multi-task loss functions are learned according to the uncertainty distribution of each task, which is not set empirically as usual. Based on these multi-task learning strategies, the proposed mtGAN can jointly achieve shadow detection and removal tasks better than existing methods. In experiments, the effectiveness of the proposed mtGAN is shown.

Augmented Reality is increasingly used for visualizing underground networks. However, standard visual cues for depth perception have never been thoroughly evaluated via user experiments in a context involving physical occlusions (e.g., ground) of virtual objects (e.g., elements of a buried network). We therefore evaluate the benefits and drawbacks of two techniques based on combinations of two well-known depth cues: grid and shadow anchors. More specifically, we explore how each combination contributes to positioning and depth perception. We demonstrate that when using shadow anchors alone or shadow anchors combined with a grid, users generate 2.7 times fewer errors and have a 2.5 times lower perceived workload than when only a grid or no visual cues are used. Our investigation shows that these two techniques are effective for visualizing underground objects. We also recommend the use of one technique or another depending on the situation.

The traditional multi-viewpoint rendering method for indoor scenes has poor lighting and shadow effects, resulting in too dark or bright indoor scene multi-viewpoint rendering. Therefore, an optimization method for indoor scene multi-viewpoint rendering based on binocular vision is proposed. This research plans light and shadow effects based on the visual relationship between point light sources and indoor scenes; sets rendering points based on binocular vision; and renders indoor scenes with multiple viewpoint angles. The simulation experiment results show that compared with the traditional rendering method, the light and shadow effect of the studied method is excellent, and the rendered indoor scene is suitable for light and dark.

Discusses the importance of shadow elimination in current target detection, expounds the basic principles of shadow formation in target detection, analyzes the key influence of shadow images on target detection, and summarizes the two aspects of shadow elimination in current target detection. There are two main algorithms: shadow elimination based on texture features and shadow elimination based on color space. The two types of shadow elimination in target detection are elaborated on the analytical digital model, and a summary analysis is given.

In order to solve the problems of degradation, uneven illumination, and shadows in the process of binarization of document images, two approaches were proposed. In this paper, many state of the art were deeply studied and an improved scheme were proposed. According to position and intensity information of each pixel in the document image, the image was mapped into 3D space. The visibility detection technology of 3D point cloud combined with convex hull calculation was used to extract background of the image and enhance the input image of any binarization algorithm. Based on this, the authors proposed another new binarization algorithm to remove shadows, in which does not need special illumination uniformity solution or shadow removal algorithm. The algorithm can generate binary images in one step. Document images with shadows were mapped into 3D space, the characteristic of the point cloud was distinctly hierarchical. The idea used its hierarchy in 3D space to calculate the set of points at convex hull several times, then selected the non-text points and segmented them from the text, and then obtained the binary image. The experimental results demonstrated the effectiveness of the two algorithms.

Target detection is one of the hotspots of image processing research. In the image, due to factors such as distance or light, it will affect the target detection result and increase the error detection rate. Moreover, the existing network training time is too long to meet the actual needs. In order to reduce the lack of light or shadow interference and other factors, based on the Faster R-CNN framework, this paper innovatively proposes a method to improve its feature network ResNet-101 to extract deep features of images.In order to shorten the running time, this paper introduces the region number adjustment layer to adaptively adjust the number of candidate regions selected by RPN during the training process. This paper conducts experiments on the PASCAL VOC data set. The experimental results show that the improved feature network model proposed has an accuracy improvement of 2% compared with the original feature network model. The results show that the target detection algorithm proposed in this paper has higher recognition accuracy than the original algorithm.

Chinese traditional opera is a valuable and fascinating heritage assert in the world as one of the most representative folk art in Chinese history. Its characteristic of ‘suppositionality’ in stage scenery provides a possibility of preservation of cultural heritage by digitization means, e.g., 3D Animation and Virtual Reality-based art show. In this novel digitization art form, the construction of virtual scenes is an important pillar--variety of created models should be accommodated to provide a vivid performance stage, including stage props and background. However, the generation of scenes based on traditional manual 3D virtual props modelling method is a tedious and strenuous task. In this paper, a novel shadow puppetry virtual stage scenes construction approach based on semantic and prior probability is proposed for the generation of compositional virtual scenes. First, primitive models based on semantics text segmentation and retrieval is provided for scene composition; and then, scene placement algorithm based on prior probability is conducive to assign these 3D models within virtual scene. This method is tested by generating the virtual performance stage for our shadow puppetry prototype system, within which various traditional art-specific 3D models are assembled. Its ease of use can assist artists to create visually plausible virtual stage without professional scene modelling skill. The user study indicates our approach’s effectiveness and its efficiency.

The detection and counting of airport passengers and baggage is an important task of airport management. When passengers arrive at the aircraft docking area from the terminal building through a shuttle bus and boarding on foot, the counting of people and baggage based on airport surveillance video is often affected by complicated climate, Dynamic background, cast shadows and other factors, the paper improves the VIBE (Visual Background Extractor) foreground detection algorithm, adds a threshold adaptive mechanism, can accurately extract the foreground image under the dynamic background of the airport, and uses the pixel distribution histogram matching method to detect ghosts Finally, based on the Gaussian mixture model, the moving target is removed from the shadow, and the moving target is accurately counted on this basis. The experimental results show that the method proposed in the paper is better than the traditional shadow removal algorithm based on traditional VIBE in removing the shadow cast by passengers.

The popularity of mobile terminals has generated massive moving objects with spatio-textual characteristics. A top-k spatial keyword query over moving objects (Top-k SKM query) returns the top-k objects, moving or static, based on a ranking function that considers spatial distance and textual similarity between the query and objects. To the best of our knowledge, there hasn’t been any research into the why-not questions on Top-k SKM queries. Aiming at this kind of why-not questions, a two-level index called Shadow and a three-phase query refinement approach based on Shadow are proposed. The first phase is to generate some promising refined queries with different query requirements and filter those unpromising refined queries before executing any promising refined queries. The second phase is to reduce the irrelevant search space in the level 1 of Shadow as much as possible based on the spatial filtering technique, so as to obtain the promising static objects, and to capture promising moving objects in the level 2 of Shadow as fast as possible based on the probability filtering technique. The third phase is to determine which promising refined query will be returned to the user. Finally, a series of experiments are conducted on three datasets to verify the feasibility of our method.

Shadow is an effective projection for presence of real objects, and has important roles in three-dimensional effects and reality of virtual space. Therefore avatar’s shadow would be able to have interactive effects with the avatar’s self in a virtual space. In our previous research, we reported that an auto-generated interactive avatar’s motion which is different from the talker’s own motion not always supports human communication. However, by using an avatar-shadow, talkers could communicate effectively without inconsistency between their own motions and their avatar’s motions in a virtual space, in which the avatar-shadow’s motions are expressed by combining the talker’s own motions with auto-generated interactive motions. Accordingly we propose an embodied entrainment system called “Avatar Twin” using shadow avatar. This system is constructed by talkers’ shadow avatars based on their own motions, and their avatars based on auto-generated entrained motions, in addition to the previous avatar-shadows. In previous studies, an avatar was a subject as a substitute for the user, and the main purpose was to investigate the effects of shadow support in communication. In this paper, we propose the master-slave inversed approach by swapping between measured and entrainment motions in shadow avatar communication system, and develop the system prototype.

Many of the existing documents are digitized using smart phone’s cameras. These are highly vulnerable to capturing distortions (perspective angle, shadow, blur, warping, etc.), making them hard to be read by a human or by an OCR engine. In this paper, we tackle this problem by proposing a conditional generative adversarial network that maps the distorted images from its domain into a readable domain. Our model integrates a recognizer in the discriminator part for better distinguishing the generated images. Our proposed approach demonstrates to be able to enhance highly degraded images from its condition into a cleaner and more readable form.

In this chapter, we propose a deep learning-based pipeline for filtering of internet-sourced images towards 3D reconstruction of heritage sites. The 3D reconstruction of heritage sites facilitates creation of virtual walk-through, digital museum and augmented reality. Using internet-sourced images for 3D reconstruction of heritage sites is challenging, as these images may contain blur, text, occlusion, shadow and many other noises. We propose to include pruning and selection of images in the pipeline to select a suitable set of images for 3D reconstruction. We propose a method for pruning of images using learning-based classification models to eliminate the contribution of unwanted images in 3D reconstruction. We also propose a method to select a suitable set of images using a combination of mean-shift and hierarchical clustering algorithms. We demonstrate the proposed pipeline by generating various 3D models of cultural heritage sites.

Dealing with the inconsistency between a foreground object and a background image is a challenging task in high-fidelity image composition. State-of-the-art methods strive to harmonize the composed image by adapting the style of foreground objects to be compatible with the background image, whereas the potential shadow of foreground objects within the composed image which is critical to the composition realism is largely neglected. In this paper, we propose an Adversarial Image Composition Net (AIC-Net) that achieves realistic image composition by considering potential shadows that the foreground object projects in the composed image. A novel branched generation mechanism is proposed, which disentangles the generation of shadows and the transfer of foreground styles for optimal accomplishment of the two tasks simultaneously. A differentiable spatial transformation module is designed which bridges the local harmonization and the global harmonization to achieve their joint optimization effectively. Extensive experiments on pedestrian and car composition tasks show that the proposed AIC-Net achieves superior composition performance qualitatively and quantitatively.

The shadow volume is utilized extensively for real-time rendering applications which includes updating volumes and calculating silhouette edges. Existing shadow volume methods are CPU intensive and complex occluders result in poor rendering efficiency. In this paper, we propose a hash-culling shadow volume algorithm that uses hash-based acceleration for the silhouette edge determination which is the most time-consuming processing in the traditional shadow volume algorithm. Our proposed method uses a hash table to store silhouette edge index information and thus reduces the time taken for redundant edge detection. The method significantly reduces CPU usage and improves algorithm time efficiency. Furthermore, for low hardware-level systems, especially embedded systems, it is still difficult to render dynamic shadows due to their high demand on the fill-rate capacity of graphics hardware. Our method has low hardware requirements and is easy to implement on PCs and embedded systems with real-time rendering performance with visual-pleasing shadow effects.

In the present scenario, most of the researchers are highly motivated to do new findings in the field of computer vision because of its tremendous applications such as video surveillance, human-machine interaction, traffic monitoring, human behavioral analysis, the guided missile, and military services, medical applications, and vehicle navigation. A typical video data frame comprises both foregrounds as well as background information. The pixel points that describe the target features in the region of interest are considered foreground information, and the rest of the feature points are treated as background information. Moving object detection plays a pivotal role in any kind of computer vision applications. Moving object detection is the process of identifying the class objects such as people, vehicles, toys, and human faces in the video sequences more precisely without background disturbances. In most cases, the existing moving object detection approaches concentrate only on the foreground information and frequently ignore the background information. As a result, trackers will be deviated away from the target and detect the non-foreground objects. Recently, several contributions have been proposed for moving object detection. However, the robustness and novelty are still challenging to achieve because of the complex environments, including illumination changes, rapid variations in target appearance, similar objects in the background, occlusions, target rotations, scaling, fast and abrupt motion changes, moving soft shadow, flat surface regions, and dynamic backgrounds. This book chapter introduces the concept of an intelligent video surveillance system and the major challenges involved in moving object detection. This chapter also deliberates the related backgrounds and their shortcomings in different perspectives. It also presents potential algorithms for efficient object detection and tracking. Finally, the book chapter is concluded with research opportunities pertaining to object detection and tracking.

Information extraction of shadow areas in hyperspectral images (HSIs) has always been a difficult problem in HSI processing. Dynamic stochastic resonance (DSR) theory has proved that the noise contained in the signal can enhance the strength of the original signal and improve the signal-to-noise ratio (SNR). And it has been applied in signal and image processing,communication and other fields. In this paper, DSR theory is introduced to the shadow enhancement in HSIs for the first time. The spatial and spectral dimensions of the shadow areas in a HSI could be enhanced by the DSR respectively. Then, the enhanced shadow should be fused with the other areas in the HSI. Finally, the fused image could be classified to explore the information in the HSI. The experimental result show that the DSR has promising prospect in the shadow enhancement in HSIs, and can help to improve the classification.