Search

Context: Research is emphasized as a critical component of Accreditation Council for Graduate Medical Education (ACGME) Osteopathic Recognition (OR) criteria, yet there remains a deficit of osteopathic contributions to the literature. Scholar 12 combines discrete research development tools into an interactive application and blog forum that guides students from research team formation with an agreed-upon query to a scholarly product and presentations. Objective: This study aims to evaluate a beta test of Scholar 12 in developing a scholarly culture within medical school education. Methods: An unblinded prospective cohort beta trial by 6 osteopathic medical students across different campuses provided feedback for improvement measures and self-assessed research skill competency before and after completing Scholar 12 on an accelerated time frame. The pre- and post-Scholar 12 surveys scored 12 skills based on learning objectives for each unit on a 5-point Likert scale. Results: The composite results from self-assessments of 6 medical students demonstrate a statistically significant improvement in research skill familiarity by the completion of Scholar 12 (p Conclusions: The osteopathic profession has opportunity to advance clinical practice and fulfill ACGME OR initiatives with evidence-based medical research. Scholar 12 is a foundational educational tool and aims to engage medical students, residents, and attendings with scholarly work, regardless of experience level. The present survey provides a preliminary measure of the efficacy of Scholar 12 in improving medical students’ knowledge of creating new scholarly work. General feedback has been communicated to the application developer and editorial staff for improvement measures before the 2020 nationwide launch. Despite the statistical significance of these students’ self-reported progress, additional beta trials; blinded, long-term evaluation of students’ and mentors’ productivity as a result of this research learning tool; and controlled comparison to other research development programs are warranted. Scholar 12 is designed to accommodate students’ academic obligations with a convenient, virtual tool to learn the research process on a flexible schedule, in order to meet generational needs.

Due to the emergence of the era of big data, cross-modal learning have been applied to many research fields. As an efficient retrieval method, hash learning is widely used frequently in many cross-modal retrieval scenarios. However, most of existing hashing methods use fixed-length hash codes, which increase the computational costs for large-size datasets. Furthermore, learning hash functions is an NP hard problem. To address these problems, we initially propose a novel method named Cross-modal Variable-length Hashing Based on Hierarchy (CVHH), which can learn the hash functions more accurately to improve retrieval performance, and also reduce the computational costs and training time. The main contributions of CVHH are: (1) We propose a variable-length hashing algorithm to improve the algorithm performance; (2) We apply the hierarchical architecture to effectively reduce the computational costs and training time. To validate the effectiveness of CVHH, our extensive experimental results show the superior performance compared with recent state-of-the-art cross-modal methods on three benchmark datasets, WIKI, NUS-WIDE and MIRFlickr.

This study aimed to report on a method for injecting botulinum toxin type A into the hamstring muscles combined with periarticular injection in total knee arthroplasty (TKA) patients. We enrolled patients who underwent elective unilateral TKA at our hospital from February 2021 to December 2021 and administered botulinum toxin type A hamstring muscle injection combined with periarticular injection. We established and reported a detailed method for this combined approach, which could provide an alternative analgesic regimen after TKA in clinical practice.

Trial design: The present study was designed to investigate the safety and effectiveness of iced normal saline combined with cocktail perfusion during total knee arthroplasty (TKA). This was a random, double-blind, parallel-group study conducted in China. Methods: Seventy patients undergoing unilateral total knee replacements were assessed in the present study. Among them, sixty patients with confirmed primary knee osteoarthritis in stage IV were recruited and divided into three groups randomly, three different intro-operative articular cavity perfusion treatments were given according to the randomized and controlled rule. One way ANOVA analysis on visual analogue scale (VAS) score, functional recovery, drainage, and edema of the affected limb were performed to assess the efficiency of the treatment in the following three days after the operation. The participants, care givers, and those assessing the outcomes were blinded to group assignment.Results: Postoperative drainage in group A (n=20) and B (n=20) reduced significantly (P0.05). The VAS score of group B was significantly lower than in group A(PConclusions: Intra-operation articular cavity perfusion therapy with iced normal saline combined with cocktail perfusion therapy can greatly reduce the early inflammation, contributing to the better rehabilitation of TKA.Trial registration: The present study was retrospectively registered on ClinicalTrials. Gov with the identifier NCT05204056 (27/11/2021).

Diesel exhaust particles (DEPs) are major constituents of air pollution and associated with numerous oxidative stress-induced human diseases. In vitro toxicity studies are useful for developing a better understanding of species-specific in vivo conditions. Conventional in vitro assessments based on oxidative biomarkers are destructive and inefficient. In this study, Raman spectroscopy, as a non-invasive imaging tool, was used to capture the molecular fingerprints of overall cellular component responses (nucleic acid, lipids, proteins, carbohydrates) to DEP damage and antioxidant protection. We apply a novel data visualization algorithm called PHATE, which preserves both global and local structure, to display the progression of cell damage over DEP exposure time. Meanwhile, a mutual information (MI) estimator was used to identify the most informative Raman peaks associated with cytotoxicity. A health index was defined to quantitatively assess the protective effects of two antioxidants (resveratrol and mesobiliverdin IXα) against DEP induced cytotoxicity. In addition, a number of machine learning classifiers were applied to successfully discriminate different treatment groups with high accuracy. Correlations between Raman spectra and immunomodulatory cytokine and chemokine levels were evaluated. In conclusion, the combination of label-free, non-disruptive Raman micro-spectroscopy and machine learning analysis is demonstrated as a useful tool in quantitative analysis of oxidative stress induced cytotoxicity and for effectively assessing various antioxidant treatments, suggesting that this framework can serve as a high throughput platform for screening various potential antioxidants based on their effectiveness at battling the effects of air pollution on human health.

Visual trackers using deep neural networks have demonstrated favorable performance in object tracking. However, training a deep classification network using overlapped initial target regions may lead an overfitted model. To increase the model generalization, we propose an appearance variation adaptation (AVA) tracker that aligns the feature distributions of target regions over time by learning an adaptation mask in an adversarial network. The proposed adversarial network consists of a generator and a discriminator network that compete with each other over optimizing a discriminator loss in a mini-max optimization problem. Specifically, the discriminator network aims to distinguish recent target regions from earlier ones by minimizing the discriminator loss, while the generator network aims to produce an adaptation mask to maximize the discriminator loss. We incorporate a gradient reverse layer in the adversarial network to solve the aforementioned mini-max optimization in an end-to-end manner. We compare the performance of the proposed AVA tracker with the most recent state-of-the-art trackers by doing extensive experiments on OTB50, OTB100, and VOT2016 tracking benchmarks. Among the compared methods, AVA yields the highest area under curve (AUC) score of 0.712 and the highest average precision score of 0.951 on the OTB50 tracking benchmark. It achieves the second best AUC score of 0.688 and the best precision score of 0.924 on the OTB100 tracking benchmark. AVA also achieves the second best expected average overlap (EAO) score of 0.366, the best failure rate of 0.68, and the second best accuracy of 0.53 on the VOT2016 tracking benchmark.

Deep learning for image caption generation makes great progress in the field of natural images. However, there are still lack of effective methods for detailed analysis and automatic description of diseases content information in ultrasound image understanding. In order to find the location of focus areas, and understand the content of focus areas conveniently, we propose a novel method of ultrasound image captioning generation based on region detection. The method simultaneously detects and encodes the focus areas in ultrasound images, then utilizes the LSTM to decode the encoding vectors and generate annotation text information to describe the diseases content information in ultrasound images. The experimental results show that the method can accurately detect the location of the focus area, and also improves 1% the scores of BLEU-1, BLEU-2 with less parameters and running time, which compared with the full-size-image captioning model for ultrasound images.

Objective To determine the value of scintillation sign from Magnetic Resonance (MR) in assessing the tear of rotator cuff before operation. Method Preoperative MRI from 52 cases were system reviewed, two independent imaging physicians were invited to read the results. The results were further confirmed by the interoperative observations from the arthroscope. Statistical analysis was performed to estimate the sensitivity and specificity of scintillation sign. Result The specificity and sensitivity of the scintillation sign are relatively high stable, when used between different imaging physician or the same imaging physician at different times. Conclusion Scintillation sign is a very meaningful sign in identifying the L-shape tear with a high sensitivity and specificity, which is helpful for a better preoperative plan for the rotator cuff repair.

Ultrasound imaging is commonly used for diagnosing breast cancers since it is non-invasive and inexpensive. Breast ultrasound (BUS) image classification is still a challenging task due to the poor image quality and lack of public datasets. In this paper, we propose novel Neutrosophic Gaussian Mixture Models (NGMMs) to more accurately classify BUS images. Specifically, we first employ a Deep Neural Network (DNN) to extract features from BUS images and apply principal component analysis to condense extracted features. We then adopt neutrosophic logic to compute three probability functions to estimate the truth, indeterminacy, and falsity of an image and design a new likelihood function by using the neutrosophic logic components. Finally, we propose an improved Expectation Maximization (EM) algorithm to incorporate neutrosophic logic to reduce the weights of images with high indeterminacy and falsity when estimating parameters of each NGMM to better fit these images to Gaussian distributions. We compare the performance of the proposed NGMMs, its two peer GMMs, and three DNN-based methods in terms of six metrics on a new dataset combining two public datasets. Our experimental results show that NGMMs achieve the highest classification results for all metrics.

We propose a technique to generate "visual explanations" for interpretability of volumetric-based 3D object detection networks. Specifically, we use the average pooling of weights to produce a Sparse Activation Map (SAM) which highlights the important regions of the 3D point cloud data. The SAMs is applicable to any volumetric-based models (model agnostic) to provide intuitive intermediate results at different layers to understand the complex network structures. The SAMs at the 3D feature map layer and the 2D feature map layer help to understand the effectiveness of neurons to capture the object information. The SAMs at the classification layer for each object class helps to understand the true positives and false positives of each network. The experimental results on the KITTI dataset demonstrate the visual observations of the SAM match the detection results of three volumetric-based models.

Detection of traffic signs and light poles using light detection and ranging (LiDAR) data has demonstrated a valid contribution to road safety improvements. In this study, the authors propose a fast and reliable method, which can identify various traffic signs and light poles in mobile LiDAR data. Specifically, they first use the surface reconstruction algorithm to extract the normal vectors of the points as one of the characteristic features and apply k-means on the characteristic features of the points to automatically segment the data into road or non-road points. They then employ sliding cuboids to search for high-elevated objects that are located near the borders and on top of the road points. They further employ the random sample consensus algorithm to remove outliers and keep the points that fall on the perpendicular planes to the road trajectory. Finally, they introduce a modified seeded region growing algorithm to remove noisy points and incorporate the shape information to reject the false objects. A set of extensive experiments have been carried out on the datasets that are captured by Utah Department of Transportation from I-15 highway. The results demonstrate the robustness of the proposed method in detecting almost all traffic signs and light poles.

For the traditional locally resonant beams there always attached the one-stage local resonator and result that the lower band gap the heavier the scattering ring. In order to resolve this problem, the flexural vibration band gap in an Euler beam with periodically arranged lightweight multistage local resonators was theoretically investigated using the transfer matrix method based on discretization of lumped mass. The present method considered a distributed spring constant, which showed fast convergence with less computational requirements. A finite element method was then employed to calculate the frequency response function of a finite sample simultaneously, which demonstrated that the results calculated using the proposed method were closer to the simulation results than those obtained using the traditional transfer matrix method. The study found that, under the same additional mass, the lightweight multistage structure had much lower beginning frequency than one-stage structure, and the total width of the gaps was basically the same. In addition, a simplified model of the beginning frequency of gaps was proposed, and the effect of scattering density on the model precision was further explored numerically. The results show that the lower scattering density, the more important the role of the rubber mass and the higher precision of the simplified model.

Ultrasound imaging is one of the most commonly used diagnostic tools to detect and classify abnormalities of the women breast. Automatic ultrasound image segmentation provides radiologists a second opinion to increase diagnosis accuracy. Deep neural networks have recently been employed to achieve better image segmentation results than conventional approaches. In this paper, we propose a novel deep learning architecture, a Multi-Scale Self-Attention Network (MSSA-Net), which can be trained on small datasets to explore relationships between pixels to achieve better segmentation accuracy. Our MSSA-Net integrates rich local features and global contextual information at different scales and applies self-attention to multi-scale feature maps. We evaluate the proposed MSSA-Net on three public breast ultrasound datasets and compare its performance with six state-of-the-art deep neural network-based approaches in terms of five metrics. MSSA-Net achieves best overall segmentation results and improves the second best approach by 1.21% for Jaccard Index (JI) and 0.94% for Dice’s Coefficient (DSC).

Facial Expression Recognition (FER) has demonstrated remarkable progress due to the advancement of deep Convolutional Neural Networks (CNNs). FER's goal as a visual recognition problem is to learn a mapping from the facial embedding space to a set of fixed expression categories using a supervised learning algorithm. Softmax loss as the de facto standard in practice fails to learn discriminative features for efficient learning. Center loss and its variants as promising solutions increase deep feature discriminability in the embedding space and enable efficient learning. They fundamentally aim to maximize intra-class similarity and inter-class separation in the embedding space. However, center loss and its variants ignore the underlying extreme class imbalance in challenging wild FER datasets. As a result, they lead to a separation bias toward majority classes and leave minority classes overlapped in the embedding space. In this paper, we propose a novel Discriminant Distribution-Agnostic loss (DDA loss) to optimize the embedding space for extreme class imbalance scenarios. Specifically, DDA loss enforces inter-class separation of deep features for both majority and minority classes. Any CNN model can be trained with the DDA loss to yield well separated deep feature clusters in the embedding space. We conduct experiments on two popular large-scale wild FER datasets (RAF-DB and AffectNet) to show the discriminative power of the proposed loss function.

Image registration is a viable task in the field of computer vision with many applications. When images are captured under different spectrum conditions, a challenge is imposed on the task of registration. Researchers carefully handcraft a local module insensitive to illumination changes across cross-spectral image pairs to tackle this challenge. We, in this paper, develop an optimized feature-based approach Single Instance Phase Congruency Feature Extractor (SIPCFE) to tackle the problem of natural cross-spectral image registration. SIPCFE uses the phase information of an image pair to quickly identify and describe reliable keypoints that are insensitive to illumination. It then employs a sequence of outlier removal processes to find the matching feature points accurately and the Direct Linear Transformation to estimate the geometric transformation to align the image pair. We extensively study the proposed approach for every module in the system to give more insights into the challenges. We benchmark our proposed method and other state-of-the-art feature-based methods developed for cross-spectral imagery on three datasets with various settings and image contents. The comprehensive analysis of cross-spectral registration results of natural images demonstrates that SIPCFE achieves up to 47.24%, 14.29%, and 12.45% accuracy improvement on the first, second, and third dataset, respectively, over the second best registration method in the benchmark.

Illumination variations significantly affect the performance of face recognition systems. This paper presents a multi-scale method based on the maximum response (MR) filter bank and the gradient of faces. The proposed method first scales the face image using a simple log function to expand darker pixels and compress brighter pixels. It then effectively employs a subset of the MR filter bank to enhance edges and partially reduce illumination. Finally, it applies an enhanced multi-scale Gradientface method, which increases discriminating abilities and captures different characteristics of the face image to produce illumination invariant feature representation. Our extensive experiments on four closed-universe face databases and one open-universe database show the proposed method achieves the best recognition accuracy when comparing with 14 recently proposed state-of-the-art methods and its four variant methods. Our evaluations using receiver operating characteristic (ROC) curves on the four closed-universe face databases and precision and recall (PR) curves on the open-universe face database also verify the proposed method has the best verification and discrimination ability compared with other peer methods.

In this paper, we propose a fast and reliable track-before-detect approach to simultaneously detect, track, and identify an unknown and variable number of resident space objects (RSOs) without any prior information and any explicit detection, which leads to better space domain awareness. Specifically, we use the point spread function concept to propose a separable likelihood function as the observation model in the random finite set-based multi-Bernoulli filtering framework. This framework clearly distinguishes RSOs from any counterfeit objects and detects and tracks them immediately after their respective appearance in background cluttered telescope imagery data. The extensive experimental results on the TAOS dataset demonstrate the robustness of the proposed method in detecting and tracking RSOs with the average optimal subpattern assignment localization error less than 2 pixels in image sequences with the signal to noise ratio as low as 9 dB and under the conditions of varying illumination and occlusion.

We report here the development of transparent and strong polymer composites reinforced by unmodified cellulose nanofibrils (CNFs) with a Pickering emulsion gelation strategy. The CNFs entangle and firmly stabilize on the surface of emulsion droplets containing polymethyl methacrylate (PMMA) solution, leading to the gelation of the emulsions. CNFs/PMMA composites were generated via vacuum filtration and solvent washing of the gel and a subsequent two-step hot pressing. The composites contained a unique self-assembled two-tier hierarchy of CNFs networks and demonstrate promising transparency, tensile strength, flexibility, and an extremely low thermal expansion. Remarkably, these properties are highly tunable with varying the concentration of CNFs and the volume ratio of the water to oil phase. This work offers a facile route to realize the well dispersion of unmodified CNFs in hydrophobic polymer matrix and achieve high performance of polymeric materials reinforced by CNFs.

Affective interaction in tutoring environments has been of great interest among several researchers in this community, which has spurred the development of various systems to capture learners’ emotional states. Young children are one of the biggest learner groups in digital learning environments, but these studies have rarely targeted them. Our current study leverages computer vision and deep learning to analyze young childrens’ learning-related affective states. We developed an effective recognition system to compute the probability for a child to present neutral or positive affective state. Our results showed that the prototype was able to achieve an average affective state prediction accuracy of 93.05%.

Traffic sign detection is considered as one of the active research topics in transportation and computer vision. The previous works mainly focus on detecting traffic signs in images or in mobile light detection and ranging (LiDAR) data. In this paper, we propose a novel deep learning method to accurately detect traffic signs by fusing the complementary features from registered airborne geo-referenced color images and noisy airborne LiDAR data. Specifically, we first segment the airborne color images to road and non-road segments by integrating various local features in an inequality constraint quadratic optimization model. Next, we find the corresponding road regions in LiDAR data and extract high elevated objects above the road. We then segment the extracted objects to different regions corresponding to traffic sign candidates using Euclidean distance-based clustering. Finally, we find the corresponding traffic sign candidates in color images, extract their deep features, and represent them in a convex optimization model to classify the candidates. A set of extensive experiments have been carried out on the airborne geo-referenced color images and noisy airborne LiDAR data captured by Utah State University from I-15 highway. The results show the effectiveness of the proposed method in detecting traffic signs.

Face recognition under illumination is really challenging. This paper proposes an effective method to produce illumination-invariant features for images with various levels of illumination. The proposed method seamlessly combines adaptive homomorphic filtering, simplified logarithmic fractal dimension, and complete eight local directional patterns to produce illumination-invariant representations. Our extensive experiments show that the proposed method outperforms two of its variant methods and nine state-of-the-art methods, and achieves the overall face recognition accuracy of 99.47%, 94.55%, 99.53%, and 86.63% on Yale B, extended Yale B, CMU-PIE, and AR face databases, respectively, when using one image per subject for training. It also outperforms the compared methods on the Honda UCSD video database using five images per subject for training and considering all necessary steps including face detection, landmark localization, face normalization, and face matching to recognize faces. Our evaluations using receiver operating characteristic (ROC) curves also verify the proposed method has the best verification and discrimination ability compared with other peer methods.

Successfully detecting shadows in still images is challenging yet has wide applications. Shadow properties and features are very important for shadow detection and processing. The aim of this work is to find some new physical properties of shadows and use them as shadow features to design an effective shadow detection method for outdoor color images. We observe that although the spectral power distribution (SPD) of daylight and that of skylight are quite different, in each channel, the spectrum ratio of the point-wise product of daylight SPD with sRGB color matching functions (CMFs) to the point-wise product of skylight SPD with sRGB CMFs roughly approximates a constant. This further leads to that the ratios of linear sRGB pixel values of surfaces illuminated by daylight (in non-shadow regions) to those illuminated by skylight (in shadow regions) equal to a constant in each channel. Following this observation, we calculated the spectrum ratios under various Sun angles and further found out four new shadow properties. With these properties as shadow features, we developed a simple shadow detection method to quickly locate shadows in single still images. In our method, we classify an edge as a shadow or non-shadow edge by verifying whether the pixel values on both sides of the Canny edges satisfy the three shadow verification criteria derived from the shadow properties. Extensive experiments and comparison show that our method outperforms state-of-the-art shadow detection methods. HighlightsWe found the ratios of regions lit by daylight vs. by skylight equal to a constant.We calculated spectrum ratios and found out four new shadow properties.Following the new shadow properties, we proposed a simple shadow detection method.We conducted most extensive experiments and comparison.We also tested our method on images without shadows.

Image registration is a viable task in the field of computer vision with many applications. Researchers propose various local modules insensitive to illumination changes across cross-spectral image pairs to handle the registration challenges under different spectrum conditions. In this paper, we develop an optimized feature-based approach to register natural cross-spectral image pairs. It works on the phase information to quickly identify and describe reliable keypoints that are insensitive to illumination. It then employs a sequence of outlier removal processes to accurately find the matching feature points and the direct linear transformation to estimate the geometric transformation to align the image pair. We benchmark the proposed method and six state-of-the-art feature-based methods on the dataset provided by Ecole Polytechnique Federale De Lausanne (EPFL), which includes 477 pairs of RGB-NIR images. The comprehensive analysis demonstrates that the proposed method achieves up to 13.90% accuracy improvement over the second best registration method.

Raman spectroscopy has been playing an increasingly significant role for cell classification. Here, we introduce a novel microfluidic chip for non-invasive Raman cell natural fingerprint collection. Traditional Raman spectroscopy measurement of the cells grown in a Polydimethylsiloxane (PDMS) based microfluidic device suffers from the background noise from the substrate materials of PDMS when intended to apply as an in vitro cell assay. To overcome this disadvantage, the current device is designed with a middle layer of PDMS layer sandwiched by two MgF(2) slides which minimize the PDMS background signal in Raman measurement. Three cancer cell lines, including a human lung cancer cell A549, and human breast cancer cell lines MDA-MB-231 and MDA-MB-231/BRMS1, were cultured in this microdevice separately for a period of three days to evaluate the biocompatibility of the microfluidic system. In addition, atomic force microscopy (AFM) was used to measure the Young's modulus and adhesion force of cancer cells at single cell level. The AFM results indicated that our microchannel environment did not seem to alter the cell biomechanical properties. The biochemical responses of cancer cells exposed to anti-cancer drug doxorubicin (DOX) up to 24 h were assessed by Raman spectroscopy. Principal component analysis over the Raman spectra indicated that cancer cells untreated and treated with DOX can be distinguished. This PDMS microfluidic device offers a non-invasive and reusable tool for in vitro Raman measurement of living cells, and can be potentially applied for anti-cancer drug screening.

Sparse representation is a viable solution to visual tracking. In this paper, we propose a structured multi-task multi-view tracking (SMTMVT) method, which exploits the sparse appearance model in the particle filter framework to track targets under different challenges. Specifically, we extract features of the target candidates from different views and sparsely represent them by a linear combination of templates of different views. Unlike the conventional sparse trackers, SMTMVT not only jointly considers the relationship between different tasks and different views but also retains the structures among different views in a robust multi-task multi-view formulation. We introduce a numerical algorithm based on the proximal gradient method to quickly and effectively find the sparsity by dividing the optimization problem into two subproblems with the closed-form solutions. Both qualitative and quantitative evaluations on the benchmark of challenging image sequences demonstrate the superior performance of the proposed tracker against various state-of-the-art trackers., Comment: IEEE International Conference on Multimedia and Expo (ICME), 2018

Utilize relationships of singular values to extract content-dependent watermark.Merge relationships of singular values to choose adaptive quantizer for each block.Apply adaptive quantization to embed secure watermark in the wavelet domain.Define a 3-level authentication process to detect authenticity and prove tampering.Use five measures to compensate misclassification and capture distortions. This paper presents a singular-value-based semi-fragile watermarking scheme for image content authentication. The proposed scheme generates secure watermark by performing a logical operation on content-dependent watermark generated by a singular-value-based sequence and content-independent watermark generated by a private-key-based sequence. It next employs the adaptive quantization method to embed secure watermark in approximation subband of each 4i?4 block to generate the watermarked image. The watermark extraction process then extracts watermark using the parity of quantization results from the probe image. The authentication process starts with regenerating secure watermark following the same process. It then constructs error maps to compute five authentication measures and performs a three-level process to authenticate image content and localize tampered areas. Extensive experimental results show that the proposed scheme outperforms five peer schemes and its two variant systems and is capable of identifying intentional tampering, incidental modification, and localizing tampered regions under mild to severe content-preserving modifications.

Face recognition under varying illumination is a challenging task. This study proposes a modified version of local directional patterns (LDP), eight local directional patterns (ELDP), to produce an illumination insensitive representation of an input face image. The proposed ELDP code scheme uses Kirsch compass masks to compute the edge responses of a pixel's neighbourhood. Then, ELDP uses all the directional numbers to produce an illumination invariant image. The authors' extensive experiments show that the ELDP technique achieves an average recognition accuracy of 98.29% on the CMU-PIE face database and 100% on the Yale B face database, and clearly outperforms the state-of-the-art representative techniques.

In this work, a format-compliant encryption method with the data embedding feature for JPEG compressed image is proposed. First, DC coefficients are encoded based on the regions induced by the textural information carried by AC coefficients. Second, AC coefficients are scanned in eight different orders and the order that results in the smallest bitstream size is selected. Next, AC coefficients from each block are extracted in the form of Run/Size and Value, and manipulated to significantly increase the scope of permutation. Then the virtual queue decomposition is proposed to embed external information. All the processes are completely reversible where the embedded information can be extracted and the original content can be perfectly reconstructed from its processed counterpart. The performance of the proposed method is verified through experiments using various standard test images and the UCID dataset. The proposed method is also compared against the conventional format-compliant encryption methods, where its superiority in terms of robustness against sketch attacks, suppression of bitstream size increment, and data embedding are highlighted. In the best case scenario, the proposed method is able to generate an encrypted image of the same size as the original image (e.g., 512×512) with more than 5800 bits of additionally embedded information while achieving a compression gain of 1%. HighlightsSignificantly increase the scope of permutation for DCT coefficients.Robust against sketch attacks.Offer extra feature of data embedding to format-compliant encryption.Design VQD (virtual queue decomposition) data representation scheme where carrier capacity can be increased without giving any impact to bitstream size.Design an adaptive scanning order to suppress the bitstream size increment.

Face recognition under illumination variations is a challenging research area. This paper presents a new method based on the log function and the fractal analysis (FA) to produce a logarithmic fractal dimension (LFD) image which is illumina- tion invariant. The proposed FA feature-based method is a very effective edge enhancer technique to extract and enhance facial features such as eyes, eyebrows, nose, and mouth. Our extensive experiments show the proposed method achieves the best recog- nition accuracy using one image per subject for training when compared to six recently proposed state-of-the-art methods.

Traditional approaches to characterize stem cell differentiation are time-consuming, lengthy and invasive. Here, Raman microspectroscopy (RM) and atomic force microscopy (AFM) - both considered as non-invasive techniques - are applied to detect the biochemical and biophysical properties of trophoblast derived stem-like cells incubated up to 10 days under conditions designed to induce differentiation. Significant biochemical and biophysical differences between control cells and differentiated cells were observed. Quantitative real time PCR was also applied to analyze gene expression. The relationship between cell differentiation and associated cellular biochemical and biomechanical changes were discussed. Monitoring trophoblast cells differentiation.

The authors propose a scalable graph-based semi-supervised ranking system for image retrieval. This system exploits the synergism between relevance feedback based transductive short-term learning and semantic feature-based long-term learning to improve retrieval performance. Active learning is applied to build a dynamic feedback log to extract semantic features of images. Two-layer manifold graphs are then built in both low-level visual and high-level semantic spaces. One graph is constructed at the first layer using anchor images obtained from the feedback log. Several graphs are constructed at the second layer using images in their respective cluster formed around each anchor image. An asymmetric relevance vector is created for each second layer graph by propagating initial scores from the first layer. These vectors are fused to propagate relevance scores of labeled images to unlabeled images. The authors’ extensive experiments demonstrate the proposed system outperforms four manifold-based and five state-of-the-art long-term-based image retrieval systems.

We propose a complementary relevance feedback-based content-based image retrieval (CBIR) system. This system exploits the synergism between short-term and long-term learning techniques to improve the retrieval performance. Specifically, we construct an adaptive semantic repository in long-term learning to store retrieval patterns of historical query sessions. We then extract high-level semantic features from the semantic repository and seamlessly integrate low-level visual features and high-level semantic features in short-term learning to effectively represent the query in a single retrieval session. The high-level semantic features are dynamically updated based on users' query concept and therefore represent the image's semantic concept more accurately. Our extensive experimental results demonstrate that the proposed system outperforms its seven state-of-the-art peer systems in terms of retrieval precision and storage space on a large scale imagery database.

The nanostructures and hydrophobic properties of cancer cell membranes are important for membrane fusion and cell adhesion. They are directly related to cancer cell biophysical properties, including aggressive growth and migration. Additionally, chemical component analysis of the cancer cell membrane could potentially be applied in clinical diagnosis of cancer by identification of specific biomarker receptors expressed on cancer cell surfaces. In the present work, a combined Raman microspectroscopy (RM) and atomic force microscopy (AFM) technique was applied to detect the difference in membrane chemical components and nanomechanics of three cancer cell lines: human lung adenocarcinoma epithelial cells (A549), and human breast cancer cells (MDA-MB-435 with and without BRMS1 metastasis suppressor). Raman spectral analysis indicated similar bands between the A549, 435 and 435/BRMS1 including ~720 cm(-1) (guanine band of DNA), 940 cm(-1) (skeletal mode polysaccharide), 1006 cm(-1) (symmetric ring breathing phenylalanine), and 1451 cm(-1) (CH deformation). The membrane surface adhesion forces for these cancer cells were measured by AFM in culture medium: 0.478 ± 0.091 nN for A549 cells, 0.253 ± 0.070 nN for 435 cells, and 1.114 ± 0.281 nN for 435/BRMS1 cells, and the cell spring constant was measured at 2.62 ± 0.682 mN m(-1) for A549 cells, 2.105 ± 0.691 mN m(-1) for 435 cells, and 5.448 ± 1.081 mN m(-1) for 435/BRMS1 cells.

Generating a georeferenced mosaic map from unmanned aerial vehicle (UAV)imagery is a challenging task. Direct and indirect georeferencing methods may fail to generate an accurate mosaic map due to the erroneous exterior orientation parameters stored in the inertial measurement unit (IMU), erroneous global positioning system (GPS) data, and difficulty inlocating ground control points (GCPs) or having a sufficient number of GCPs. This paperpresents a practical framework to orthorectify and georeference aerial images using the robustfeatures-based matching method. The proposed georeferencing process is fully automatic and does not require any GCPs. It is also a near real-time process which can be used to determine whether aerial images taken by UAV cover the entire target area. We also extend this framework to use the inverse georeferencing process to update the IMU/GPS data which can be further used to calibrate the camera of the UAV, reduce IMU/GPS errors, and thus produce more accurate mosaic maps by employing any georeferencing method. Our experiments demonstrate the effectiveness of the proposed framework in producing comparable mosaic maps as commercial soft-ware Agisoft and the effectiveness of the extended framework in significantly reducing the errors in the IMU/GPS data. © 2016 Society of Photo-Optical Instrumentation Engineers (SPIE)

Intrinsic natures of different appearance between sub-regions of objects and non-objects in optical flows lead to more visual consistency for object proposals. Hence, visual variations in different sub-regions in video sequences over time is a good indicator for likeliness of objects. We propose a method that dynamically measures the objectness of each proposal by exploiting temporal consistency within each optical flow. We develop a block-based feature representation using object's spatial property and define an objectness measure using the temporal changes of this spatial representation. As a result, the proposed temporal objectness learns good object proposals over a short period (e.g., less than 1 second). The proposed method is model-free and can be used to simultaneously learn and track object proposals without training. Experiments on a video dataset shows that the proposed approach significantly outperforms state-of-the-art methods in terms of precision-recall.

We propose to combine short-term block-based fuzzy support vector machine (FSVM) learning and long-term dynamic semantic clustering (DSC) learning to bridge the semantic gap in content-based image retrieval. The short-term learning addresses the small sample problem by incorporating additional image blocks to enlarge the training set. Specifically, it applies the nearest neighbor mechanism to choose additional similar blocks. A fuzzy metric is computed to measure the fidelity of the actual class information of the additional blocks. The FSVM is finally applied on the enlarged training set to learn a more accurate decision boundary for classifying images. The long-term learning addresses the large storage problem by building dynamic semantic clusters to remember the semantics learned during all query sessions. Specifically, it applies a cluster-image weighting algorithm to find the images most semantically related to the query. It then applies a DSC technique to adaptively learn and update the semantic categories. Our extensive experimental results demonstrate that the proposed short-term, long-term, and collaborative learning methods outperform their peer methods when the erroneous feedback resulting from the inherent subjectivity of judging relevance, user laziness, or maliciousness is involved. The collaborative learning system achieves better retrieval precision and requires significantly less storage space than its peers. © 2011 Wiley Periodicals, Inc. © 2011 Wiley Periodicals, Inc.

This paper presents a novel semi-fragile watermarking scheme for image content authentication with tampering localization. The proposed scheme uses a non-traditional quantization method to modify one chosen approximation coefficient of each non-overlapping block to ensure its robustness against incidental attacks and fragileness against malicious attacks. The image content authentication starts with extracting watermark using the parity of quantization results from the probe image, where the round operation is used to ensure the semi-fragile property. It then constructs a binary error map and computes two authentication measures with M"1 measuring the overall similarity between extracted and embedded watermarks and M"2 measuring the overall clustering level of tampered error pixels. These two measures are further integrated to confirm the image content and localize the possible tampered areas. Our experimental results show that our scheme outperforms four peer schemes and is capable of identifying intentional tampering and incidental modification, and localizing tampered regions.

This paper presents a novel Mod4 steganographic method in discrete cosine transform (DCT) domain. Mod4 is a blind steganographic method. A group of 2x2 spatially adjacent quantized DCT coefficients (GQC) is selected as the valid message carrier. The modulus 4 arithmetic operation is then applied to the valid GQC to embed a pair of bits. When modification is required for data embedding, the shortest route modification scheme is applied to reduce distortion as compared to the ordinary direct modification scheme. Mod4 is capable in embedding information into both uncompressed and JPEG-compressed image. To compare Mod4 with other existing methods, carrier capacity, stego image quality, and results of blind steganalysis for 500 various images are shown. Visual comparison of three additional metrics is also presented to show the relative performance of Mod4 among other existing methods.

This paper presents a content-based digital image-watermarking scheme, which is robust against a variety of common image-processing attacks and geometric distortions. The image content is represented by important feature points obtained by our image-texture-based adaptive Harris corner detector. These important feature points are geometrically significant and therefore are capable of determining the possible geometric attacks with the aid of the Delaunay-tessellation-based triangle matching method. The watermark is encoded by both the error correcting codes and the spread spectrum technique to improve the detection accuracy and ensure a large measure of security against unintentional or intentional attacks. An image-content-based adaptive embedding scheme is applied in discrete Fourier transform (DFT) domain of each perceptually high textured subimage to ensure better visual quality and more robustness. The watermark detection decision is based on the number of matched bits between the recovered and embedded watermarks in embedding subimages. The experimental results demonstrate the robustness of the proposed method against any combination of the geometric distortions and various common image-processing operations such as JPEG compression, filtering, enhancement, and quantization. Our proposed system also yields a better performance as compared with some peer systems in the literature.

In this paper, a novel automatic image annotation system is proposed, which integrates two sets of support vector machines (SVMs), namely the multiple instance learning (MIL)-based and global-feature-based SVMs, for annotation. The MIL-based bag features are obtained by applying MIL on the image blocks, where the enhanced diversity density (DD) algorithm and a faster searching algorithm are applied to improve the efficiency and accuracy. They are further input to a set of SVMs for finding the optimum hyperplanes to annotate training images. Similarly, global color and texture features, including color histogram and modified edge histogram, are fed into another set of SVMs for categorizing training images. Consequently, two sets of image features are constructed for each test image and are, respectively, sent to the two sets of SVMs, whose outputs are incorporated by an automatic weight estimation method to obtain the final annotation results. Our proposed annotation approach demonstrates a promising performance for an image database of 12000 general-purpose images from COREL, as compared with some current peer systems in the literature.