Qiong Yang

Ranking with Uncertain Labels

Most techniques for image analysis consider the image labels fixed and without uncertainty. In this paper, we address the problem of ordinal/rank label prediction based on training samples with uncertain labels. First, the core ranking model is designed as the bilinear fusing of multiple candidate kernels. Then, the parameters for feature selection and kernel selection are learned by maximum a posteriori for given samples and uncertain labels. The convergency provable Expectation-Maximization (EM) method is used for inferring these parameters. The effectiveness of the proposed algorithm is finally validated by the extensive experiments on age ranking task. The FG-NET and Yamaha aging database are used for the experiments, and our algorithm significantly outperforms those state-of-the-art algorithms ever reported in literature.

A symmetric patch-based correspondence model for occlusion handling

Occlusion is one of the challenging problems in stereo. In this paper, we solve the problem in a segment-based style. Both images are segmented, and we propose a novel patch-based stereo algorithm that cuts the segments of one image using the segments of the other, and handles occlusion areas in a proper way. A symmetric graph-cuts optimization framework is used to find correspondence and occlusions simultaneously. The experimental results show superior performance of the proposed algorithm, especially on occlusions, untextured areas and discontinuities

Recent advances in subspace analysis for face recognition

Given the unprecedented demand on face recognition technology, it is not surprising to see an overwhelming amount of research publications on this topic in recent years In this paper we conduct a survey on subspace analysis, which is one of the fastest growing areas in face recognition research We first categorize the existing techniques in subspace analysis into four categories, and present descriptions of recent representative methods within each category Then we discuss three main directions in recent research and point out some challenging issues that remain to be solved.

Progressive Cut: An Image Cutout Algorithm that Models User Intentions

In this article, we presented a progressive-cut algorithm for background and foreground segmentation. We first analyzed the user intention behind the additional user-specified stroke, and then incorporated the user intention into the graph-cut framework: we derived an eroded graph to prevent overshrinkage, and added a user attention term to the energy function to compress overexpansion in low-interest areas. Experiments showed that the new algorithm outperforms existing graph-cut methods in both accuracy and speed, and it effectively removes the fluctuation effect, making results more controllable with fewer strokes.

Salience Preserving Image Fusion with Dynamic Range Compression

Gradient conveys important salient features in images. Traditional fusion methods based on gradient generally treat gradients from multichannels as a multi-valued vector, and compute its global statistics under the assumption of identical distribution. However, different source channels may reflect different important salient features, and their gradients are basically non-identically distributed. This prevents existing methods from successful salience preservation. In this paper, we propose to fuse the gradients from multi-channels in the concept of saliency. We first measure the salience map of each channels gradient, and then use their saliency to weight their contribution in computing the global statistics. Gradients with high saliency are properly highlighted in the target gradient, and thereby salient features in the sources are well preserved. Furthermore, we handle the dynamic range problem by applying range compression on the target gradient, and thereby halo effect is effectively reduced.

Progressive cut

Recently, interactive image cutout technique becomes prevalent for image segmentation problem due to its easy-to-use nature. However, most existing stroke-based interactive object cutout system did not consider the user intention inherent in the user interaction process. Strokes in sequential steps are treated as a collection rather than a process, and only the color information of the additional stroke is used to update the color model in the graph cut framework. Accordingly, unexpected fluctuation effect may occur during the process of interactive object cutout. In fact, each step of user interaction reflects the users evaluation of previous result and his/her intention. By analyzing the users intention behind the interaction, we propose a progressive cut algorithm, which explicitly models the users intention into a graph cut framework for the object cutout task. Three aspects of user intention are utilized: 1) the color of the stroke indicates the kind of change s/he expects, 2) the location of the stroke indicates the region of interest, 3) the relative position between the stroke and the previous result indicates the segmentation error. By incorporating such information into the cutout system, the new algorithm removes the unexpected fluctuation effect of existing stroke-based graph-cut methods, and thus provides the user a more controllable result with fewer strokes and faster visual feedback. Experiments and user study show the strength of progressive cut in accuracy, speed, controllability, and user experience.

Imlooking: image-based face retrieval in online dating profile search

Textual search, the approach used by the majority of existing online dating sites, successfully covers a variety of attributes, such as age range and gender, but falls short when searching for facial features. Meanwhile, by using images as the query in a search, current image-based face-retrieval applications ease the challenge of textual description from users, but only focus on finding the same person. We believe there is a gap that needs to be filled in image-based face retrieval to further support the interpersonal search scenarios on Internet dating sites. Therefore, we are introducing a profile search prototype -- ImLooking - using an augmented image-based face retrieval filter. First, we present a prototype design and offer technical support. In a user study, participants quickly felt at home in user interface and acclimatized to the way the prototype operates. In addition, they reported they enjoyed the interaction process.

An Effective System for Optical Microscopy Cell Image Segmentation, Tracking and Cell Phase Identification

The lacking of automatic screen systems that can deal with large volume of time-lapse optical microscopy imaging is a bottleneck of modern bio-imaging research. In this paper, we propose an effective automated analytic system that can be used to acquire, track and analyze cell-cycle behaviors of a large population of cells. We use traditional watershed algorithm for cell nuclei segmentation and then a novel hybrid merging method is proposed for fragments merging. After a distance and size based tracking procedure, the performance of fragments merging is improved again by the sequence context information. At last, the cell nuclei can be classified into different phases accurately in a continuous hidden Markov model (HMM). Experimental results show the proposed system is very effective for cell sequence segmentation, tracking and cell phase identification.

Detection of Roads in SAR Images using Particle Filter

A novel method is presented to detect roads in synthetic aperture radar (SAR) images. A multi-segmented poly-line model is introduced to provide a more accurate description of the road as well as to ensure the road curves smoothness in the model level. We then solve the road detection problem using the Bayesian tracking theory, where the particle filtering algorithm is adopted to provide a simple and consistent framework. The effectiveness and robustness of the proposed method is demonstrated by experimental results.

Incorporating Generic Learning to Design Discriminative Classifier Adaptable for Unknown Subject in Face Verification

In recent years, there has been a growing interest on the verification of unspecific person, which requires the system adaptable for unknown new subject. Most of previous works used generative methods. In this paper, we propose a discriminative method, Bayesian Competitive Model, to explicitly handle the person-unspecific problem. The key idea originates from the observation that it is possible to design a discriminative classifier adaptable for unknown new subject when generic learning is applied. The generic learning functions in two aspects: First, it learns the generic distribution of faces, and thus provides a MAP framework for verification. Second, it learns the intra-personal variations of numerous known persons to infer the distribution of the unknown new subject. Both distributions are formulated in GMM model, respectively. To further improve the performance, we integrate Bayesian Competitive Model with a generative classifier based on confidence. A number of experiments on the BANCA dataset demonstrate the effectiveness of the new algorithm in handling the personunspecific problem, and its advantage over existing algorithms.