Brais Cancela

Improving retinal artery and vein classification by means of a minimal path approach

This paper describes a technique for the retinal vessel classification into artery and vein categories from fundus images within a framework to compute the arteriovenous ratio. This measure is used to assess the patient condition, mainly in hypertension and it is computed as the ratio between artery and vein widths. To this end, the vessels are segmented and measured in several circumferences concentric to the optic nerve. The resulting vessel segments at each radius are classified as artery or vein independently. After that, a tracking procedure joins vessel segments in different radii that belong to the same vessel. Finally, a voting system is applied to obtain the final class of the whole vessel. The methodology has been tested in a data set of 100 images labeled manually by two medical experts and a classification rate of over 87.68 % has been obtained.

On the Automatic Computation of the Arterio-Venous Ratio in Retinal Images: Using Minimal Paths for the Artery/Vein Classification

Abnormalities in the retinal vessel tree are associated with different pathologies. Usually, they affect arteries and veins differently. In this regard, the arteriovenous ratio(AVR) is a measure of retinal vessel caliber, widely used in medicine to study the influence of these irregularities in disease evolution. Hence, the development of an automatic tool for AVR computation as well as any other tool for diagnosis support need an objective, reliable and fast artery/vein classifier. This paper proposes a technique to improve the retinal vessel classification in an AVR computation framework. The proposed methodology combines a color clustering strategy and a vessel tracking procedure based on minimal path approaches. The tests performed with 58 images manually labeled by three experts show promising results.

Open-world Person Re-Identification by Multi-Label Assignment Inference.

(c) 2014. The copyright of this document resides with its authors. It may be distributed unchanged freely in print or electronic forms.

Hierarchical framework for robust and fast multiple-target tracking in surveillance scenarios

Multiple-target tracking is a challenging field specially when dealing with uncontrolled scenarios. Two common approaches are often used, one based on low-level techniques to detect each object size, position and velocity, and other based on high-level techniques that deal with object appearance. None of these methods can deal with all possible problems in multiple-target tracking: environment occlusions, both total and partial, and collisions, such as grouping and splitting events. So one solution is to merge these techniques to improve their performance. Based on an existing hierarchical architecture, we present a novel technique that can deal with all the mentioned problems in multiple tracking targets. Blob detection, low-level tracking using adaptive filters, high-level tracking based on a fixed pool of histograms and an event management that can detect every collision event and performs occlusion recovery are used to be able to track every object during the time they appear within the scene. Experimental results show the performance of this technique under multiple situations, being able to track every object in the scene without losing their initial identification. The speed processing is higher than 50 frames, which allows it to be used under real-time scenarios.

Multiple human tracking system for unpredictable trajectories

Tracking multiple objects into a scene is one of the most active research topics in computer vision. The art of identifying each target within the scene along a video sequence has multiple issues to be solved, being collision and occlusion events among the most challenging ones. Because of this, when dealing with human detection, it is often very difficult to obtain a full body image, which introduces complexity in the process. The task becomes even more difficult when dealing with unpredictable trajectories, like in sport environments. Thus, head-shoulder omega shape becomes a powerful tool to perform the human detection. Most of the contributions to this field involve a detection technique followed by a tracking system based on the omega-shape features. Based on these works, we present a novel methodology for providing a full tracking system. Different techniques are combined to both detect, track and recover target identifications under unpredictable trajectories, such as sport events. Experimental results into challenging sport scenes show the performance and accuracy of this technique. Also, the system speed opens the door for obtaining a real-time system using GPU programing in standard desktop machines, being able to be used in higher-level human behavioral systems, with multiple applications.

Contextual and skin color region information for face and arms location

Interest in intelligent human-computer interfaces has grown in recent years due to the possibilities that they offer. To these systems, two of the most important sources of interaction are the face and the arms gestures. Different face detection approaches have been made up to date, while arms detection is still a challenging task. This paper describes a methodology for the location of faces and arms in color images combining color information with region information and domain knowledge information. The obtained method is able to work very accurately regardless of races and skin colors, poses, resolutions, lighting conditions, and so on. It has been tested with a representative range of different arm positions, achieving encouraging results.

On the use of a minimal path approach for target trajectory analysis

Vision-based action recognition has multiple applications, mainly focused in video surveillance systems. The art of labeling each target behavior in crowded scenarios is a complicated field since usually we do not have visual confirmation of the parts of a target to infer its behavior. Thus, trajectory analysis becomes a good choice to try to infer knowledge about target movements. Most of the contributions to this field involve a training period in which we obtain information a priori about the environment, storing a dataset with all the possible usual routes. Based in the minimal path theory using geodesic active contours, we present a novel architecture where no initial information about the scene is needed, while it is possible to include it if necessary to specify constraints. Experimental results in two different application domains show the performance and flexibility of this method, being able to be used in multiple trajectory analysis problems.

Automatic Eye Gesture Recognition in Audiometries for Patients with Cognitive Decline

This paper provides a specifically adapted methodology for supporting the audiologists when testing the hearing of patients with cognitive decline or other communication disabilities. These patients can not interact with the audiologist conventionally, but they often express gestural reactions when they perceive the auditory stimuli typically associated to the eyes region. From a video sequence captured during the hearing evaluation, we analyze the movements in the area of the patient’s eyes, so we can detect these gestural reactions. We define a set of different gestures for classification, based on the expert knowledge. The proposed method achieves an accuracy of the 90.65% when classifying these movements, showing their separability, and therefore, the possibility of interpreting them with high-level information as positive reactions to the auditory stimuli.

Automatic processing of audiometry sequences for objective screening of hearing loss

Audiology is the branch of science that deals with hearing, balance, and related disorders. Detecting patients with slow responses to auditory stimuli is relevant because this slowness could be due to other cognitive problems or conditions, which should be studied carefully. In this paper, we present an automatic methodology for processing video sequences recorded during the performance of hearing tests to patients. This screening method allows us to measure the patients response times to the auditory stimuli sent to them, and based on these times, to identify those patients with response times abnormally slow. The method is tested on individuals taken at random from a standard population, and based on the obtained results, it is confirmed that the proposed method is valid for the automatic detection of patients with slow response times, and it also serves to the experts as a tool for the accurate and objective measurement of these times.

Path analysis in multiple-target video sequences

Path analysis becomes a powerful tool when dealing with behavior analysis, i. e., detecting abnormal movements. In a multiple target scenario it is complicated to obtain each object path because of collision events, such as grouping and splitting targets, and occlusions, both total or partial. In this work, a method to obtain the similarity between different trajectories is presented, based in register techniques. In addition, an hierarchical architecture is used to obtain the corresponding paths of the objects in a scene, to cope with collision events. Experimental results show promising results in path analysis, enabling it to establish thresholds to abnormal path detection.