Xavier Roca

Exploiting multiple cues in motion segmentation based on background subtraction

This paper presents a novel algorithm for mobile-object segmentation from static background scenes, which is both robust and accurate under most of the common problems found in motion segmentation. In our first contribution, a case analysis of motion segmentation errors is presented taking into account the inaccuracies associated with different cues, namely colour, edge and intensity. Our second contribution is an hybrid architecture which copes with the main issues observed in the case analysis by fusing the knowledge from the aforementioned three cues and a temporal difference algorithm. On one hand, we enhance the colour and edge models to solve not only global and local illumination changes (i.e. shadows and highlights) but also the camouflage in intensity. In addition, local information is also exploited to solve the camouflage in chroma. On the other hand, the intensity cue is applied when colour and edge cues are not available because their values are beyond the dynamic range. Additionally, temporal difference scheme is included to segment motion where those three cues cannot be reliably computed, for example in those background regions not visible during the training period. Lastly, our approach is extended for handling ghost detection. The proposed method obtains very accurate and robust motion segmentation results in multiple indoor and outdoor scenarios, while outperforming the most-referred state-of-art approaches.

Expression of CD44H and CD44v3 in normal oesophagus, Barrett mucosa and oesophageal carcinoma.

AIMS: To examine CD44H and CD44v3 expression in normal gastric and small bowel mucosa, normal and Barrett oesophagus, and oesophageal epithelial malignancies (squamous cell carcinoma and adenocarcinoma). METHODS: Ninety five specimens, comprised of 40 of normal oesophageal, gastric and small bowel mucosa, 22 of Barrett oesophagus (two with dysplastic changes), 20 of resected adenocarcinomas, and 13 of squamous cell carcinoma, were evaluated. The samples were fixed in formalin and subsequently stained with anti-CD44H and anti-CD44v3 monoclonal antibodies using the avidin-biotin peroxidase technique. RESULTS: In contrast to normal oesophagus, which showed positivity for both CD44 epitopes (CD44H and CD44v3) in the basal third of the epithelium, antral and intestinal subtypes of Barrett oesophagus expressed CD44H only, the distribution being focal in non-dysplastic and diffuse in dysplastic Barrett mucosa. Similarly, normal antral glands and small bowel epithelium were focally immunopositive for CD44H at the base of the crypts. All squamous cell carcinomas were diffusely positive for both isoforms, whereas 75% (15/20) of the adenocarcinomas expressed CD44H and 60% (12/20) expressed CD44v3. CONCLUSIONS: CD44H is expressed in the proliferating areas of both normal squamous epithelium and Barrett mucosa. CD44H expression seems to increase progressively in dysplasia and infiltrating carcinoma, similar to the process described in the stomach. CD44v3 expression, usually not observed in normal or neoplastic gastric mucosa, was present in normal squamous epithelium and oesophageal squamous cell carcinoma. CD44v3 immunoreactivity was also identified in 60% of adenocarcinomas. These findings suggest that CD44v3 may play a role in the development of oesophageal carcinoma of both squamous and glandular types.

New autofocusing algorithm for cytological tissue in a microscopy environment

We present a new autofocusing method as a part of a general project to automatize a transmission microscope for inspection of cytological tissue. These images present several cell superpositions that lead to the appearance of false edges under defocus conditions. This means that some autofocusing algorithms, based on derivatives, may not work properly. The method that we propose allows us to solve this problem. The paper is divided in two parts. In the first we present a new focusing criterion, the dynamic focus criterion (DFC). It is computed by comparing (point to point) the gray-level values of two images instead of comparing the energy of the first or second derivatives. This allows us to avoid certain bad effects on the shape of the criterion function, which are due to the structure of the cytological images and the transmission of light. In the second part we propose an optimal search algorithm to find the best-focused image using the DFC. Optimality is given in terms of the number of lens movements. The algorithm considers not only the unimodal property, as the classical Fibonacci search does, but also the symmetry of the criterion function. Finally, some results on its performance are given.

Interpretation of complex situations in a semantic-based surveillance framework

The integration of cognitive capabilities in computer vision systems requires both to enable high semantic expressiveness and to deal with high computational costs as large amounts of data are involved in the analysis. This contribution describes a cognitive vision system conceived to automatically provide high-level interpretations of complex real-time situations in outdoor and indoor scenarios, and to eventually maintain communication with casual end users in multiple languages. The main contributions are: (i) the design of an integrative multilevel architecture for cognitive surveillance purposes; (ii) the proposal of a coherent taxonomy of knowledge to guide the process of interpretation, which leads to the conception of a situation-based ontology; (iii) the use of situational analysis for content detection and a progressive interpretation of semantically rich scenes, by managing incomplete or uncertain knowledge, and (iv) the use of such an ontological background to enable multilingual capabilities and advanced end-user interfaces. Experimental results are provided to show the feasibility of the proposed approach.

Quality control of safety belts by machine vision inspection for real-time production

Safety belts are specific fabrics manufactured to ensure the highest performance. Their manufacturing process not only has to assure its endurance to high tension strength, but also has to guarantee its correct visual appearance. Safety belts must not contain fiber breaks, knots, thickness variations, etc. Such defects imply the unfulfillment of rigorous safety standards. We describe the development of a computer vision inspection system, which control safety belts at a speed rate of 2 m/s. This inspection rate has been achieved by means of a parallel architecture and the use of optimized vision algorithms.

Robust and Efficient Multipose Face Detection Using Skin Color Segmentation

In this paper we describe an efficient technique for detecting faces in arbitrary images and video sequences. The approach is based on segmentation of images or video frames into skin-colored blobs using a pixel-based heuristic. Scale and translation invariant features are then computed from these segmented blobs which are used to perform statistical discrimination between face and non-face classes. We train and evaluate our method on a standard, publicly available database of face images and analyze its performance over a range of statistical pattern classifiers. The generalization of our approach is illustrated by testing on an independent sequence of frames containing many faces and non-faces. These experiments indicate that our proposed approach obtains false positive rates comparable to more complex, state-of-the-art techniques, and that it generalizes better to new data. Furthermore, the use of skin blobs and invariant features requires fewer training samples since significantly fewer non-face candidate regions must be considered when compared to AdaBoost-based approaches.

Solving the multi object occlusion problem in a multiple camera tracking system

An efficient method to overcome adverse effects of occlusion upon object tracking is presented. The method is based on matching paths of objects in time and solves a complex occlusion-caused problem of merging separate segments of the same path.

Trajectory occlusion handling with multiple-view distance-minimization clustering

A robust and efficient method for overcoming the negative effects of long-time occlusion in the tracking process is presented. The proposed approach is based on the matching of multiple trajectories in time. Trajectories are sets of 2-D points in time and in a joint ground plane of the world coordinate system. In order to avoid mismatches due to possible measurement outliers, we introduce an integral distance between compared trajectories. The proposed method can also be considered as an interpolation algorithm for a disconnected trajectory during the blackout. Thus this technique solves one of the most difficult problems of occlusion handling: the matching of two unconnected parts of the same trajectory.

Efficient discriminative multiresolution cascade for real-time human detection applications

Human detection is fundamental in many machine vision applications, like video surveillance, driving assistance, action recognition and scene understanding. However in most of these applications real-time performance is necessary and this is not achieved yet by current detection methods. This paper presents a new method for human detection based on a multiresolution cascade of Histograms of Oriented Gradients (HOG) that can highly reduce the computational cost of detection search without affecting accuracy. The method consists of a cascade of sliding window detectors. Each detector is a linear Support Vector Machine (SVM) composed of HOG features at different resolutions, from coarse at the first level to fine at the last one. In contrast to previous methods, our approach uses a non-uniform stride of the sliding window that is defined by the feature resolution and allows the detection to be incrementally refined as going from coarse-to-fine resolution. In this way, the speed-up of the cascade is not only due to the fewer number of features computed at the first levels of the cascade, but also to the reduced number of windows that need to be evaluated at the coarse resolution. Experimental results show that our method reaches a detection rate comparable with the state-of-the-art of detectors based on HOG features, while at the same time the detection search is up to 23 times faster.

Expression patterns of cyclins D1 and E in condyloma acuminatum in comparison with psoriatic proliferative lesions

Human papillomavirus (HPV) is able to subvert the host cell replication machinery so as to foster viral reproduction. Specifically, HPV infection is known to induce expression of proliferation antigens such as Ki67 and proliferative cell nuclear antigen (PCNA) in differentiated keratinocytes which have ceased to replicate. In order to determine whether cyclin D1 or cyclin E deregulation is also a feature of HPV infection, an immunohistochemical investigation of cyclin D1, cyclin E, Ki67, and PCNA expression has been carried out in 38 cases of HPV 6/11‐related condyloma acuminatum (CA). Results were compared with those obtained from 15 psoriatic proliferative lesions. Whereas 35 (92·1 per cent) CA samples exhibited positive nuclear immunostaining for cyclin E, no cyclin D1 immunoreaction was detected in any of the CA samples studied. All psoriatic lesions showed immunostaining for both cyclins. All CA cases revealed a positive immunoreaction for Ki67 and 33 for PCNA, both in the parabasal and in the differentiated upper epithelial layers. Parabasal keratinocytes of psoriatic lesions were always positive for both Ki67 and PCNA. These results indicate that in the onslaught of HPV 6/11 upon the keratinocyte replication machinery, cyclin E, PCNA, and Ki67 are amongst the targeted cell cycle modulators, whereas cyclin D1 is spared the main effects of virus–cell interplay. In contrast, both cyclins seem to be induced in psoriasis, a non‐viral proliferative skin condition.