Mohamed Chaouki Babahenini

Real-time wrist localization in color images based on corner analysis

Hand detection and gestures recognition have become very popular in recent human-computer interaction systems. Although several methods of hand detection have been proposed in the literature, they exist few methods that use the wrist as a factor of detection, others impose constraints on the length of the sleeves and on the orientation of the hand. In this work, we present a new two-stage algorithm of wrist localization designed for hand detection and gestures recognition systems. The first stage of the algorithm consists in separating the skin region containing the hand from the background, and in the second stage, the wrist is localized from the resulted skin mask. The main contribution of the proposed method is based on the analysis of corners along the contour of the skin masks to localize the wrist emplacement. Based on an evaluation on 437 color images with their ground-truth and three sets of skin masks, we compared our method with other efficient methods of literature and the results obtained were very satisfactory.

Topological synchronization mechanism for robust watermarking on 3D semi-regular meshes

This paper presents a contribution in matter of robustness of a previous method of 3D objects watermarking [10] based on wavelet decomposition on semi-regular triangular meshes. This decomposition gives rise to a multiresolution mesh where the watermark could be embedded at each level of resolution, in this case we talk about a hierarchical watermarking. In practice the robustness of the previous method against geometrical attacks was high only in the low resolution which makes it far from being a hierarchical method. Our analysis made on the fragility of that method allowed us to reveal the cause which was not primarily the bad bits quantization of the watermark, but rather the alteration of the sequencing of the embedded bits (synchronization). Our contribution is to replace the previous geometrical mechanism with a new topological synchronization mechanism, which has resulted a real hierarchical and robust watermarking.

Automatic Hand Detection in Color Images based on skin region verification

Among the modern means of communications that appeared recently, there is the natural computer interaction using hands. Several methods have been proposed for their detection in the literature, and the common methods are based on skin color. The majority of these methods perform on images containing the hand region only or on images containing the hand and the forearm regions, but few of them have been interested to deal automatically for both situations.In this paper, we propose a new method of automatic hand detection in color images based on skin region verification. Using the 2D properties of hand components (fingertips, hand palm, wrist) we verify if the skin region treated contains the hand region with or without the presence of the forearm, and for each situation the hand is detected differently. Compared to other methods of literature, our method avoids several scenarios of misdetection and process in a consistent manner through the verification and detection stages. Using a database containing 899 color images of hand gestures, we evaluated the efficiency of our method and we compared it to other methods of literature, the results obtained were very satisfactory.

Improving the Method of Wrist Localization Local Minimum-Based for Hand Detection

Nowadays, hand detection and gestures recognition have become very popular in human computer interaction systems. Several methods of hand detection based on wrist localization have been proposed but the majority work only with short sleeves and they are not efficient in front of all the challenges. Hand detection based on wrist localization proposed by Grzejszczak et al. (Proceedings of the 8th International Conference on Computer Recognition Systems CORES 2013 439–449, 2013), Nelpa et al. (Man Mach Interact 3(242):123–130, 2014) [3, 4] use the property of local minima along the contour of the skin mask obtained in the first stage to detect the wrist position. Although this technique provides good results where the skin mask contains the hand and the forearm, it still sensitive to the short contour where the skin mask contains the hand region only which generate false detection of the hand. We present in this paper an assessment of this method where the skin mask contains the hand region only. The main idea is based on the 2D shape properties of the hand and its components. Using 134 color images with their ground- truth, we evaluated the method enhanced and the results obtained were very satisfactory compared to the original one.

Puzzle based system for improving Arabic handwriting recognition

Abstract Several researches have been done through the last years to improve the recognition rate of Arabic handwritten recognition systems. The use of different post-processing techniques for word selection methods such as voting and contextual information was the choice of many systems. In our previous works, we proposed a technique that uses SVM classifier to recognize Arabic handwritten based on two passes horizontal and vertical. In this work, we add a Puzzle algorithm as a post-processor to improve the recognition rate, especially for ambiguous characters. Our method uses a set of stages (filtering, segmentation, features extraction, classification, and post-treatment) and leads to a better classification rate. The approach is tested on Tunisian database IFN/ENIT for handwritten Arabic. It gives encouraging results and opens other perspectives in the domain of Arabic handwritten recognition.

Facial Emotion Detection in Massive Open Online Courses

Recently, the Massive Open Online Course (MOOC) has appeared as a new emerging method of online teaching with the advantages of low cost and unlimited participation as well as open access via the web. However, the use of facial emotion detection in MOOCs is still unexplored and challenging. In this paper, we propose a new innovative approach for facial emotion detection in MOOCs, which provides an adaptive learning content based on students’ emotional states and their profiles. Our approach is based on three principles: (i) modeling the learner using the MOOC (ii) using of pedagogical agents during the learning activities (iii) capturing and interpreting the facial emotion of the students. The proposed approach was implemented and tested in a case study on the MOOC.

Efficient inverse transform methods for VPL selection in global illumination

In computer graphics, designing efficient Global Illumination methods is a hot research topic. These methods consist in computing the light distribution inside a 3D scene. There exist several global illumination-based rendering methods, but one popular approach is based on Virtual Point Light (VPL). It is a two-step approach. First, the algorithm generates VPLs that act as secondary light sources (indirect illumination). Second, the radiance of a pixel is computed by summing the contributions of a small set of VPLs (rather than all the VPLs) selected randomly. The most active issues rely on how to select a small set of VPLs that contribute more to the final image. In this paper, we propose two new VPL selection methods using the inverse transform method. To provide realistic images, we propose a Multiple Importance Sampling technique combining an inverse transform method with a gathering approach. The obtained results demonstrate the effectiveness of our methods in terms of image quality and rendering time.

Full body adjustment using iterative inverse kinematic and body parts correlation

In this paper, we present an iterative inverse kinematic method that adjust 3D human full body pose in real time to achieve new constraints. The input data for the adjustments are the starting posture and the desired end effectors positions -constraints-. The principal idea of our method is to divide the full-body into groups and apply inverse kinematic based on conformal algebra to each group in specific order, our proposed method involve correlation of body parts. In the first part of the paper we explain the used inverse kinematic when handle with one and multiple constraints simultaneously and in the case of the collision induced by the joints with the objects of the environment. The second part focuses on the adjustment algorithm of the full body using the inverse kinematic described above. Comparison is made between the used inverse kinematic(IK) and another inverse kinematic that have the same principle. In the case of multiple tasks simultaneously, our inverse kinematic gives results without con ict. With presence of obstacles, our IK allows to avoid collisions too. Preliminary results of the adjustment method show that it generates new realistic poses that respect quickly new constraints. The tests made on our adjustment method show that it resolves the motion retargeting problem.

Combining visual hull and stereovision technique for new 3D reconstruction method

The 3D reconstruction consists to generate a 3D model from information inspired from a real scene. It is a very interesting alternative, because it aims to improve the modeling of 3D environments, in terms of accuracy and design speed, as well as the level of realism. We have proposed a new approach for 3D reconstruction from images, by combining in new way two IBMR [10] (Image Based Rendering and Modeling) methods: The first is the stereovision and the second is the visual hull.

Retinal blood vessel segmentation using the elite-guided multi-objective artificial bee colony algorithm

Retinal vessel segmentation constitutes an essential part of computer-assisted tools for the diagnosis of ocular diseases. In this study, the authors propose an unsupervised retinal blood vessels segmentation approach based on the elite-guided multi-objective artificial bee colony (EMOABC) algorithm. The proposed method exploits several criteria simultaneously to improve the accuracy of the segmentation results. An energy curve function is used to calculate the values of the thresholding criteria, in order to reduce the noise response from lesions and select the optimal thresholds that separate the blood vessels from the background. In order to achieve computational speed up, a stopping criterion method is used to adjust the parameters of the EMOABC algorithm. The proposed method is computationally simple and faster than most of the available unsupervised algorithms, demonstrating fast convergence to the final segmentation. Additionally, the proposed vessel segmentation method outperforms the metaheuristics vessels segmentation algorithms reported in the literature. The achieved mean discrepancy metrics for the proposed approach are 94.5% accuracy, 97.4% specificity and 73.9% sensitivity for DRIVE database, and 94% accuracy, 96.2% specificity and 73.7% sensitivity for STARE database.