{Energy minimization by graph cuts is increasingly used in computer vision. One of the most attractive aspects of this approach is that it allows, under certain conditions, to find global minima of functions, avoiding the pitfall of classical local minima. In this paper, we describe our semisupervised approach of visual objects segmentation from a priori information, reducing the space of research. Then, we present different applications and we discuss the application of this method on color images and we evaluate a number of functions of penalties used for the construction of the graph.},

{Marbling (intramuscular fat tissue) in beef meat is one of the most important criteria for quality, notably juiciness, in meat grading systems. Visual inspection of the meat surface is the common way to assign quality grading level, which is accomplished by authorized experts called graders. In the last years, several works were proposed in order to introduce computer vision on meat quality evaluation. In these works, meat grading was based exclusively on the analysis of meat surface images. In this paper, a new technique using near-infrared light in transmission mode is used to evaluate the beef meat quality based on the marbling detection. It is demonstrated that using near-infrared light in transmission mode, it is possible to detect the fat not only on the surface, as in traditional methods, but also under the surface. Also, in combining the analysis of the two sides of the meat simple, it is possible to estimate the volumetric marbling. We compared results from traditional techniques and those provided by the nearinfrared camera using light transmission. The result of this new study showed that using near-infrared light in transmission mode is a valuable technique to evaluate meat quality, thereby demonstrating the possibility of implementing this approach in a vision system},

{The classification is a problem that occurs in several areas of application, as the classification of emails, speech recognition, signature recognition and face recognition ... for this purpose several methods are applied in different approaches. In this article we introduce the artificial immune system for the plantar classification, we chose IRIS as example to apply artificial immune system. We initially present an overview on the mechanism of natural immunity, thereafter we will give stages of learning artificial immune systems and ultimately results obtained by applying these methods to the classification of plants IRIS},

{Exploiting images coming from different sensors is an important challenge in the remote sensing field. Integration of new knowledge is crucial to help the user interpret satellite images and track their spatio-temporal changes over time. Thus, we propose a new approach to process multi-date satellite images. Our approach combines knowledge discovery from satellite image databases, and fusion methods in order to find out new and relevant knowledge useful to create decision making and prevision models. The choice of the proposed architecture is motivated by two reasons. First, we need to process imperfection related to the knowledge discovery and interpretation processes. Second, we should integrate new, valid, potentially useful and ultimately understandable knowledge hidden in databases. Our work is based on three concepts (multi-agent systems, case based reasoning and rule based reasoning) and is validated through the use of two optical satellite images coming from Landsat 7 representing the region of Matmata (South of Tunisia)},

{In this paper, we propose a method that analyzes human behaviour in a shopping context. This method is based on motion detection and object tracking. Our research is developed in order to build a real-time application that will analyze people’s behaviour while shopping. Specifically, we detect customer interactions with specific products},

{The methods of Content Based Image Retrieval (CBIR), initially carry out an index in a base of images. It consists of extracting the low level descriptors (color, texture, form). In order to be able to establish a criterion of similarity between them, for each user’s request, a distance is calculated according to these descriptors and the images closest to the request are turned over. The main purpose of this article is to present a
method of segmentation resulting from the field of Data Mining. This allows us to partitionate the whole base of images into group of similar images which will be stored in a particular tree structure called "generic Quad-Tree”; this fact yields to the minimization of the storage space of information. This method is an adaptation of the so-called CHAMELEON, which is a hierarchical segmentation applied to images. It requires the definition of a measurement of dissimilarity (or similarity) between images represented out of a quad-tree},

{One issue of great importance in the industrial monitoring is constituted by the possibility of making a false detection alarms and possibly predict failure. These issues fall under the supervision dynamic method in terms of moving towards the concept of proactive maintenance, and more particularly to the prognosis. In this context, our work is oriented to the temporal Neurofuzzy systems; they are characterized by their ability to represent industrial system by numeric and symbolic datas at the same time in a single tool. Also, we have chosen to integrate a fuzzy inference systèm in the neural network architecture of Elman. The production system is chosen workshop clinkering of SCIMAT, cement Ain Touta "Batna" Algeria},

{Among the existing image fusion algorithms, Intensity- Hue-Saturation (IHS) technique is the most widely used one. When the IHS method is used for IKONOS imagery, there is a significant color distortion, due mainly to the range of wavelengths in an IKONOS Pan image. Based on the fact that the grey values of Pan in the green vegetated regions are far larger than the grey values of intensity (I) and that vegetation ppears of relatively high reflectance in Near Infrared (NIR) and Pan bands, we propose a new technique to solve this problem. This approach makes se of the Normalized Difference Vegetation Index (NDVI) to identify the vegetation area and then enhances it in the green (G) band by using the red (R) and the NIR bands. We obtain an intensity image with grey values comparable to the Pan’s grey values. Hence the color distortion in the fused image is reduced. Visual and statistical analyses prove that the concept of the proposed method is promising, and it significantly improves the fusion quality compared to conventional IHS techniques},

{Segmentation is a fundamental step in image description or classification. In recent years, several computational models have been used to implement segmentation methods but without establishing a single analytic solution. In this paper, the problem of textured images segmentation upon an unsupervised scheme is addressed. Until recently, there has been few interest in segmenting images involving possible complex random texture patterns. To overcome these adversities, we proposed a cascade clustering method combining a statistical features and fuzzy c-means (FCM) clustering algorithm. For textured image segmentation, the performance of the proposed approach is compared to the standard FCM},

{Image registration is an inevitable problem arising in many image-processing applications whenever two or more images of the same scene have to be compared pixel by pixel. The increased volume of satellite images has reinforced the need for automatic image registration methods. Several techniques have been developed for many different types of applications and data. In this paper, two new feature-based approaches to automated image-to-image registration are presented. The characteristic of the first approach is that it combines an invariant moment shape descriptor with improved chaincode matching to establish correspondences between the potentially matched regions detected from the two images. It is robust in that it overcomes the difficulties of control-point correspondence by matching the images both in the feature space, using the principle of minimum distance classifier (based on the combined criteria), and sequentially in the image space, using the rule of root mean square error (RMSE). This method works well for image pairs in which the contour information is well preserved, such as the optical images from Landsat and Spot satellites. For the registration of the optical images with synthetic aperture radar (SAR) images, we propose another method based on the wavelet transform, this second method uses spectral information of the images and their local wavelet transform modulus maxima to extract a set of control points. The experimental result demonstrates the robustness, efficiency and accuracy of
the two algorithms},

{The objective of this paper is the segmentation of handwritten Arabic script into connected components using a combination between Hough
Transform (HT) and Mathematical Morphology (MM). We start by a presentation of these two methods, their applications and their limits. According to the complexity of the segmented document, different kinds of combinations between these two methods are done in order to extract blocks, words, sub-words and some times characters. The main contribution of MM is binarisation, elimination of background and noisy, construction and extraction of straight lines and connected components. The HT is used essentially for line detection. Different kinds of lines can be extracted from a document and can be useful for its segmentation: the script baseline, pre-printed lines and horizontal and vertical lines used for the identification of boxes. Pre-printed lines and boxes are generally used for user enter handwriting script. For a document, three levels of segmentations can be done: segmentation of document in blocks, blocks in connected components and connected components in characters. We are interested in this paper by the two first level of segmentation. Extracted connected components from an Arabic document can in many cases be isolated characters. The application of the two proposed methods and their combination is done on several kinds of documents. Handwritten multilingual bank check, newspapers and printed formulas. The extraction rate, on CENPARMI Arabic check database and on the IFN/ENIT database attempts 99.63% for block documents and 85% for sub-words},

{The Fit is one of the parameters characterizing the concept of comfort footwear. Yet there is no system at present, for its validation on the finished product. This paper presents the study of the morphology of the foot and footwear analysis to be carried out through the forms and models of shoes concerned},

{The three-dimensional medical imagery appeared with the focusing of various methods of Image Acquisition System such as tomography X (CT), ultrasonic echography and the Magnetic Resonance Imaging (MRI), the obtained images are a series of 2D slices. The data visualization is not sufficient to infer the characteristics necessary for the diagnosis. In order to propose a powerful toolkit for data analysis and visualization models 3D to the medical use, there is recourse on the 3D reconstruction. The Principle of reconstruction of three-dimensional shape from stacks of 2D parallel images. (Slices ‘serial’) consist to extract geometrical structures representing the three-dimensional image. We present in this paper the method of «Marching cubes» that permit to obtain very good approximations of the original surface and associated with algorithms of improvement and smoothing. The surfaces obtained are very close to reality},

{An efficient, and robust License plate recognition system become more and more a necessity in the management of many domains as traffic flow and security control, management of parking lots and management of stolen cars and criminals pursuits. In This study we propose an automatic Tunisian license plate recognition system. As all the ALPRS, we will go through four steps. The first step consists of the extraction of the license plate of the vehicle from a green scale image captured by a digital camera by base essentially on an edge detection method. This plate will be segmented on characters to be scanned by an optical characters recognition system (OCR). A number of image processing will be
applied to the system in order to have the best results in quick time},

{During management of a patient, many sources of information are provided to medical doctor. Imaging with different modalities provides structural and functional information. A step of combination and synthesis is required for precise diagnosis and adequate treatment. This step is long and tedious. To cope with this difficulty, multimodal fusion of images is a new tool for diagnosis. It’s a computer automated technique that replace the subjective mental fusion. In this work, we compare different techniques of fusion based on the pyramidal decomposition of images (Laplacian, FSD, Gradient and RATIO). Our study suggests that the best technique for brain tissue classification is based on the RATIO algorithm},