Multi-object tracking is a challenging task, especially when the persistence of the identity of objects is required. In this paper, we propose an approach based on the detection and the recognition. To detect the moving objects, a background subtraction is employed. To solve the recognition problem, a classification system based on sparse representation is used. With an online dictionary learning, each detected object is classified according to the obtained sparse solution. Each column of the used dictionary contains a descriptor representing an object. Our main contribution is the representation of the moving object with a descriptor derived from a novel representation of its 2-D position and a histogram-based feature, improved by using the silhouette of this object. Experimental results show that the approach proposed for describing moving objects, combined with the classification system based on sparse representation provides a robust multi-object tracker in videos involving occlusions and illumination changes.

The compositing of videos is considered one of the most important steps on the post-production process. The compositing process combines several videos that may be recorded at different times or locations into a final one. Computer generated footages and visual effects are combined with real footages using video compositing techniques. High reality shots of many movies were introduced to the audience who cannot discover that those shots are not real. Many techniques are used for achieving high realistic results of video compositing. In this paper, a survey of video compositing techniques, a comparison among compositing techniques, and many examples for video compositing using existing techniques are presented.

The paper presents a design scheme to provide a faster implementation of multiplication of two signed or unsigned numbers. The proposed scheme uses modified booth's algorithm in conjunction with barrel shifters. It provides a uniform architecture which makes upgrading to a bigger multiplier much easier than other schemes. The verification of the proposed scheme is illustrated through implementation of 16x16 multiplier using ISIM simulator of Xilinx Design Suite ISE 14.2. The scheme is also mapped onto hardware using Xilinx Zynq 702 System on Chip. The performance is compared with existing schemes and it is found that the proposed scheme outperform in terms of delay.

Robotics has enabled the lessening of human intervention in most of the mission critical applications. For this to happen, the foremost requirement is the identification of objects and their classification. This study aims at building a humanoid robot capable of identifying objects based on the characters on their labels. Traditionally this is facilitated by the analysis of correlation value. However, only relying on this parameter is highly error-prone. This study enhances the efficiency of object identification by using image segmentation and thresholding methods. We have introduced a pre-processing stage for images while subjecting them to correlation coefficient test. It was found that the proposed method gave better recognition rates when compared to the conventional way of testing an image for correlation with another. The obtained results were statistically analysed using the ANOVA test suite. The correlation values with respect to the characters where then fed to the robot to uniquely identify a given image, pick the object using its arm and then place the object in the appropriate container. 

Handwritten character recognition complexity varies among different languages due to distinct shapes, strokes and number of characters. Numerous works in handwritten character recognition are available for English with respect to other major languages such as Bangla. Existing methods use distinct feature extraction techniques and various classification tools in their recognition schemes. Recently, Convolutional Neural Network (CNN) is found efficient for English handwritten character recognition. In this paper, a CNN based Bangla handwritten character recognition is investigated. The proposed method normalizes the written character images and then employ CNN to classify individual characters. It does not employ any feature extraction method like other related works. 20000 handwritten characters with different shapes and variations are used in this study. The proposed method is shown satisfactory recognition accuracy and outperformed some other prominent exiting methods.

Texture image retrieval plays a significant and important role in these days, especially in the era of big-data. The big-data is mainly represented by unstructured data like images, videos and messages etc. Efficient methods of image retrieval that reduces the complexity of the existing methods is need for the big-data era. The present paper proposes a new method of texture retrieval based on local binary pattern (LBP) approach. One of the main disadvantages of LBP is, it generates 256 different patterns on a 3x3 neighborhood and a method based on this for retrieval needs 256 comparisons which is very tedious and complex. The retrieval methods based on uniform LBP's which consists of 59 different patterns of LBP is also complex in nature. To overcome this, the present paper divided LBP into dual LBP's consisting four pixels. The present paper based on this dual LBP derived a 2-dimensional dual uniform LBP matrix (DULBPM) that contains only four entries. The texture image retrieval is performed using these four entries of DULBPM. The proposed method is evaluated on the animal fur, car, leaf and rubber textures. 

The scale of salient object in an image is not a known priori, therefore to detect salient objects accurately multiple scale analysis is used by saliency detection models. However, multiple scale analysis makes the saliency detection slow. Fast and accurate saliency detection is essential to obtain Region of Interest in image processing applications. This paper proposes a scale space reduction with interpolation to speed up the saliency detection. To demonstrate the concept, this method is integrated with Hypercomplex Fourier Transform saliency detection which reduced the computational complexity from O(N) to O(N/2).

The purpose of this paper is to encode a color video by wavelet transformation. Therefore, we propose a new hybrid approach which combines a fractional motion estimation technique. Several studies were carried out to reduce the spatial and temporal redundancies, hence at the level of spatial video coding, we use a new approach based on sub-bands coding through a discrete wavelet transformation. This technique is based on the principle of the EZW algorithm of Shapiro. It proceeds by separating the encoding of the signs and the magnitudes of wavelet coefficients. Then, at the level of temporal compression, we propose a study of motion estimation with different accuracy based on image interpolation to improve the quality of predicted frame. Next, we present a representation reducing the size of the motion vector field and we compress it by two of entropic coding approaches namely Huffman coding and arithmetic coding.
The proposed video codec was applied on a video sequence with different sizes (CIF and QCIF) and different dynamics. The obtained results, in terms of objective assessment (PSNR, the SSIM and VQM), were satisfactory compared with other video coding standards. We have also proposed a subjective evaluation and the results are compared to those obtained by H.264/AVC standard.