Advanced Materials Research Vols. 945-949

Abstract: In order to improve the real-time performance and accuracy of the vehicle type recognition in real traffic scenes, a novel method of vehicle type recognition based on template matching is proposed. First, points are adopted with non-uniform sampling based on salient regions in contours of vehicle. Then, parallel template matching is introduced to realize the rapid vehicle type recognition through lookup tables. The results show that the algorithm can recognize vehicle type in traffic scenes efficiently.

Abstract: The purpose of this paper is to research application of speed-up robust feature (SURF) based on the region of interest for workpiece matching and positioning. Thresholding is a simple but important method to perform image segmentation. In order to reduces the complexity of the data and simplifies the process of recognition, the image is segmented by threshold value method which eliminates and suppresses useless information of image background. The image matching algorithm shows a better performance on real-time than the standard SURF and it succeeds in accelerating the speed of image pre-processing before image matching. In addition, the good robustness and adaptability of SURF are maintained. Compared with the traditional algorithm, improved algorithm enhances the efficiency of vision inspection system and can be used in other applications of image matching.

Abstract: For multiple objects tracking in complex scenes, this paper proposes a new tracking algorithm for multiple moving objects. The algorithm makes likelihood calculation by using new DG_CENTRIST feature and color feature, and then calculates the overlapping ratio of the tracking object and the object in the current frame using coincidence degree to measure the occlusion. The algorithm has good robustness and stability, and the experiment results show that this method can effectively improve the accuracy of the multiple target tracking.

Abstract: In order to improve the accuracy of people counting in video surveillance, the method for people counting based on the analysis of the mass is proposed. The novel algorithm of objects tracking is designed to aim at people counting, and the people counting model is obtained by training a support vector machine (SVM) classifier with the input of the feature of mass. The experimental results show that the accuracy of counting is over 93%.

Abstract: To overcome illumination changes and pose variations, a pose-invariant face detection method is presented. First, an illumination compensation method based on reference white is presented to overcome the lighting variations. The reference white is obtained according to the component Y from YCbCr color space. Then, a mixture face model is constructed by the Cb and Cr from YCbCr color space and H from the HSV color space to extract faces from colorful image. At last, an eyes model is designed to locate eyes in the obtained face images, which can distinguish face from neck and arms ultimately. The presented method is conducted on the CASIA face database. The experimental results have shown that our method is robust to pose changes and illumination variations, and it can achieve well performance.

Abstract: This paper has stated the theory of wavelet threshold denoising , combining the theory analysis and simulation results,the paper discusses several kinds of factors which affect the denoising capability in a complete denoising algorithm.That provides the date reference of threshold denoising methods in actual image process.

Abstract: The traditional sports training coach is one-to-one or one-to-many face-to-face training, it not only take the limitation of time and place, but also the high cost of training, coach resources nervous. In order to solve this problem, a growing number of professional sports training or national training began to apply motion capture to sports training. Kinect through the accurate grasp athlete body contour and body position to determine the position player action, and action corresponding to the role of sports training or operation. This article is based on motion capture device, combined with the feature of sports activities, to provide effective solutions for sports training, so as to enhance the level of sport athletes.

Abstract: The technique of wavelet transform is a good method for segmentation of complex image edge. This paper applied Two-B spline wavelet to identify the image edge and removed the false edges better. Based on edge detection,the system implemented the computing on rocks area and perimeter with Pixel statistics arithmetics.

Abstract: Image segmentation is one of the basic problems of image processing, also is the first essential and fundamental issue in the solar image analysis and pattern recognition. This paper summarizes systematically on the image segmentation techniques in the solar image retrieval and the recent applications of image segmentation. Then the merits and demerits of each method are discussed in this paper, in this way we can combine some methods for image segmentation to reach the better effects in astronomy. Finally, according to the characteristics of the solar image itself, the more appropriate image segmentation methods are summed up, and some remarks on the prospects and development of image segmentation are presented.

Abstract: Adopting the Verification Methodology Manual’s (VMM) hierarchical structure, this paper presents a design of available verification platform based on System Verilog adopted. The platform completed can implement constrained-random test, directed test, and error stimulus test with high efficiency; moreover, gain maximum code reuse. Using Direct Programming Interface (DPI), the verification platform can conveniently link C++ with the model that realized the function of Design Under Test (DUT), and then to test it. At last the paper shows the experiment results to prove the effectiveness and practicality of the platform by verification sample of HOG chip.