Applied Mechanics and Materials Vols. 263-266

Abstract: At present, objects detection, identification and tracking according to machine vision technology, have been widely used in various economic aspects. In this paper, a real-time monitoring system was discussed. This surveillance system mainly combines two sub-systems, motion detection and objects tracking. An Adaptive Gaussian Background Model was established in order to automatically update the background and detect the outline of moving objects. By analyzing different algorithms, this paper brings out approaches to promote the performance.We proposed CamShift algorithm to complete motion detection and objects tracking, which applied for static background video sequences. And the experimental results show that our method can achieve the pre-determined targets.

Abstract: This paper presents a simple method of circle pose estimation based on binocular stereo vision. It takes the projective equation of a circle as the basis, and gives the closed form solution of the pose parameters. Since there are two possible sets of pose parameters for a circle from one calibrated perspective view, the stereo vision constraints are incorporated and the accurate pose parameters are determined. Experiments using computer simulated data and real data demonstrate the robustness and accuracy of our method.

Abstract: In the projection system, the projected image should be able to meet the normal visual needs of observer, so that the viewer can get the correct projection. When the projection screen surface is the irregular free surface, the projection geometric correction of the projected image becomes the research focus. For the question of irregular projection screen surface, this paper presents effective geometric correction method. The geometric correction method based on the geometry of irregular projection screen surface and ideal location of viewport to calculate the points’ three-dimensional coordinate values. In this paper, the projection geometric correction method is widely applicable to various projection screen surface environments.

Abstract: In this paper, we present a comparison of the scale-invariant feature transforms (SIFT)-based feature-matching scheme and the speeded up robust features (SURF)-based feature-matching scheme in the field of vehicle logo recognition. We capture a set of logo images which are varied in illumination, blur, scale, and rotation. Six kinds of vehicle logo training set are formed using 25 images in average and the rest images are used to form the testing set. The Logo Recognition system that we programmed indicates a high recognition rate of the same kind of query images through adjusting different parameters.

Abstract: This paper describes the techniques used in visual based hand gesture recognition systems. The study is discussed from three aspects: the two categories, the five components, and the methods of feature extraction of visual based hand gesture recognition systems. The two categories are 3D model based systems and appearance model based systems. The five components are image sequences capture, pre-processing, hand regions detection, feature extraction and gesture classification. The methods of feature extraction are Hidden Markov Model (HMM), Artificial Neural Networks (ANN), and Support Vector Machines (SVM). The main ideas of each technique are described in detail.

Abstract: Data association problem has been an important issue for the multiple vehicles tracking in a vehicle tracking system. In this paper, we present an efficient data association method to track multiple vehicles in a sequence of traffic video frames. We first introduce the compact rectangular region-of-interest (crROI) that tightly encloses a vehicle and has the rotation-invariant property. The subsequent processing is based on the crROI instead of a vehicle image itself to avoid the processing overhead. Next, we extract the features from the crROI such as shape, size, and spatial relationship. Using these features, we define the similarity metric between two vehicles, and present the association method that matches a vehicle in a frame with the corresponding vehicle in its consecutive frame. An experimental result shows that the proposed method identifies and tracks vehicles effectively and efficiently in the curve or crossroad environment where multiple vehicles appear.

Abstract: By studying the principles of the mean filter, we obtain a class of partial differential equations based on mean filter, which used for image zooming. Experimental results show that our method is feasible. It also confirmed the importance of the partial differential equations in image processing.

Abstract: Image segmentation is an important processing step for finger-vein identification technique. But it’s difficult to extract precise details because of the irregular noise and shades around finger-vein. In this paper, a threshold image method is proposed for finger-vein segmentation. The new method computes thresholds for every pixel, so every pixel has a best threshold to get good segmentation. In order to compute the thresholds robustly and rapidly, we develop an optimal OTSU method. Firstly, the pixels with larger brightness than mean brightness are set to background directly. Then, the brightness is confined in a small region when computing threshold by using an optimal OTSU method. The experiments show that the proposed method can obtain good segmentation of finger vein images while costing small time consuming.

Abstract: Brain surgery is generally guided by brain anatomical image, tumor removal maybe injure patients’ important tissues and functional areas, and result in death and permanent disability, these important tissue and functional areas are invisible in the anatomical image. This paper presents an image fusion software system, which can merge lesion, important tissues, brain functional image, brain atlas, fiber tract into anatomical image, and show them in 3D image. With the help of this system, surgeons can avoid important tissues and functional areas when they design surgical approach, they can also minimize intraoperative risk and postoperative deformity by the guidance of fusion image. Experiments show that the image fusion system is feasible and applicable to surgery.

Abstract: This paper studies the application of surface fitting technology in medical tomography. In clinical research, in the result of early detection of cancer increasingly critical, the accuracy and efficiency become the most important issues in medical tomography. However, recent research on medical tomography only use curve fitting technology and only can fit single point heat source, which cause the analyzed result not accurate enough and complex the process of medical tomography. To improve the problem, surface fitting technology is applied in medical tomography, and then information of multiple heat sources can be directly fitted out. The experiments have confirmed: using Custom Equation fitting method in the Curve Fitting Toolbox of MATLAB can accurately fit out heats as well as locations of multiple heat sources (information of diseases area). The result improves the accuracy and efficiency of the Tomography and provides the base for the design.