To determine the position of the subastral point in the base map and reduce the computational burden, an approach was developed based on scale invariant feature transform (SIFT) and match between salient regions. The salient regions of the base map can be determined by manual or algorithm in advance. And also, the salient regions of the descent image can be obtained by saliency computation. The extraction of SIFT feature was only performed on the salient regions of the base map and the descent image. These feature points were used to match between the two images. The method of maximum likelihood estimation sample consensus (MLESAC) was employed to eliminate the wrong matches. Then the correct matching points were used to determine the transform matrix between the base map and the descent image. The position of the probe can be predefined in the descent image. Through the transform matrix, the position of the subastral point can be determined in the base map. The experimental results demonstrate that the proposed approach can determine the position of the subastral point simply by matching in the salient regions rather than traversing the entire image to search for the matching points so as to reduce the cost of comparing all regions in two images.
In order to compare the effects of traditional LBP operators and various improved Local Binary Patterns (LBP) operators’ ability on feature extraction and classification. In this paper, the LBP feature is extracted by slidin g window method. The integrity of the texture information extracted by different LBP operators and the com putational complexity and detection effect are compared. The effects of sliding window parameters on the de tection effect is studied respectively.
Moving object detection is a major research direction of video surveillance systems. This paper proposes a novel approach for moving object detection by fusing information from the laser scanner and infrared camera. First, in accordance with the feature of laser scanner data, we apply robust principal component analysis (RPCA) to studying moving object detection. Then the depth and angle information of moving objects is mapped to the infrared image pixels so as to obtain the regions of interest (ROI). Finally, moving objects can be recognized by making investigation of the ROI. Experimental results show that this method has good real-time performance and accuracy.