Font Size: a A A

Photoelectric Search With An Image Motion Compensation System

Posted on:2015-02-18Degree:MasterType:Thesis
Country:ChinaCandidate:X ZhouFull Text:PDF
GTID:2268330425987718Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
Image stabilization technology is widely used in mobile robotics,military reconnaissance, security and surveillance, precision-guided weapons field, intelligent vehicles and other field. Image stabilization technology have developed from mechanical image stabilization technology, optical image stabilization technology, the electrical and mechanical image stabilization technology, to the electronic image stabilization technology, which is rapidly developed in recent years. With the development of computer technology and large-scale integrated circuits, electronic image stabilization technology has become the inevitable trend.In the image motion compensation algorithms, the block matching method is the most used method. However, the block matching method has the block effect. This task research on the motion detection and tracking of targets in the aerial video, which focuses on the CCHKLT algorithm. In the CCHKLT algorithm, the corner detection is used in the image to get the corner sets in the image. Using this corner sets and the KLT optical flow tracking algorithm in next images, and get the corresponding corner sets. Match the two image sets of the before and after frames, and then calculated homography relation between two frames. Finally, this method derives to the transform relationship between the initial reference frame and the current frame. In this method,correcting the transformation matrix properly is a must. When the rotation angle corresponding to the transition matrix reaches a certain threshold, we will correct the transition matrix of the images. Then use this image as the reference frame, continue to repeat the above algorithm, until the end of the video image. After CCHKLT algorithm, the difference between the reference image and the motion-compensated inter-frame image is reduced. And then use this algorithm in the moving targets detection and tracking. In practice, we found that CCHKLT algorithm is more precise and faster than phase method and method handled directly, and suitable for the use in translation, rotation and scaling conditions.
Keywords/Search Tags:Motion compensation, filtering, corner detection, search, track
PDF Full Text Request
Related items