Real-time Video Stitching Method Based on Improved SURF
Lei Fe · Computer Technology and Development · 2015
Video stitching has great significance in real life. In this paper,achieve the function that stitches the multi-channel videos to form a large field of view video in real-time. To make the system has good real-time feature,firstly isolate the video streaming by DirectShow and use an improved SURF algorithm for image registration. The improved SURF algorithm restricts the scope in the step of feature point extraction and improves the feature descriptor,which can greatly speed up the extraction. M atch feature points based on the nearest neighbor and second-closest neighbor distance ratio method,and then use RANSAC approach to purify and calculate the transformation matrix. Finally,only register the first frame of the video image and use the calculated transformation matrix to stitch the other frames,instead of registering and fusing each frame of image,which can greatly accelerate the speed of the video mosaic. The experimental results demonstrate that this method can generate the seamless and clear video with wide scene.