Research On Panorama Synthesis Technology

Panorama is not only widely used in modern economic life, including atmospheric monitoring, satellite remote sensing, television advertising and 3D scene reconstruction, etc., but also a hot research issue in the fields of computer vision, image processing and information fusion. Under the support of the funds of Fujian Science and Technology Department(2011H0027) and Fujian Natural Science Foundation (2011J01268), we study some related technologies of panorama synthesis and use video images to synthesize panoramic images. Finally, we design and implement a set of software of video panorama synthesis. The details are as follows:(1) Research on adaptive color image registration algorithm. This research is based on improved SURF descriptor. As one of the key technologies of panorama synthesis, image registration, whose accuracy determines greatly the effect of image fusion and the final effect of panorama synthesis, is the foundation of panorama synthesis. Considering that the SURF algorithm does not support color image registration and the existing image matching algorithm performs ineffectively in many general cases, this study proposes a color image adaptive registration algorithm based on improved SURF descriptor. Experiments proves that the proposed algorithm not only preserves the prominent performance of the SURF algorithm and increases the registration capability of color image, but also greatly improves the adaptive matching ability and the accuracy of registration.(2) Research on image fusion algorithm of panoramic image stitching. This research is based on contrast pyramid and Contourlet transform. Considering the characteristics of panorama stitching, this paper proposes an image fusion algorithm which is based on contrast pyramid decomposing and combines color space conversion and Contourlet transform in order to attain smooth and continuous effect of panorama stitching. Experiments proves that the proposed algorithm takes good advantages the contour features of Contourlet transformation, presents the detail information of images well and achieves good effects on panorama seamless synthesis.(3) Based on the above researches, we design and implement a set of software of video panorama synthesis by studying the synthesis technologies of panorama and applying the research achievements. This software can not only open and play a variety of formats of video files, such as AVI, MPEG, etc., but also extract images from the playing video by manual interaction. Moreover, this software can synthesize cylindrical panoramas and planar panoramas through internal programs automatically. The crop tool in the software can crop those synthetic panoramic images preliminarily, and then those can be saved and exported as "jpg", ".bmp" and ".png" files by the software.By analysing the related technologies of panorama synthesis, this research has designed and implemented a new kind of image registration algorithm and image fusion algorithm and further developed a video panorama synthesis software applying the Opencv and Directshow on VS2010 platform. The experiments show that the video panorama synthesis software is of great practical value.