Research And Implementation On The Key Technologies Of Image Mosaic

With the rapid development of digital image technology in recent years,the demand for high-definition and wide-angle image has been also on the increase. However, the image collection device used for shooting superwide-angle image is quite expensive while the image is also subject to a high level of distortion. As this reason, some people have proposed to acquire the wide-angle and high-definition integral image by computer.Starting from the flowchart of image mosaic technique, this paper has given a complete introduction of these steps and relevant key technologies necessary for image mosaic. What is more, this paper has also analyzed and compared the principles of those classical algorithms in the field of image mosaic. Lastly, this paper has carried out the simulated realization of the three classical algorithms with MATLAB software,which namely SIFT feature detection algorithm, SURF feature detection algorithm and HARRIS corner detection algorithm.On the basis of the study of traditional image mosaic technique, this paper has looked deep into a brand-new image mosaic technique which is known as Half-Projective transform with low-distortion image mosaic algorithm. This algorithm is a spatial combination of a projective transformation and a similarity transformation. The projective transformation is used to achieve the accurate alignment while the similarity transformation aims at the independency between images.Because the two transformations are based on the progressive mode, there will not be any obvious seam in the image. In the simulation experiment,the new and old mosaic techniques have been compared. It has beenproved that the new algorithm can enlarge the imaging view without causing serious image distortion.