Research of Image Mosaic Technology
|School||Harbin Institute of Technology|
|Course||Instrument Science and Technology|
|Keywords||Image Mosaic Image Registration Image Fusion Harris Corner Detection Reject False Matches|
Image mosaic is a technology that carries on the spatial matching to a series of image which are overlapped with each other, and finally builds a seamless and high quality image which has high resolution and bid eyeshot compared with a single image. With the rapid development of digital imaging technology, digital imaging equipments have been widely used. However, in some situations, the digital image and video can not met the demands because of the limited field of view. This paper studies the digital image mosaic technology to generate large field-of-view image from image sequences.The thesis first introduces the present research status of image mosaics technology, and the characteristics and application domains of mosaics, thus has demonstrated the broad application prospects of image mosaics technology. Then the thesis discusses the basic steps of image mosaic, image acquisition methods and models of image mosaic. Moreover, it researches with emphasis in the two core technology of image mosaic, which are the image registration technology and the image fusion technology. The common image registration methods and image fusion methods are introduced, and their performance as well as the existing problems is analyzed.The methods of image registration can be classified into two categories: the intensity-based matching approach and the feature-based matching approach. In this paper, the characteristics and the application fields of these two common methods are dissertated, and the feature-based matching approach is discussed as a main point. As one of the most common methods, the largest advantage of the feature-based approach is its ability of translating the analysis of the whole image into its features that contains the feature point, the feature curve, etc. And as a result, it speeds up the image process. In this paper, the current feature detection methods are discussed, and the common point-feature detection methods are emphasized. Finally, after the precise and fast feature-point detection, improved some the effective algorithms based on Harris point-matching.The intensity information and position information of interest point is utilized in this algorithm. Interest points are firstly extracted by the Harris ankle detector in this method, and initial matches are extracted by the method of cross correlation, then the false matches are rejected according to the matching degrees of feature points and count of Projection transform parameters. Transform parameters are determined by the matches, and changing weighted average method is used to blend images. The experimental results prove that the method can significantly reduce the time of matching and solve the problem of joint. The proposed algorithm is accurate in matching and having good capabilities. The paper will be helpful to future research about this area.