Study On Key Algorithm Of Digital Image Mosaic

As a key technology in image rendering research area, Image mosaic is the most important and rapidly developed technology in the field of digital image processing and virtual reality. Image mosaic is a high-resolution image processing method, which is used to splice a series of images with overlapping areas in the same scene into a wide viewing angle image, and make it close to the original image, with little distortion and no obvious seams. The study of image mosaic has gradually developed for the demand of application in practice, and has become the focus of computer vision and computer graphics research. Image mosaic is widely used in digital video, image compression, virtual reality technology, medical image analysis and other fields.The quality of image mosaic depends mainly on the accuracy of image registration. As a result, image registration algorithm is the key point of image mosaic algorithm. There are main two kinds of image registration algorithms. One is algorithm based on region, which use the relationship of gray of two images to decide image coordinate parameter, including pixel matching method based on dimension, frequency-domain algorithm. The other is based on feature, which use distinct features to estimate the transform of images instead of all information from pictures, including Corner Detection algorithm, SIFT algorithm, and algorithm based on region features.Based on basic technologies of image mosaic, various of image registration algorithm are studied in this paper, in order to find different image mosaic algorithm that can be used in different conditions and improve the efficiency and accuracy of image mosaic.The main contents of this paper are as follows:1. The basic principles and processes of image stitching are studied in this paper, including image de-noising method and image gray value deviation of the amendment and geometric deformation of the amendment in image pre-processing, advantages and disadvantage of image registration method, image fusion methods, and a proper image fusion methods for image mosaic.2. A new image mosaic algorithm based on ratio template matching is presented . By introducing gradient factor to enhance the robustness of the template. it corrects the exposure difference in the stitching image, and applies the fade-in and fade-out method to make the stitched image seamless and smooth. Experimental results testify that this algorithm improves the reliability and increases the speed of matching and is easier to operate.3. A new image mosaic algorithm of regional characteristics matching based on fuzzy sets recognition method is presented in this paper. Using shape features such as perimeter, area, flattening and aspect ratio as regional features, introducing fuzzy sets recognition method, it is possible to identify the right matching points. Then image mosaic can be achieved by adopting 8 parameter projection conversion principle and fade-in-fade-out method. It is proved by experiments that this technique with strong robustness is easy to realize.4. With regard to the large amount of calculation in image mosaic algorithm, a image mosaic algorithm based on extended phase correlation of edge is presented in this paper. The iterative threshold segmentation approach is introduced to detect image edge. Then the image translation, rotation and scale changes are calculated by the extended phase correlation method. These parameters are utilized to stitch images. Finally, image fusion can be achieved by fade-in-fade-out method. It is proved by experiments that this algorithm is simple to calculate, and can effectively achieve the image mosaic.5. A new image mosaic algorithm based on Harris Corner Detection is presented in this paper, which use Harris operator to detect corners in images as well as correlation feature to identify matching pair, and restricted matching pair algorithm to get rid of false match and find proper matching points to stitch images. Finally, we use fade-in and fade-out method to achieve image fusion. Experiments have proved that this algorithm can effectively improve the accuracy and robustness of image mosaic.Based on the method of image stitching, algorithms in template matching, regional feature matching, phase correlation matching and focus matching image registration methods are improved respectively. It has been tested by experiments that the new algorithm can effectively improve the accuracy and robustness of image stitching.