The Research of Image Mosaic Techniques Based on Optimized SIFT Algorithm

Image mosaic refers to the process of stitching multiple images those have overlapping areas of small perspective and low resolution into a panoramic image with high resolution and wide perspective through the corresponding image registration and fusion algorithm. In the mosaic of panoramic images, the traditional SIFT algorithm has large amount of calculation that leads to mismatching and unsatisfactory splicing effect in the process of generating feature vectors and performing feature matching. To this end, this paper proposes an optimized SIFT algorithm. The optimization algorithm, at the first time, introduces the Laplacian operator in order to sharpen the edges of the image. Then, based on the SIFT algorithm, matching the feature points by bidirectional matching algorithm. Finally, in the part of image fusion, an algorithm of luminance weight fusion in HSI color space is proposed. Experiments show that compared with the traditional SIFT algorithm, the proposed optimization algorithm can effectively reduce the error matching and improve the matching accuracy of feature points. In the image fusion part, the phenomenon of ghost image and the sudden change of luminance during image mosaic is effectively eliminated, besides the fusion effect is optimized, and ends with a good image mosaic result.