Focusing on the application problem according to feature matching and panoramic image stitching, comparing direct stitching and smoothing after feature matching algorithms. Through time performance, synthesis accuracy, and synthesis efficiency, comprehensively compared their effects of synthesizing panoramic images to find a better solution for panoramic splicing. In the study, there are three groups of image pairs with parallax and brightness contrast, then use SIFT algorithm to conduct feature point and matching feature point. The RANSAC algorithm obtains the homograph matrix and completes the panoramic image direct splicing and smoothing test through the matrix transformation. Because of the luminance difference, the panoramic picture after direct splicing will show more obvious light and dark. Though the smoothing treatment could be performed rigidly in the light and dark difference, the deformation, and heavy shadow are easier produced when the disparity and brightness difference are large.