A Novel Method for Quality Assessment of Image Stitching Based on the Gabor Filtering
Shangdong Zhu, Yunzhou Zhang, Tao Liang, Tongbo Liu, Yanjiao Liu · 2018
Predicting perceptual quality is the main goal of the Image Quality Assessment (IQA) and image stitching is an exceedingly important branch in the field of computer vision, especially for panoramic maps. Nevertheless, there are few existing objective IQA methods suitable for the quality assessment of stitched images. In order to better reflect people’s subjective feelings and verify the validity of stitching and fusion algorithms in the reverse direction, an overall IQA framework is proposed in this paper based on characteristics of the image stitching and the Human Vision System (HVS), named Difference Information Sensing Model (DISM). The presented method first uses the Gabor filtering to detect the edge of the image, then constructs the difference image, and finally forms the assessment model combined with the Just Noticeable Difference (JND) threshold. Meanwhile, an improved edge detection algorithm is proposed, which filters out more high-frequency information and preserves the edge of the image skeleton. Furthermore, we developed an edge difference image generation algorithm which can obtain the butt seam, dislocations and brightness differences in the process of stitching. Experimental results demonstrate the efficacy of the proposed method outperforming prior work for quality assessment of image stitching.