Research On Region-based Image Stitching Of Rotating Objects And Quality Assessment

Image stitching technology is an important technical tool in the field of image processing,and it has a wide range of application prospects in computer vision,medical images,defect detection,etc.At present,the commonly used image stitching methods are feature matching stitching method and deep learning stitching method.However,the feature matching method is susceptible to parallax and deformation in the process of image stitching on the surface of an object,while the deep learning method requires high training cost.In this paper,we use the region-based image stitching method to stitch the surface images of objects and evaluate the quality of the stitched images.The main work of the paper is as follows.First,a rotating table system is designed for acquiring object surface images.The object is placed in the center of the motor-driven controlled rotating table,and a smartphone camera is used to capture the video of the object rotation.Based on the hardware configuration of the rotating table and the camera frame rate,the frame images for stitching are selected from the video files,and the alignment area of the extracted images is determined by the statistical features of the image pixels.Next,a combined Wiener filter is designed to improve the blurring problem of the stitched images.The subjective quality assessment algorithm with average opinion score and the objective quality assessment algorithm with peak signal-to-noise ratio and structural similarity are used to assess the quality of the stitched images with different number of iterations.Finally,the objective quality assessment algorithm is used to study the effects of hardware system parameters such as the vertical distance from the object to the camera,the amount of object offset center,and the rotation angular velocity of the object on the quality of the stitched images.