PANORAMIC STITCHING ALGORITHM OF ORAL LATERAL CEPHALOGRAM SEQUENCE IMAGES BASED ON IMPROVED NORMALIZED CROSS-CORRELATION REGISTRATION
Dental lateral cephalogram imaging is performed by linear scanning with a small field plate detector.In order to obtain the lateral cephalogram panorama,this paper designs a stitching algorithm based on improved normalized cross-correlation registration.Gaussian mixture model was used to segment image preprocessing.The normalized cross-relation value in the region of interest was used as the measure function,and registration accuracy was improved by adopting multi-resolution strategy based on wavelet transform and particle swarm optimization.The image fusion was performed on the registered images,and a panoramic view of the oral cavity head side was obtained after multiple rounds of continuous stitching.The experimental results show that the registration error fluctuation of the improved algorithm is reduced from[-6,8]to[-1,1],and the registration result reaches the sub-pixel level.Subjective and objective analysis shows that the stitched panorama obtained by the proposed algorithm has no stitching gaps and ghosts,and the quality evaluation index is the best.
国家科技期刊平台
NSTL
万方数据
维普
