Super Resolution Reconstruction Of Ultrasonic Sequence Images Based On Improved Optical Flow Algorithm

The advantages of non-invasive,cheap price and real-time motion display of tissues and organs make the ultrasonic imaging to be one of the four imaging diagnostic technologies.The ultrasonic imaging has been the first selection for clinical application.However,the limitation of system cost and the ultrasonic characteristics of itself,the resolution of the ultrasonic images are low,so does the contrast of the organs and tissues.The image quality can't meet the requirements of the diagnosis of small lesions.Thus the disadvantage restricted the further development of ultrasonic diagnosis.For this matter,we can try to do the ultrasonic sequence images super-resolution reconstruction to enhance the useful information of the image.After analyzing the current situation of super-resolution reconstruction and the processing of ultrasonic imaging,based on the motion estimation for the ultrasound imges,this paper try to do some research on the ultrasonic sequence images super-resolution reconstruction to improve the quality of the images,so that they can show more details of tissues and organs.The research in this paper has important value for clinic applications.Based on the problems and research achievements of super-resolution reconstruction technique,this paper has accomplished the non-rigid motion estimation algorithm for ultrasound image sequences.By analizing the effect of pyramid Lucas Kanada optical flow algorithm,this paper used the optical flow algorithm followed the idea of nonlinear diffusion to reconstruct the discontinuity area in optical flow field,after the post-processing of the motion victors,the accuracy of motion estimation has been improved,so that the algorithm is more suitable for motion estimation of ultrasound image sequences.Based on the traditional iterative back project method and the research of ultrasonic image observation model,then,this paper used bicubic interpolation to realize the initial estimation of high resolution images,combined the motion vector matrix,we can reconstruct the high-resolution images,the filter processing is the last step.This paper adopts image simulation data,ultrasonic simulation data and the ultrasonic image data of human body to test the algorithms,the results showed that:the nonlinear diffusion optical flow algorithm is more suitable for the motion estimation,moreover,the vector denoising algorithm can effectively suppress noise;super resolution reconstruction algorithm based on optical flow can improve the resolution of ultrasonic images.Especially compared with the traditional IBP algorithm,this algorithm has better result.This paper used the algorithm for ultrasonic image datas,achieved tissue strain estimates,the result shows more details.