Super-resolution Reconstruction System Based On Sequence Images

Image has already been the world leading data carrier.With the advent of AI's age,images can not only be visible and clear but also more distinguishability to meet people's need.And it's hard to find electronic products with the function of shooting not working well now in the market.However,we can't acquire more detail information from the image as for some restriction on physical condition sometimes.Super-resolution image reconstruction now can be used to produce contents beyond the cut off spatial frequency of the optics system so that images with higher resolution and higher definition can be seen.For that purpose,we study how to build a super-resolution image reconstruction system with sequence images.The essential principle to reconstruct with sequence images is that spatial resolution can be improved using time resolution.Making the assumption the we have acquired images with high quality and definition,there will be three steps to reconstruct higher resolution image.They are image matching,image interpolation and image reconstruction.The method of scale-invariant feature transform,SIFT in short is already being used in the first step.And we propose a new method for gray correction based on statistics according to bi-direction matching using SIFT method.In the last step,we make use of projection onto convexs method,POCS in short.In almost all literatures,the authentication of POCS theory is not enough.In this paper,we expound a sufficient demonstration upon all of those conclusion proved by previous research production.First of all,we introduce the research background,research purposes,research significance,basic conception and the development of super-resolution reconstruction.The processed experimental data in this paper taken from a real scene using a camera equipment with good quality.We want to generate an image with higher resolution.And the experiment result come from the system we built proved that we achieve the goal.This paper also offers a comprehensive view of system super-resolution reconstruction and points out the key problems have to be settled.Then the module of image match and the module of image reconstruction would be given more detail descriptions on those central concerns as to the improvement on them.The flow chart of system super-resolution reconstruction would be presented in the end with experimental evidence.It can be proved that we achieve our goal in this paper.