Image Super-Resolution Technology Applied In Multi-Aprture Imaging Systems

The imaging with WFOV(Wide field of view)and high resolution has always been an important issue for researchers in optical engineering.Designing a kind of optical system with high resolution and WFOV is often difficult.Even if it can be achieved,its system complexity and geometric size will also greatly increase.The rise of computational imaging technology gradually breaks the contradiction that the traditional optical system with WFOV and high resolution could not be attained at the same time.The multi-aperture imaging system depends on the splice of sub images of multi-aperture imaging system,by which the larger field of view can be effectively obtained.However,the pixel splice often wastes the aliased information of each sub aperture,which can contribute to enhancing image resolution.Therefore,through making reasonable use of aliased information,researching the image super-resolution technology of multi-aperture imaging system to realize the resolution promotion based on WFOV of multi-aperture system can break through the limitation of traditional optical design,which is of great significance.This paper mainly includes the following aspects about the image super-resolution technology of multi-aperture imaging system:In terms of theoretical image super-resolution analysis of multi-aperture imaging system,this paper puts forward the basic conception that the image registration which reaches to the precision with sub pixel level is the key to realize image super-resolution restoration.There are three aspects of the key technology of image super-resolution including image registration,establishing the model of image transformation and image re-sampling technology.Through analyzing the various existing key technologies as well as the advantages and disadvantages of their applications on the multi-aperture system,this paper establishes a set of image super-resolution algorithm which is more perfect for the multi-aperture imaging system.Through analyzing the actual multi-aperture prism imaging system,this paper puts forward an object-space transformation model based on ray tracing,by which good flat field correction for sub images can be achieved for sub images.In addition,a fast image fusion algorithm of multi-aperture imaging system can be proposed which is based on characteristics if this concept is combined with the image registration technology based on characteristics,and the splice for full field of view of multi-aperture prism imaging system can be realized if applying this algorith m to the actual system.On the basis of the above work,and directing at the imaging characteristic that the sub images in multi-aperture prism system do not conform to a certain transformation model,this paper designs a kind of multi-aperture image registration algorithm based on the amalgamation of region and characteristic.Applying this algorithm to practical application,and ultimately,analyzing and discussing this algorithm,we can successfully achieve the splice of sub-pixel level among each sub image.The final reconstruction result can be obtained by re-sampling the images,and the reconstructed image has higher resolution compared with the image acquisition.