Research On Method Of Viewpoint Generation And Hole Filling Based On One-dimensional Integrated Imaging

Compared to traditional integrated imaging,the one-dimensional integrated imaging system abandons the stereoscopic effect in the vertical direction,leaving only the stereoscopic effect in the horizontal direction.While satisfying the stereoscopic perception,it greatly reduces the storage space of data and alleviates the large attenuation of resolution caused by integrated imaging.However,at present,there is no 3D film source suitable for a one-dimensional integrated imaging system.Therefore,the scope of application and field of application of this method are greatly limited.In addition,when generating a stereo film source,we often need to identify and separate objects in the background and screen for matching points.This paper proposes an array map generation and filling algorithm based on a onedimensional integrated imaging system.This algorithm fully absorbs the idea of DIBR,and uses optical flow method and Criminisi image repair algorithm to fill the array map;Secondly,this paper also uses histogram equalization,edge detection,and region growing methods to separate foreground objects from the orange background;Finally,this paper uses the cooccurrence matrix and the sum of squared difference(SSD)algorithm to frame the foreground objects in the video to achieve the purpose of filtering out redundant matching points.The research contents of this paper include:We propose an array map generation and filling algorithm based on one-dimensional integrated imaging.The principles of optical flow method and Criminisi image restoration algorithm are introduced and compared in detail,And we use the above two algorithms to fill the array image with holes,and generate a one-dimensional integrated imaging viewpoint image with any number of viewpoints,which solves the problem of the lack of 3D film sources in one-dimensional integrated imaging.We propose a method for front background segmentation.We use histogram equalization,edge detection algorithms,and region growing methods to segment foreground objects,while referring to the grayscale information and color information of the image to make the segmentation more accurate.We propose a method for frame selection and matching point screening of foreground objects.This frame selection algorithm uses the co-occurrence matrix and the sum of squared difference(SSD)algorithm to select the frame of the object,which is not susceptible to the influence of large blocks of color.Using this algorithm to screen the matching points of the object,we can effectively eliminate the matching point pairs that we don't need.In this paper,we use 20 frames of video images divided into 10 groups to verify the effectiveness of the algorithm.