Research On Depth Of Field Enhancement And De-Occlusion In Integrated Imaging

Three dimensional(3D)image and visualization has become a hot topic in recent years.Integral imaging(II)is a promising 3D imaging technology,which can provide full parallax and continuous viewpoint images,requires incoherent light,and needs no special glassed.However,due to the property limitation of imaging sensor,the II system often suffers low resolution,narrow viewing angle and limited depth of field.And when the 3D scene exists occlusion,the quality of 3D reconstruction image will be seriously degraded.In this paper,the depth of field enhancement and occlusion removal problems in integrated imaging are studied as follows:For the limitation problem of the depth of field(DOF)in integral imaging,the paper proposes a DOF enhancement system based on multi-focus images with different viewpoints.The system combines object extraction technology with image fusion technology.In this system,contour-based object extraction technology is used to extract objects from element images with different parallax information,which overcomes the limitations of unregistered images.In terms of efficiency,the redundant computation caused by matching the whole image is avoided,and the way of matching only on the contour greatly reduces the workload.In addition,because of the problem of size change of the object in the multi-focus element image,the size correction method based on the minimum boundary rectangle is used to adjust all the extracted objects to the same size.Then all target images are fused by the image-block-based image fusion technology to obtain a set of all-focus element images and finally carry out three-dimensional reconstruction.The innovation of this method lies in the realization of image fusion on multi-focus element images at different viewpoints,achieving the purpose of enhancing the depth of field in the integral imaging system.The experimental results verify the feasibility of this method.For the occlusion problem of the integral imaging technology,this paper proposes a new occlusion removal method based on epipolar plane image(EPI).In the process of extracting occlusion,the object extraction technique based on contour is used to extract the occlusion of element image.Due to the influence of different parallax,the size of three-dimensional object in element image will change slightly.And this method solves this problem well,and greatly reduces the computational cost.After extracting the occlusion object,the system extracts the EPI with abundant line texture information and uses the polar plane image to analyze the texture structure of the occlusion region.The method of EPI restoration based on spinning parallelogram operator is used to ensure the texture of the image by searching the lines of damaged pixels.Then the restoration of the damaged image after occlusion extraction is completed.Finally,We obtained a group of high-quality element images after occlusion removal and reconstructed them in an integrated imaging system to improve the image quality of 3D reconstructed images.