A Method of Restoring Highlight Area Based on Light Field Image

Abstract

A method for repairing highlight areas based on light field images, including: acquiring a four-dimensional light field image and a corresponding depth image; extracting a center viewpoint image from the four-dimensional light field image, and initially determining the spatial domain coordinates of the highlight target point, according to the input depth image Refocus the four-dimensional light field image to obtain the angular domain characteristics of the highlight target point and divide it into saturated highlight points and unsaturated highlight points; perform intrinsic image decomposition on the image from one viewpoint or multiple viewpoints to obtain the image's intrinsic reflection attributes , find the intrinsic reflection information corresponding to the highlight target point; for the unsaturated highlight point, use the local area characteristics under multi-viewpoints to separate the diffuse reflection component, and combine the intrinsic reflection information determined in step A3 to repair the unsaturated highlight point; A5: For saturated highlight points, use the diffuse reflection component of adjacent pixels to propagate, and combine the intrinsic reflection information determined in step A3 to repair the saturated highlight points. The present invention can be used to improve the quality of images containing high-gloss surfaces.

Description

technical field [0001] The invention relates to the fields of computer vision and digital image processing, in particular to a method for repairing highlight regions based on light field images. Background technique [0002] Highlights, also known as specular reflections. In the field of computer vision and pattern recognition, the highlights of images bring difficulties and challenges to the realization of many applications. Highlight is actually a very common phenomenon in real scenes. It is the change of the color and brightness of the surface of the object caused by the change of light under different viewing angles, reflecting the optical reflection characteristics of the surface of the object. In digital images, high-light pixels often have high brightness, thus covering the color, outline, and texture of the object surface. Saturated highlights directly lead to the loss of local area information, so highlights are usually considered as image defects. At present, man...

AI Technical Summary

Problems solved by technology

Since the traditional camera only records the information of one viewpoint when taking pictures, it can only focus to one depth, so most of the light information of the scene is lost

Images