A shadow detection and elimination method for complex scene video

Abstract

The invention discloses a video shadow detection and elimination method based on depth information. First, the depth information of the image is used to estimate the normal information and point cloud position information of each pixel point. By comparing each pixel point and its spatio-temporal local part in the video stream, The feature similarity between neighborhood pixels is used to estimate the shadow confidence value of each pixel, and the Laplacian operator is used to further optimize the shadow confidence to obtain the final shadow detection result. Finally, the shadow detection result is used to construct a The illumination recovery optimization equation of the video flow is used to obtain the final shadow elimination result. The present invention has the following advantages: it uses texture filtering to effectively reduce the interference of texture information on shadow detection, uses the Laplacian operator to optimize the initial shadow confidence, and obtains more complete shadow detection results, and uses chromaticity constraints and previous and subsequent frames. Correlation eliminates shadows, which can effectively ensure the chroma invariance and inter-frame continuity of the results.

Description

technical field [0001] The invention belongs to the technical field of video processing, in particular to a method for detecting and eliminating video shadows in complex scenes. Background technique [0002] Shadows are common natural phenomena in our daily life. They can provide important information for the understanding of visual scenes, such as lighting environment, scene geometry and other information. This information plays an important role in applications such as lighting analysis, relighting, and augmented reality. Therefore, effectively detecting and removing shadows is an important topic in the field of computer vision. However, automatic detection and removal of shadows is a very difficult task, which is not only affected by local texture and material information, but also needs to consider the global structure information and lighting environment information in the scene. Most of the existing shadow detection and elimination algorithms are based on local chrom...

AI Technical Summary

Problems solved by technology

There are two main reasons why it is difficult to eliminate shadows in complex scenes. First, the material texture information in complex scenes is relatively rich, and the distribution of shadows is also messy and not concentrated, which increases the difficulty of detection. Even with the prior knowledge of human interaction, complex The shadow scene will increase the burden of labeling, and it is difficult to perform efficient batch processing; secondly, due to the difficulty of labeling and the lack of corresponding data sets, it is difficult to use deep learning methods to eliminate shadows in large scenes.

Images