Depth image completion method and device

Abstract

The invention provides a depth image completion method and device. The method comprises the steps of obtaining an original color image and a corresponding original depth image; obtaining a first depthimage by using a first depth neural network based on the original color image; obtaining a second depth image by using a second depth neural network based on the original depth image and intermediatefeature images generated by the intermediate layers of the first depth neural network; and combining the first depth image and the second depth image to obtain a final depth image.

Description

technical field [0001] The present disclosure relates to the field of image processing, and more specifically, to a method and device for depth image complementation. Background technique [0002] High-quality and complete depth image information plays a vital role in many applications based on depth information, such as 3D reconstruction, autonomous driving, augmented reality, robotics, etc. However, the current consumer-level depth cameras have problems such as poor image quality, sparse depth images, or missing depth values ​​such as holes. For this existing problem, the current depth map completion algorithms are mainly divided into two categories, one is the traditional method based on filtering, and the other is the deep learning method that fills the depth value by building a regression model. [0003] The traditional method is mainly based on filtering and Markov random field model to expand and fill the depth image and constrain it with the help of texture informat...

AI Technical Summary

Problems solved by technology

This type of method has blurred output images, unclear edges, and unsatisfactory effects on edge parts and large-scale depth missing parts.

Images