Depth image generation method and apparatus

Abstract

Embodiments of the invention disclose a depth image generation method and apparatus. A specific embodiment of the method comprises the steps of obtaining a target depth image; inputting the target depth image to a pre-trained convolutional neural network, thereby generating a restored image of the target depth image, wherein the convolutional neural network is obtained by the following training steps of obtaining a training sample set, wherein each training sample comprises a first sample depth image and a second sample depth image obtained by shooting a same object by a first depth camera anda second depth camera, and the second depth camera is superior to the first depth camera in at least one preset depth camera performance parameter; by taking the first sample depth image and the second sample depth image in the training sample of the training sample set as an input and an expected output of an initial convolutional neural network respectively, training the initial convolutional neural network; and determining the initial convolutional neural network obtained by training as the pre-trained convolutional neural network. Therefore, the restored image of the depth image is generated.

Description

technical field [0001] The embodiments of the present application relate to the field of computer technology, in particular to the field of computer vision technology, and in particular to a method and device for generating a depth image. Background technique [0002] In computer vision systems, 3D scene information provides more possibilities for various computer vision applications such as image segmentation, target detection, and object tracking. a wide range of applications. The gray value of each pixel in the depth image can be used to represent the distance from a certain point in the scene to the shooting device. [0003] At present, most of the depth images are acquired through depth sensors. However, the cost of depth sensors that can capture better quality depth images is relatively high, and high-quality depth sensors are not available on many occasions, and the depth images captured by ordinary depth sensors may often have low pixels and low signal noise. Ther...

AI Technical Summary

Problems solved by technology

However, the cost of depth sensors that can capture better quality depth images is relatively high, and high-quality depth sensors are not available on many occasions, and the depth images captur

Images