Method and device for generating target depth information in real time

Abstract

The invention provides a method for generating target depth information in real time. The method comprises the following steps: S1, acquiring a target diffraction stray light spot image sequence of a target in the field of vision; S2, getting a plurality of original first frames of target depth maps, comparing each original first frame of target depth map with the deviation value of the known depth of a corresponding reference diffraction stray light spot image, and selecting the depth information of a corresponding region when the deviation value is the minimum to obtain the optimized first frames of target depth maps; S3, carrying out diffraction stray light spot cross-correlation matching processing on a reference diffraction stray light spot image block and a second frame of target diffraction stray light spot image to obtain a second frame of target depth map; and S4, getting a next frame of target depth map. The invention further provides a device using the method for generating target depth information in real time. The method and the device have the following beneficial effects: the distance between a camera and a person or other targets in the field of vision can be acquired in real time, the depth information can be updated in real time, and the device can be applied to a motion capture system, a machine vision system or the like to be used as information input of a specific application system.

Description

technical field [0001] The invention relates to image processing technology, in particular to a method and device for generating target depth information in real time. Background technique [0002] With the development of 3D technology, more and more technical applications such as stereoscopic display, machine vision, and satellite remote sensing need to obtain the depth information of the scene. Currently, depth cameras are capable of obtaining depth information of objects within the camera's field of view. [0003] There are three main architectures for depth cameras, namely binocular stereo vision method, time-of-flight (TOF) method and structured light method. The binocular stereo vision method uses the stereo vision of two lenses to obtain depth information about the distance between the surface of the object and the camera. The time-of-flight method uses the principle of radar measurement to measure the depth of the three-dimensional scene. The camera emits active li...

AI Technical Summary

Problems solved by technology

[0005] (1) The generation speed of depth information is slow, and real-time acquisition cannot be realized;

[0006] (2) The accuracy of the obtained depth information cannot reach the ideal accuracy effect, and the edge processing is not good, for example, where the target depth data is discontinuous;

[0007] (3) Application scenarios, the measurement range is too limited to meet the needs of some applications

Images