Test method and test system based on depth estimation performance of depth camera

Abstract

The invention discloses a test method and a test system based on depth estimation performance of a depth camera. The test method comprises the following steps of constructing a scene, and erecting thedepth camera and a laser total station; storing the color image and the depth image; the laser total station scanning the field to obtain a projected color image and a depth map; evaluating an overlap degree projection error between the projected color image and a color image acquired by a camera; if the projection error does not meet the requirement, carrying out calibration again; and if a projection error meets the requirement, comparing and evaluating the depth map acquired by the camera and the depth map projected by the total station pixel by pixel, and calculating the error condition of depth estimation of the depth camera. The method is advantaged in that the true value of the depth map is obtained through field scanning of the laser total station, joint calibration of the cameraand the total station and point cloud back projection, more comprehensive evaluation is conducted by directly comparing the depth map, meanwhile, rich scenes are provided, physical equipment is not limited, and higher precision is achieved.

Description

technical field [0001] The invention relates to the technical field of sensor testing, in particular to a testing method and testing system based on the depth estimation performance of a depth camera. Background technique [0002] In recent years, a depth camera refers to a visual sensor with depth measurement capabilities. It has been widely used in unmanned systems such as robots, drones, and unmanned vehicles. The depth camera obtains the depth value of each pixel in the image. It is mainly divided into three methods: binocular, structured light, and light time of flight to form a depth map containing depth information. [0003] At present, the evaluation of the depth estimation performance of the depth camera often adopts the test and evaluation of a single feature point or the test and evaluation based on the calibration board. The test and evaluation based on a single feature point has a certain degree of one-sidedness and it is difficult to reflect the performance of ...

AI Technical Summary

Problems solved by technology

[0003] At present, the evaluation of the depth estimation performance of the depth camera often adopts the test and evaluation of a single feature point or the test and evaluation based on the calibration board. The test and evaluation based on a single feature point has a certain degree of one-sidedness and it is difficult to reflect the performance of the depth estimation of the entire image. Based on The test and evaluation method of the calibration board has the problem of a single scene and cannot well represent the depth estimation performance in actual use.

Images