Research On Automatic Calibration Method Of Depth Camera

With the increasing popularity of artificial intelligence technology,depth vision has set off a wave of visual revolution,and will certainly become the core content of the field of machine vision.Depth camera has many advantages,such as real-time depth information,not affected by illumination,small size and so on.It has gradually become an indispensable hardware condition in the manufacturing and scientific research process.But the key prerequisite for achieving all these is that the depth camera can complete accurate and fast calibration.Based on the principle of camera depth measurement and self-calibration technology,two methods of automatic calibration of depth camera with simple operation and accurate and reliable results are proposed in this paper.Firstly,based on the research background and current research situation at home and abroad,we analyze the shortcomings of traditional calibration methods,clarify the research significance of depth camera calibration,and introduce the important geometric principles related to camera calibration technology,which paves the way for the follow-up research work.Secondly,based on the classical calibration theory,we propose a method of automatic calibration of depth cameras,which uses the corresponding relationship between three-dimensional feature points and pixels in depth maps to establish the mathematical model of depth cameras to solve the initial values of the camera's internal and external parameters,and obtains the optimal values of camera parameters through general optimization.Then,aiming at the shortcomings of the three-dimensional feature point calibration method,such as poor real-time performance and poor global search ability of the general non-linear optimization method,we propose an improved depth camera self-calibration algorithm based on homography matrix,and calculate the distortion parameters by establishing the distortion model of the depth camera,and determine the optimal internal and external parameters of the depth camera by combining the immune clone selection optimization algorithm.Finally,we build a visual experiment platform,use Microsoft Kinect V2 depth camera to obtain depth scene map,use Python language to write calibration program for calibration experiments,and do comparative experiments of classical calibration methods.The accuracy and practicability of the calibration algorithm proposed in this paper are proved by the comparison and analysis of the experimental results.