Research On Binocular Distance Measurement System Based On Humanoid Robot

With the continuous development of computer vision technology,binocular stereo vision technology based on the principle of parallax has been applied to various fields.At the same time,the robots have also gradually walked into people's lives,the research on humanoid robots technology have also attract extensive attention worldwide.The binocular vision system is a process of simulating human vision stereo perception,its principle is to observe the same object in the same scene from two different perspectives and to obtain two digital images in different perspectives,and to acquire three-dimensional object coordinate information which uses the related technologies.In order to realize the idea that humanoid robots can recognize complicated targets and measure the actual distance of the target,in view of some problems such as inaccurate matching,large amount of calculation,poor ranging accuracy and poor real-time performance of some binocular ranging methods.And a binocular method was proposed in this paper that is quick recognize target and measure distance which based on humanoid robots.This thesis mainly focuses on camera calibration,image acquisition,target recognition,stereo matching and distance measurement of these parts of the content of research and experimental verification.Firstly,the camera calibration was realized by the checkerboard method,and the basic information of the camera is acquired,and two parallel cameras are used to capture the image of the target object in space.Then the fast object recognition method based on multi-feature detection is used to identify the target object in the image,and the Hough transform is used to extract the shape features of the target object,the reverse projection algorithm is used to extract the color features of the target object,and the FAST(Features from Accelerated Segment Test)algorithm is used to extract the corner features of the target object.Next,On the basis of recognition,the SIFT(Scale-invariant Feature Transform)algorithm is used to extract and match the feature points,and the RANSAC(Random Sample Consensus)algorithm is used to eliminate the mismatched pairs.Finally,the actual distance is calculated based on the principle of triangulation.Using the method proposed in this paper,many experiments were carried out to compare the experimental data with the actual distance,and the analysis of error was give.Experimental results show that the proposed method based on humanoid robots recognition and the method of binocular distance measurement proposed in this paper have obvious improvement compared with the traditional methods in measurement accuracy and speed.To a certain extent,the accuracy and real-time performance of humanoid robot target recognition and binocular distance measurement have been improved.The work done in this paper has established a solid foundation for humanoid robots to realize higher difficulty motion and robot navigation.