Research On Brush Writing Robot System Based On Machine Vision

With the rapid development of robotics technology,robots have become a trend in various industries to replace humans.In order to inherit and promote traditional Chinese culture,this thesis proposes a brush writing robot system based on machine vision.By collecting the image of the text to be copied,the characteristics of the text to be copied are extracted,and the robot is guided to complete the writing,and the similarity of the writing result is evaluated.Through the combination of modern technology and traditional culture,the inheritance and promotion of brush writing is realized.The main research contents of this thesis are as follows:(1)Designed a brush writing robot system based on machine vision,analyzed and selected the hardware of the system,adopted the software scheme of the modular design system,and completed the calibration of the writing robot vision system through the Matlab calibration toolbox.(2)Aiming at the problems of low contrast and noise of the text image to be copied,preprocessing algorithms such as fuzzy set image enhancement and filtering are used to improve the quality of the text image to be copied,and the image is segmented by the maximum between-class variance method.The "Zhang thinning algorithm" is used to extract the skeleton to obtain feature information such as endpoints,inflection points and intersections.A multi-scale morphological gradient edge detection algorithm is proposed to obtain the width information of the text with the largest inscribed circle.(3)Aiming at the stability problem in the robot writing process,the motion trajectory of the brush writing robot is simulated and analyzed,and a trajectory planning method combining fifth-order polynomial and B-spline interpolation is proposed,which is based on skeleton features in joint space and Cartesian space.The two-stage motion trajectory planning of the writing and the brush leaving the thesis.A simulation model of the brush writing robot is constructed in Matlab,and the stability of the robot writing is improved through simulation analysis.(4)In order to objectively evaluate the results of the copying,a similarity evaluation method of the copied text image is proposed.Through the analysis of histogram method,the perceptual hash algorithm and the gray-level co-occurrence matrix algorithm,the three aspects of the pixel,position and contour of the image before and after the copy are analyzed.Features,set the weight coefficients of different algorithms,comprehensively evaluate the effect of copy writing.(5)This thesis builds an experiment platform for a brush writing robot based on machine vision,and conducts copying experiments on three fonts: Kaiti,official script and cursive script.The experimental results show that the method in this thesis can copy individual characters one by one,with a similarity of 90% the above.