Research On Mapping Algorithm Of 3D Vision Sensor Based On Laser Speckle

The 3D vision sensors are the "eyes" of artificial intelligence.They can acquire the threedimensional information of the surrounding environment like human eyes.By comparing the existing 3D vision sensor products at home and abroad,it is found that domestic products start relatively late,the quality and accuracy are relatively poor.To this end,a set of 3D vision sensor ranging algorithm based on laser speckle is developed,a ranging system is built,and then this sensor is applied in the field of mobile robot map construction,and a three-dimensional grid map is established.The specific work is as follows:(1)This paper researched the ranging theory of 3D vision sensor based on laser speckle.The speckle autocorrelation analysis of the speckle image proves that speckle has unique characteristics;the speckle 3D vision sensor ranging model is established,the depth ranging formula is derived,and the relationship between the offset and the measured depth is analyzed.(2)This paper developed a 3D vision sensor ranging algorithm based on laser speckle.The problem of poor robustness of the speckle ranging algorithm can be solved using the locally adaptive binarization;a backtracking algorithm is used to extract the center pixel coordinates of the speckle to simplify the speckle information;in order to improve the processing speed of the ranging algorithm,the matching search range is reversed according to the relationship between the offset and the measured depth,and the matching search range is narrowed,thereby reducing the calculation amount of the algorithm;for the errors that may occur in the matching results,the matching results are screened by establishing screening criteria,which enhances the confidence of the matching results and finally the development of the ranging algorithm is completed.(3)This paper built a 3D vision sensor ranging system based on laser speckle and carry out related experiments.The ranging system is built by using a infrared camera and a speckle emitter,in order to verify the advantages of the ranging algorithm in accuracy and stability,this paper uses the ranging system to carry out ranging algorithm experiments,including local adaptive binarization experiment,the experiment of speckle descriptor window size setting,and the experiment of the ranging algorithm accuracy in this paper under different measuring distances.(4)This paper built the three-dimensional grid map.Aiming at the problem that the ideal ISM measurement model is not accurate enough,the real ISM model is obtained by convolution of the ideal ISM measurement model and the Gaussian noise of the measurement sensor to improve the accuracy of the ISM model;In order to obtain the ISM measurement model of the 3D vision sensor,the variance of the 3D vision sensor at different measurement distances is obtained,and the relationship between the variance and the measurement distance is obtained using quadratic curve fitting,and then the ISM model of the 3D vision sensor is determined;an experimental platform for the construction of three-dimensional grid maps is built,conduct experiments on the algorithm of three-dimensional grid map construction,and prove that the map construction algorithm in this paper has the advantages of high precision.