Research on Human-Computer Interaction Distance Measurement Method Based on Binocular Vision
With the rise of human-computer interaction and unmanned driving,machine vision has be-come one of the key research areas at present.Three distance measurement methods,namely infrared ran-ging,ultrasonic ranging,and binocular ranging,were compared.A human-computer interaction distance measurement method based on binocular vision was proposed using computer vision technology,and a face detection and logistic regression model were established.First,the binocular cameras were calibrated using the binocular camera calibration toolbox.Then,the collected pedestrian images were imported into the computer.For images not detected,predictions were made using the logistic regression model.Stereo matching was performed using the improved BM algorithm to obtain the disparity image.Finally,depth information for the corresponding position was obtained by clicking the mouse,thereby achieving distance measurement.Experimental results show that the proposed binocular distance measurement method is effi-cient and accurate.The ranging error is guaranteed to be within 4%in the range of 1 800mm to 3 600mm,indicating good ranging performance.
NETL
NSTL
万方数据
