Research On Detection Algorithm Of Night Vision Applying In Apple Harvesting Robot

With the development of machine vision technology,robotics and artificial intelligence,great progress has been made in the research of fruit and vegetable harvesting robots.However,due to the complex operating environment,the picking efficiency can not meet the actual production needs.As an important part of apple picking robots,machine vision system plays an important role in improving the recognition and location performance of the target fruit.It is the key technology to improve the working efficiency of picking robots.At the same time,in order to achieve full use of time and improve the efficiency of picking,the robot picking operation time is prolonged and the night working mode is realized.Therefore,the research on detection algorithm of night vision applying in apple harvesting robot has important theoretical significance and practical value.First of all,this paper studies the noise reduction of apple image at night.Towards Gaussian noise processing problem,a Fast ICA noise reduction algorithm based on maximum likelihood deviation removal is adopted in this paper.Based on the signal processing,the algorithm regards the noisy image as the mixture of noise and source image,and redues the effect of noise by correcting the deviation.Experimental results show that this algorithm can effectively improve the image quality and reduce the influence of noise.In order to accurately extract night-time target apple,In this paper,the a* component of L*a*b*,the Cb component of YCbCr and the R component of RGB are fused in a certain ratio.And under the a*CbR fusion space,the image executes threshold segmentation.Compared with Otsu method,chromatic aberration method and K-means algorithm,the experimental results show that a*CbR-color space fusion method can accurately and efficiently extract the target apple from the complex environment at night.Then,against the problem of low accuracy and low effectiveness for the target recognition,using the round-like characteristics of apple,three-point fixed circle method based on genetic algorithm is proposed to improve the target fruit recognition and localization.Compared Hough transform detection algorithm and three-point fixed circle method,the experimental results show that this method can improve the precision of target fruit recognition and reduce the error caused by the three-point fixed circle method.In the end,based on the GUI compilation toolbox in MATLAB,a nighttime apple image recognition and location software is designed,and it includes the functions of image noise reduction,image segmentation and nighttime apple recognition with occlusion and overlap,and the graphical user interface is also established.