A litchi fruit recognition method in a natural environment using RGB-D images. (April 2021)

• Record Type: Journal Article
• Authors: Yu, Lianyi; Xiong, Juntao; Fang, Xueqing; Yang, Zhengang; Chen, Yunqi; Lin, Xiaoyun; Chen, Shufang
• Is Part Of: Biosystems engineering. Volume 204(2021)
• Page Start: 50
• Page End: 63
• Publication Date: 2021-04
• Subjects: litchi -- RGB-D image -- depth segmentation -- visual recognition; Bioengineering -- Periodicals; Agricultural engineering -- Periodicals; Biological systems -- Periodicals; Génie rural -- Périodiques; Systèmes biologiques -- Périodiques; 631
• DOI: 10.1016/j.biosystemseng.2021.01.015
• Languages: English
• ISSNs: 1537-5110

Abstract

Abstract : In order to achieve accurate detection of litchi fruits in their natural environment, a ripe litchi recognition method with a Red-Green-Blue Depth (RGB-D) camera, which can be used to estimate fruit yields, is proposed in this paper. First, both colour and depth images of litchis are collected using an image capturing system with an RGB-D camera. Then, redundant image information outside the effective picking range of the manipulator is excluded using depth image segmentation. Finally, the random forest binary classification model is trained employing colour and texture features to recognise litchi fruits. Simultaneously, an approach based on multi-scale detection and a non-maximum suppression Algorithm is proposed to further improve the fruit detection precision. Visual recognition experiments using the proposed method to recognize red and green litchis are designed, and the method achieves a recognition accuracy of 89.92% for green litchis and 94.50% for red litchis. Highlights: A method to recognize different varieties of litchis effectively was proposed. Visual system was designed to avoid fruit misjudgment due to uneven illumination. Depth segmentation proposed effectively improves recognition accuracy.