Localisation of litchi in an unstructured environment using binocular stereo vision. (May 2016)

• Record Type: Journal Article
• Authors: Wang, Chenglin; Zou, Xiangjun; Tang, Yunchao; Luo, Lufeng; Feng, Wenxian
• Is Part Of: Biosystems engineering. Volume 145(2016:May)
• Page Start: 39
• Page End: 51
• Publication Date: 2016-05
• Subjects: Binocular stereo vision -- Wavelet transform -- Litchi recognition -- Stereo matching; Bioengineering -- Periodicals; Agricultural engineering -- Periodicals; Biological systems -- Periodicals; Génie rural -- Périodiques; Systèmes biologiques -- Périodiques; 631
• DOI: 10.1016/j.biosystemseng.2016.02.004
• Languages: English
• ISSNs: 1537-5110

Abstract

Abstract : The major constraints for a litchi harvesting robot were to recognise and locate litchi in an unstructured environment with varying illumination and random occlusion. A rapid and reliable method based on binocular stereo vision was developed with the aim of effectively recognising and locating litchi in the natural environment. The method involved the application of wavelet transform to a pair acquired images of litchi to normalise illumination of an object surface. A litchi recognition algorithm based on K -means clustering was presented to separate litchi from leaves, branches and background. A matching algorithm to locate litchi based on a label template was discussed. Litchis with a similar label template were matched according to the preset threshold by traversing a litchi label template of a left image in a right image to find optimal matching. The experimental results showed that the proposed recognition method could be robust against the influences of varying illumination and precisely recognising litchi, the highest average recognition rate for unoccluded and partially occluded litchi was 98.8% and 97.5% respectively. From 100 pairs of tested images of unoccluded and partially occluded litchis 98% and 94% were successfully matched, respectively. Errors had no significant difference and they were less than 15 mm when the measuring distance was between 300 mm and 1600 mm under varying illumination and partially occluded conditions. Highlights: An algorithm for recognising litchi under varying illumination was developed. A label template-based matching algorithm was proposed for unstructured environments. System can recognise and match non-occluded and occluded litchis. Errors were less than 15 mm when the measuring distance was 300–1600 mm.