Recognition technique of crop row centre lines based on mathematical morphology

Qing Wei,Zhihuai Mao,Fan Yang,Fei Jia
DOI: https://doi.org/10.13031/2013.37302
2011-01-01
Abstract:In field operation, the machine vision of an intelligent tractor needs to use centre lines of crop rows as driving courses to navigate the vehicle autonomously. To recognize these centre lines from crop images fast and effectively, this reported research made an experimental analysis of an early paddy field image based on mathematical morphology, which was 456 pixels width and 292 pixels height. The study indicated that the image was preprocessed from color into binary by selection of an appropriate threshold based on bimodal distribution. Further image processing showed that the boundaries of crop rows was detected obviously after the row crop and background were segmented effectively by morphologic methods. Based on pixel distribution of these longitudinal boundaries, final processing accomplished the crop row centre lines by robust regression. The results showed that the recognized lines were expressing the crop row centers well, when they were overlying in the rude image. This study revealed that a recognition technique of crop row centre lines based on mathematical morphology could be an effective way to give the location and direction of crop rows accurately.
What problem does this paper attempt to address?