Measurement of harvesting width of intelligent combine harvester by improved probabilistic Hough transform algorithm

Aiming at the problem that the harvesting width measurement of rice or wheat is susceptible to interference and poor real-time performance, an improved probabilistic Hough transform (PHT) width measurement algorithm based on morphological vertical line extraction is proposed. Inverse perspective mapping (IPM) restores the parallel relationship of the vertical line of the three-dimensional world. Image enhancement of the object region is used to highlight differences between the harvested and unharvested region. The median filter of large size filter window eliminates the interference of local noise. Improved PHT based on the vertical line extraction of the threshold segmentation realizes the real-time accurate measurement of the harvesting width during the operation of the intelligent combine harvester. Experiments demonstrate that the improved PHT can accurately obtain the harvesting width error of not more than 64 mm at an average success rate of 96.025%, and can measure the harvesting width in real time at 77.675 ms/frame. (C) 2019 Elsevier Ltd. All rights reserved.