Image Recognition Technology in Intelligent Cotton Harvesting Machine

In view of the functional demands of the electrical control system for the intelligent cotton harvesting machine, the machine vision technology is employed. A cotton recognition method is proposed based on Hough transform. When the cottons gather together or overlap, the image data are partially lost or incomplete. To precisely identify the hidden cotton in the natural environment, firstly the image is segmented by R-B channel in RGB color model to reduce the computational complexity. Then the minimum external rectangle is obtained by labeling its region. Finally, contour information is acquired through the boundary trace approach in the effective region and the randomized Hough transform (RHT) method is realized according to the mathematic model of cotton contour. Results show that the recognition method is feasible when the amount of cotton contour information is lost less than 50%, which can meet the accuracy and speed requirements of target recognition for the intelligent cotton harvesting machine.