Abstract
In this study, stereo vision-based crop height measurement was proposed to develop an autonomous tractor, which can control ground clearance by crop height. The crop height was measured automatically by processing the disparity map generated from stereo images matching. The image processing consists of thresholding, edge detection, and area clustering, and the highest position is selected as the crop height. The proposed technique was tested under the various camera position, and the camera was set at five different height levels: 420, 440, 460, 480, 500 mm on the front of the tractor, which was located 2 m from the target crops. The result showed that the error was less than 5% in most cases. Especially cabbage showed 1.7% error which was the best performance among the 5 crops. Considering the overall results, it is possible to measure the crop height automatically with the stereo vision-based image processing. In order to use our approach to farm machinery in field, it is necessary to optimize the image processing mechanism and to validate the developed algorithm in the various environments.
Figures & Tables
Fig. 1. Disparity map process
