Easy El – sowing line detection

Specifications

The system is based on a single-board computer using an ARM processor. Based on the image from the camera, characteristic points are detected, which are individual plants. Then, the lines in which crops have been sown are detected, so as to distinguish them from weeds. Information about the positions of the detected lines is successively transferred to the machine controlling the position. The first stage of the project was the selection of an algorithm for detecting lines. We tested two promising algorithms: CRD and Hough. The test results are presented in the attached graph. We then tested the performance of RPI in comparison to a mid-range laptop. The difference is significant, but the frequency of 2.5 Hz was sufficient to provide shifts in the required time. The lines look irregular and quite random because at this stage the main goal was to determine the minimum requirements of the system. The photos show the resulting operation of the algorithm for detecting the position of the beginning of the crop line, and show the calculated difference by which the harvesting device must move to be exactly above the plants intended for harvesting. Thanks to this, the tractor operator does not have to manually calculate this value, but can do it based on the image from the camera.