The uncut lawn cognition algorithm based on image analysis

An algorithm was proposed for recognizing uncut lawn in order to improve the efficiency of robotic mowers. Image data were captured on uncut lawn. After operations on image data of edge detection, image binaryzation, and image erosion, freeman chain code was used for larger target contour extraction and contour filling, and then the filling area was thinned and the thinned skeleton was pruned. After these operations, the shape features like the length of grass and the ratio of grass length to width were used to recognize the uncut lawn. Collecting 50 images of the uncut lawn respectively on sunny days and overcast days. The experimental results show that this algorithm has good cognition properties, the cognition ratios of 50 images were 84% and 90%.