1997 Volume 59 Issue 3 Pages 37-45
3-D image recognition experiment of crisp head lettuces was done to develop a selective harvesting robot of crisp head vegetables. Lettuces in a single row that were measured by the previously reported 3- D vision sensor were recognized. The 3-D image was smoothed in the first process. Then the 3-D image was binarized to recognize the location and the size of the crisp head. 60 pieces of 62 crisp heads above a proposed diameter could be recognized to be good for harvest. And all of 60 lettuces below the diameter could be judged to be immature. Two large leaves were recognized to be crisp heads. The average position error was 12.4mm, and the average diameter error was 13.4mm.