VISION NAVIGATION BASED ON AGRICULTURAL NON-STRUCTURAL CHARACTERISTIC

摘要

This paper studies the algorithm of extract navigation route by machine vision based on agricultural non-structural characteristic. Firstly, a new algorithm was used to detect the ending region of farmland based on cluster's method. Secondly, moving window was used to examine white pixels of the binary image to definite target regions. Thirdly, Improved Hough transformation was used to extract navigation routes. The experiment proved that this algorithm can accurately extract the column information of green crops. The average time required for processing a 320*240 pixels color image is 57.404ms and the accurate recognition rate reaches 99%.