The method of coordinate recognition for maize straws under canopy by monocular vision

In large areas of corn, the lane line planning and navigation within the crop rows play a crucial supporting role in intelligent farm machinery operations such as spraying, weeding and testing. Vehicle navigation is mostly by GPS and radar. These two kinds of navigation can't complete the information needed for crop inter-row navigation line segments. Such precision navigation and obstacle avoidance ability is poor. In this paper, morphological filtering method for reducing image weeds and leaves interference straw recognition is proposed. A ridge line extraction algorithm to extract the center line of the straw, finally extracts straw coordinates, and realizes the purposes of recognition straw by monocular visual ultimately. Tested in fields which have not been weeded, the result of those not detected straws accounted for 10.9%. The result of those not detected straws in field after weeded accounted for 8.5%. Results obtained by untreated field and treated field shows that filtration system to weed effect is very obvious.