Simultaneous mapping and crop row detection by fusing data from wide angle and telephoto images

Abstract

This research introduces a machine vision method capable of performing simultaneous mapping and crop row detection with increased accuracy, which makes it ideal for practical applications like spraying. An experimental vehicle was equipped with a Real-Time Kinematic Global Positioning System (RTK-GPS), a Fiber Optic Gyroscope (FOG) and a new type of camera developed by Fujifilm Corporation. The camera can shoot wide angle and telephoto images simultaneously, so it was possible to use it in combination with the RTK-GPS and image processing algorithms to build a field map containing up to ten crop rows. In addition, it was possible to use the crop row detection data obtained from the telephoto image to increase the accuracy of the data obtained from the wide angle image by means of a simple data fusion technique. The results show a reasonable amount of theoretical error for the resulting map, which contains an increased number of crop rows in comparison to traditional mapping methods. Additionally, results also evaluate the error and noise reduction for the crop row detection algorithm; which suggest that the machine vision method introduced in this research displays accuracy improvement in comparison to other methods.