An Improved Adaptive Threshold RANSAC Method for Medium Tillage Crop Rows Detection

Abstract

In view of crop rows with irregular leaves due to different growth conditions in a single frame image, A line extraction method combining vertical projection and adaptive threshold RANSAC is proposed. Firstly, the backbone intervals are obtained through vertical projection map. Then, fitting parameters of different crop rows are obtained from their respective vertical projections. Finally, the adaptive threshold RANSAC fitting is performed in different crop rows fitting intervals obtained by filtered. Experiments show that the detection rate of single frame image is 96.76%, and the accuracy is 92.18%, which are higher than other algorithms. The average detection fitting time is 243ms, which can meet the real-time detection requirements of agricultural machinery.