In order to solve the problems such as illumination interference

poor real-time and large error for recognition of path on orchard mobile robot

An orchard mobile robot path recognition system is proposed

Firstly

the interference of different light is reduced by constructing light illumination

then the fusion of Otsu and Ncut algorithm to realize segmentation is proposed

and the orchard road boundary is detected by Sobel operator

Finally

least square method is used for detecting the navigation center line. To realize the real-time acquisition and system visualization of navigation information

the path recognition system is designed by Qt and Open CV under Linux system. The results have demonstrated that the average navigation center line offset distance is 4. 67 pixel and feflection angle is 2. 88°

the time consumption for single image is about 102. 1 ms

compared with the path recognition system using VB and MATLAB

the designed system has stronger robustness and real-time capability.