Automatic water line detection for an USV system

Real-time and accurate detection of the sailing area or water area is an enabling technique for an unmanned surface vehicle (USV) system. However, research on water line detection based on optical images is rare, and current water line detection method is difficult to be used in real application. An automatic water line detection method based on optical images is proposed and validated with a practical USV system. First, the basic principle of structure extraction is introduced to preprocess the original river images so that the textures resulted from the USV motion, wind and illumination is removed; Then, gray level co-occurrence matrix (GLCM) and local binary patterns (LBP) are combined to calculate the texture information of the river images, and a water line detection method is proposed based on the entropy indicator and Hough transformation. Finally, the water lines of many practical images captured by our USV system moving on an inland river are tested with our new proposed detection method. The results show that the proposed algorithm is an effective method for detecting the complicated water lines in real applications.