Ship recognition from high resolution remote sensing imagery aided by spatial relationship

Target recognition is of great importance for information extraction from high resolution remote sensing image. As an important kind of man-made objects, ship recognition is a key point to many applications, such as vessel monitoring and marine traffic. As spatial relationship is invariant to topology change, a method of ship recognition from high resolution remote sensing imagery aided by spatial relationship is proposed and implemented. The method includes four critical steps: water segmentation, potential ship detection, seed growing and result creation. Experiments show that this method is robust to object position, orientation, scale, and intensity, and achieve a high accuracy of ship recognition.