中文摘要：

为了更好地检测停在港口岸上的船只，提出了一种角点及角度检测方法，以用于对前端有尖角的船只检测。该法以图像在某一点的极坐标变换为基础，将梯度图像的灰度值在每个方向上叠加，再检测累加值最大的两个方向，得出两个方向上分别的累加值和方向角，再通过一个权值计算公式来设定该点为角点的可能性。结合高斯平滑，该方法能方便地找出不同尺度的角点，及其对应角度的位置，从而有效检测出船的前端点及船的方向，进而确定船的位置。该法克服了其他方法无法分割出停在岸边的船的缺点，对船只的检测提供了值得借鉴的经验。

英文摘要：

To have better detection of vessel on shore in the port, an angle detection method was presented to detect vessels with acute angle in their front part. On the basis of the polar coordinate transformtion of a point in image, this method a ceumulates the gray values of gradient images in each direction, detects the two directions of largest accumulated value,obtains the accumulated values and direction angles in both directions,and finally,sets the possibility of the point as angle point using a weight calculating formula. This method,combined with Gauss smoothing,can easily find out the angle point of different scales and the position corresponding to the point,and can effectively detect front endpoint and direction of vessel,and can conveniently to determine the position of vessel. This method overcomes the drawbacks of other methods that can not distinguish the vessels on shore,and provides a good support for vessel detection.