"投票"概率变换在纹理背景中实现水下目标检测的研究

针对水下激光目标成像时图像对比度低、噪声高等问题。采用并改进了“投票”概率变换，导入边缘方向信息和边缘连续性信息，提出两种抗噪声的“投票”概率变换算法。通过实验检测水下模拟水雷目标，证明这一方法比传统算法有更好的效果。

Underwater Target Detection Against Texture Background Using Hough Transform

The image of underwater laser imaging system (UWLIS) normally has poor contrast and high noise. To detect edges under this circumstance, two algorithms for improving anti noise performance of Hough transform are designed. One is by adding the additional information of the direction of edge points, another by checking the connetivity between the edge points. The experiment of the mine like object recognition with two algorithms described above is given. This result shows that the improved algorithms are evidently superior to the conventional algorithms.