复杂背景下扩展目标多尺度小波分割策略

用大尺度小波检测主要轮廓边缘及抑制背景细纹和噪声，用小尺度小波进行边缘的精确定位检测，并分别计算自适应的双阈值；对用较大阈值限幅输出的2幅边缘图像进行相与，并去掉短的离散短边缘得到组合边缘输出图；结合在较小阈值限幅输出的大尺度小波边缘检测出图像及其相位图，对有相近幅度和相位的边缘进行连接，得到最后的边缘分割结果。实验结果表明，该分割策略是有效的。

Mutiscale Wavelet Based Segmentation of Extended Target in Complex Environment

In this method,big-scale wavelet was used to detect main profile of the target and to restrain noise and sharp edges of background,and edge detection with precise location of the image was performed with small-scale wavelet,meanshile,two adaptive thresholds were calculated respectivly,two edge maps at each bigger threshold were performed logical-AND operation,short discrete edges were taken out from the result edge map,the remained edges with similar amplitude and phase were connected according to the edge map at smaller threshold and the phase map of big-scale wavelet edge detection,at last,closed profile of the target was gained.Experiments turned out the algorithm works efficiently.