一种抑制声呐图像散斑噪声的形态学滤波器

声呐图像散斑噪声抑制是声呐图像处理领域重要的难题。设计一种抑制声呐图像散斑噪声的形态学滤波器。该滤波器中,对于每一个尺度,利用不同方向的2种结构元素分别对声呐图像进行闭开、开闭运算,将运算结果以图像熵比值作为融合准则进行图像融合得到该尺度下的一级滤波图像;利用该尺度下另外2种不同方向的结构元素分别对一级滤波图像进行闭开、开闭运算,将运算结果以图像熵比值作为融合准则进行图像融合得到该尺度下的二级滤波图像。然后,将不同尺度下的二级滤波图像以图像熵比值作为融合准则进行图像融合得到最后的滤波图像。实验结果表明,与中值滤波、Lee滤波、Kuan滤波、Frost滤波等比较,设计的滤波器有更好的抑制声呐图像散斑噪声能力,而且不降低图像边缘保持能力。

A morphological filter for despeckling of a sonar image

It is important and difficult to despeckle a sonar image for sonar image processing. The design method of the morphological filter is described for despeckling of a sonar image. In the filter, for each scale of structure elements, the sonar image is performed respectively with close open or open close operations by means of two kinds of the structure elements with different directions. Four images are obtained and combined to achieve the first level filtered image with the image entropy ratio. The first level filtered image is performed respectively with close open or open close operations by means of the two other structure elements with two another different directions under that scale. Four another images are obtained and combined to obtain the second level filtered image with that scale by means of image entropy ratio. Finally, the second level filtered images with different scales are combined to obtain the final filtered image by means of the image entropy ratio. Experimental results show that the designed morphological filter can better despeckle without decreasing the ability of preserving image edges.