基于多分辨分析的合成孔径声纳相干斑抑制研究

相干斑噪声的存在使合成孔径声纳图像不能正确反映目标的反射特性,严重影响了图像的质量,降低了对图像的分析和理解性能。一个良好的合成孔径声纳图像相干斑抑制算法,必须在有效抑制相干斑噪声的同时,尽量保持图像中的边缘、点目标等细节信息。小波变换具有多分辨特性,可以利用其多分辨特性进行图像噪声消除。由于合成孔径声纳图像相干斑具有乘性噪声的性质,这就需要对其合成孔径声纳图像进行对数变换,把乘性噪声转换成加性噪声。然后对加性噪声进行小波多分辨分析和阈值处理,剔除小于阈值的小波系数后进行逆小波变换和指数变换,从而获得抑制相干斑噪声后的新的合成孔径声纳图像。从相干斑抑制的结果可以看出,基于多分辨分析的相干斑噪声抑制算法能够有效降低合成孔径声纳图像中的相干斑噪声。

Speckle suppression for Synthetic Aperture Sonar imaging based on multiple resolution analysis

The presence of speckle noise makes Synthetic Aperture Sonar (SAS) image incorrectly discribe the reflective properties of objects, seriously affects the image quality, and reduces analysis and understanding performance of SAS image. A good speckle noise suppression algorithm must effectively suppress speckle noise and keep the edges, point objects and other details information of SAS image at the same time. The wavelet transform has multiple resolution characteristics, so we can use the multiple resolution property to suppress speckle noise. Because of the multiplicative characteristics of speckle noise in SAS image, the original SAS image will be logarithmic transformed in order to get additive noise image. Then wavelet multiple resolution threshold processing can be used in additive noise image, and finally we can recover the data by index transformation. From the results of speckle suppression, we can see, the speckle noise suppression algorithm based on multiple resolution analysis can significantly reduce the speckle noise in SAS image.