联合NSCT模糊域的声呐图像目标增强方法研究

Research on sonar image target enhancement method combined with NSCT fuzzy domain

提出一种基于非下采样轮廓波变换(NSCT)结合模糊域的噪声抑制和目标增强方法。使用NSCT将含噪声呐图像分解为不同的频率和方向响应子带,利用V-Bayes估计理论推导了模型的相应非线性二元阈值函数对NSCT系数的子带进行去噪。结合领域信息和空间信息构造模糊特征,再构建出模糊隶属度收缩函数对降噪后的NSCT分量执行二次降噪,最后以重建降噪后的声呐图像数据,使用综合数据和效果图示例来证明所提出的方法能够有效的去除声呐图像噪声,且优于基于小波变换的阈值去噪方法。

A non-subsampled contourlet transform(NSCT) combined with adaptive threshold is proposed for noise suppression and target enhancement. First, NSCT can be used to decompose noise-containing sonar images into different frequency and directional response subbands. Then, we use the adaptive threshold operator to de-noise the subband of the NSCT coefficient. Combined with the feature of field information and spatial information to construct the fuzzy, to build up the fuzzy membership degree after the contraction function of noise reduction of NSCT secondary noise component implementation, and finally to rebuild the sonar image data of the noise reduction after using sample data to prove that the proposed method can effectively remove noise sonar image, and is superior to the threshold denoising method based on wavelet transform.