论文部分内容阅读
合成孔径雷达SAR图像的相干成像特性,不可避免的形成特有的相干斑点噪声,严重影响图像的地物信息提取和分类,需要进行去噪预处理。针对SAR图像斑点噪声的特点,针对SAR图像斑点噪声的特点,对图像进行小波变换分解,提出模糊聚类和软阈值收缩去噪的方法,利用模糊C均值聚类将小波系数分成包含信号能量和只包含斑点噪声能量两大类,对前一类小波系数进行软阈值降噪处理,而对后一类小波系数直接置零。实验结果的目视效果和评价指标均表明,小波模糊聚类和软阈值收缩有效地去除了SAR图像斑点噪声,图像视觉效果清晰,较好地保持地物目标的边缘等图像细节信息,去噪效果优于小波软阈值收缩。
SAR imaging SAR imaging coherent imaging characteristics, the formation of unique coherent speckle noise is inevitable, seriously affecting the image feature extraction and classification, the need for preprocessing. Aiming at the characteristic of speckle noise in SAR images, aiming at the characteristic of speckle noise in SAR images, the image is decomposed by wavelet transform. The methods of fuzzy clustering and soft threshold shrinkage denoising are proposed. The wavelet coefficients are divided into two categories: Only contains the speckle noise energy two categories, the former type of wavelet coefficient soft threshold denoising, and the latter type of wavelet coefficient directly set zero. The visual effects and evaluation indexes of the experimental results show that the wavelet fuzzy clustering and the soft threshold shrink effectively remove the speckle noise of the SAR image, the visual effect of the image is clear, and the image detail information such as the edge of the object object is well maintained. The effect is better than wavelet soft threshold contraction.