关于人体图像精确处理算法改进

About the Body Image Precision Processing Algorithm is Improved

对人体图像进行精确处理,可以提高成像质量。由于图像内包含大量的乘性噪声,必须针对每个像素点设定不同的阈值标准,通过自适应设定的阈值标准将噪声与目标划分开来,才能完善目标图像的细节信息,精确处理图像。传统的小波变化降噪算法采用人为设定一个阈值,不能设定合理的阈值标准,导致有的区域降噪效果差,有的区域降噪过度,损失图像的边缘细节信息,不能精确地处理图像。提出采用模糊聚类的小波降噪算法不均衡噪点模糊成像改进方法。先利用模糊算法将原图划分背景与目标两部分,通过计算类间方差最大值获取图像的最优阈值,并估计出各个图像区域的噪声模糊均差。采用模糊决策的方法将图像中所有小波系数进行分类,利用阈值对只含有噪声的小波系数进行收缩,将只含有噪声的小波系数归零,并将处理后的小波系数进行M带小波变换,有效的完成对传统小波降噪算法不均衡噪点模糊成像改进。仿真结果证明,改进的方法为图像降噪提高了质量。

In this paper, we proposed a modification method for fuzzy imaging of imbalance noise point using wavelet de-noising algorithm based on fuzzy clustering. Firstly, we used the fuzzy algorithm to divide the original drawing into background and objective and obtained the optimal threshold value of image through calculating the maximum of interclass variance, we also estimated the noise fuzzy mean deviation of each image region. Then we used fuzzy decision method to classify the wavelet coefficient in image and used threshold value to shrink the wavelet coefficient only having noise. The wavelet coefficient was made zero. Finally, we made the M-band wavelet transformation to the processed wavelet coefficient and completed modification on traditional fuzzy imaging of imbalance noise point using wavelet de-noising algorithm. The simulation results show that the modification method can improve the image de-noising quality.