摘要
肉品图像中的斑点噪声与肌内脂肪的颜色特征相似,要准确提取肌内脂肪,就必须先对斑点噪声进行滤除。在小波软阈值法滤除SAR图像中的斑点噪声算法基础上,结合肉品图像的特点进行了改进。首先,使用由耿则勋提出的算法,在小波分解时对左右边界进行处理,重构时外推值为0,对离散小波变换的边界效应进行处理,精确重建图像,消除小波变换的边界效应。选择了D8小波,3级分解后,在不同的阈值下进行了实验。结果表明:阈值的选取影响去噪效果,但在所有参数都相同时,改进算法消除边界效应的同时,在斑点指数和方法噪声两个客观评价指标均优于其他几种边界延拓方式,图像失真最小。
Speckle noises are similar to intramuscular fat pixels in meat image, and they must be removed if intramuscular fat pixels need to be correctly separated from meat muscle pixels. According to the features of meat images, this paper presents an improved algorithm based on wavelet soft -thresholding filters,which were applied to remove the speckle noises in SAR images. In order to avoid boundary errors caused by discrete wavelet transform, an algorithm presented by Geng zexun is applied firstly. The extrapolated boundary values can be obtained by a formula when the discrete data are decomposed by wavelet, as well as zero when reconstruction. With boundary processing, the image can be perfectly reconstructed when the beef image is decomposed with Daubechies wavelets.The beef images are decomposed three levels by Daubechies wavelet, whose filters length is eight, and different thresholds are selected for removing the speckle noise in beef images.Experimental results show that the denoising performance is related to the soft-thresholding values,but when all the parameters are same, the improved algorithm can avoid the boundary errors with less distorted and its overall performance is superior to other boundary extension ways, especially this two objective evaluation indicator including the speckle noise index and method noise values are better than those of others.
出处
《微计算机信息》
2011年第7期22-24,共3页
Control & Automation
基金
基金申请人:彭增起
项目名称:猪肉品质智能化无损检测技术研究
基金颁发部门:国家科技部863项目(2008AA10Z211)
基金申请人:贾渊
项目名称:猪肉关键品质自动分级技术的计算机软件系统研究
基金颁发部门:西南科技大学(08zx7101)
关键词
肉品图像
边界效应
小波软阈值法
斑点噪声
meat image
boundary errors
wavelet soft-thresholding filter
speckle noise
