改进权值非局部均值超声图像去噪

发布时间：2018-06-09 11:14

本文选题：图像去噪 + 非局部均值　；参考：《中国图象图形学报》2017年06期

【摘要】：目的超声图像斑点噪声会影响诊断的准确性和可靠性。通过分析超声图像斑点噪声统计模型,结合非局部均值滤波算法,提出一种基于超声斑点噪声模型的改进权值非局部均值(NLM)滤波算法。方法算法针对超声图像灰度信息对图像进行预处理,利用超声图像斑点噪声模型改进传统NLM算法的权值计算函数,基于图像特征确定最优采样间隔进行下采样,利用改进后的权值计算函数对图像进行NLM去噪处理。结果分别采用人工合成与真实超声图像对本文算法性能进行测试,并与传统非局部均值滤波算法、非局部总变分(NLTV)等算法进行去噪效果比较,同时采用均方误差、峰值信噪比和平均结构相似性作为滤波算法性能的客观评价指标。本文算法能快速完成超声图像的去噪处理,峰值信噪比较其他算法可以提高0.2 d B以上,可以降低均方误差,提高平均结构相似性,缩短处理时间,并得到较好的图像质量和视觉效果。结论根据超声图像斑点噪声模型对NLM算法的权值计算函数进行优化,使得NLM图像滤波算法能更好地适用于超声图像的去噪,基于超声斑点噪声模型的改进权值NLM算法相较于其他算法,滤波效果更佳,适合超声图像去噪。
[Abstract]:Objective the speckle noise of ultrasonic images will affect the accuracy and reliability of diagnosis. By analyzing the statistical model of speckle noise in ultrasonic images and combining the non-local mean filtering algorithm, an improved weighted non-local mean (NLM) filtering algorithm based on ultrasonic speckle noise model is proposed. Methods the image was preprocessed according to the gray level information of ultrasonic image. The traditional NLM algorithm was improved by using speckle noise model, and the optimal sampling interval was determined based on the image features. The NLM denoising process is carried out by using the improved weight calculation function. Results the performance of the proposed algorithm was tested by artificial and real ultrasound images, and compared with the traditional non-local mean filtering algorithm and the non-local total variational NLTV algorithm. Meanwhile, the mean square error was used. Peak signal-to-noise ratio (PSNR) and average structure similarity are used as objective evaluation indexes of filtering algorithm performance. Compared with other algorithms, the peak signal-to-noise ratio (PSNR) can be increased by 0.2 dB, and the mean square error can be reduced, the average structural similarity can be improved, and the processing time can be shortened. And get better image quality and visual effect. Conclusion based on the speckle noise model of ultrasonic image, the weight calculation function of NLM algorithm is optimized, which makes NLM image filtering algorithm more suitable for ultrasonic image denoising. Compared with other algorithms, the improved weighted NLM algorithm based on ultrasonic speckle noise model has better filtering effect and is suitable for ultrasonic image denoising.
【作者单位】：杭州电子科技大学生命信息与仪器工程学院;
【基金】：国家高技术研究发展计划(863计划)(2015AA020303) 国家自然科学基金项目(KYZ193713071,KYZ193716041)~~
【分类号】：TP391.41

【相似文献】