目的：超声图像斑点噪声会影响诊断的准确性和可靠性。本文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(non-local means, 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分块采样,利用改进后的权值计算函数对分块图像进行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,并与现成的算法进行去噪效果比较,同时采用均方误差和峰值信噪比作为滤波算法性能的客观评价指标。本文算法可以快速完成超声图像的去噪处理,提高峰值信噪比,降低均方误差,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。
Objective: Medical ultrasound imaging is widely used in clinical diagnosis, especially in pregnant women and fetus, because of its advantages such as non-invasive, inexpensive, convenient, real-time and so on. However, due to the influence of the ultrasonic imaging principle, the ultrasonic image will inevitably be disturbed by the speckle noise during the generation process, which not only reduces the quality of ultrasonic image, but also makes the identification and analysis of image detail more difficult. In this paper, an improved non-local means (NLM) image denoising algorithm based on noise model of ultrasonic image is proposed. Method: The statistical model of the speckle noise is obtained based on the probability distribution of the ultrasonic image, then using the Bayesian formula and the speckle noise model to improve the weight function of the non-local means filter algorithm. The weight function of the traditional NLM algorithm is based on Gaussian distribution, so it can suppress the Gaussian noise very well, but it is not suitable for speckle noise. In this paper, the weight function is improved based on the speckle noise model, so the algorithm can be better applied to ultrasonic image. The algorithm preprocesses the image according to the characteristics of the proposed weight function, which makes the algorithm obtain better denoising effect. Then optimize the sampling interval, so that the algorithm could maintain the effect of reducing noise while reducing the processing time. Finally the improved non-local means algorithm is applied to ultrasonic image denoising. Result: The experiments on phantom images and real 2D ultrasound datasets show that the proposed algorithm outperforms other related well accepted methods, both in terms of objective and subjective evaluations such as mean square error (MSE), peak signal-to-noise ratio (PSNR), as well as the computational time. The filtered images of proposed algorithm have a higher PSNR value with other despeckling algorithms, which means the proposed algorithm can preserve the details of the image information better and the filtered image has the similar edges with the noise-free image. In the comparison of MSE values, the proposed algorithm has the lower values, which means the algorithm can preserve the structure information of the original image better. Consider the computational time, the proposed algorithm does not take superiority in the aspect of time consuming. The experiment of real 2D ultrasound images is also conducted, and we can find the proposed algorithm get a better visual effect. Conclusion: Since the speckle noise reduces the quality of ultrasonic image and limits the development of automatic diagnostic technology. According to the speckle noise model of ultrasonic image, the weight function of NLM algorithm is optimized, which makes NLM algorithm more suitable for ultrasonic image denoising. Experimental results show that the proposed algorithm is better than other algorithms and is suitable for ultrasonic image denoising.