自适应梯度幅值和形态学组合滤波算法

针对传统梯度幅值滤波算法中阈值需人工确定而无法正确判断噪声点和产生新噪声点的问题，提出了一种自适应梯度幅值和多结构元素形态学组合滤波算法。首先使用梯度幅度均值与方差的和作为阈值，以及Otsu算法自适应阈值两种自适应滤波算法对图像进行滤波，以自动确定滤波阈值，最大限度滤除噪声点，保持图像细节；然后使用多结构元素组成串联形态学复合滤波器对自适应梯度幅值滤波后的图像进一步滤波，以解决在自适应滤波过程中产生的新噪声问题。通过实验结果对比与分析，该算法滤波后的图像视觉效果明显好于传统滤波方法，从而验证了该算法既能保持图像细节，又能实现较好的图像滤波效果。

Hybrid filtering algorithm based on adaptive gradient magnitude and morphological operations

In view of the defects of the traditional filtering algorithm based on gradient magnitude, in which the artificial threshold can not determine the noise points correctly and bring out the new noise points, a hybrid filter algorithm based on adaptive threshold gradient magnitude and morphological operations was proposed. Firstly, the original image was filtered by using the mean and variance of the gradient magnitude, and the Otsu algorithm was improved to adaptive threshold value, which could determine the threshold automatically and filter noise, and maintain the image detail. Then the series filter composited by multi-structure elements morphology operations was used to filter the image generated by the adaptive gradient magnitude filter algorithm, which could eliminate the new noise generated in the adaptive filtering process. Finally, the experimental results and analysis show that the visual effect of the proposed algorithm is better than the traditional filtering method, the advantages of better maintaining image detail and filter effect of the hybrid algorithm are demonstrated.