基于对象识别和分割的图像噪声抑制

Image noise reduction method based on object recognizing and segmentation

在医学图像降噪中,不损失图像细节是至关重要的。传统的方法是直接对整个图像进行滤波处理。然而这种操作在减少噪声的同时也会破坏图像的细节。因此,关键的问题就是怎样在保持图像细节的同时能够减少噪声。为了解决上面提到的问题,在这篇论文中提出了一种模拟生物视觉的图像处理方法,首先提取图像的边缘信息,将图像分割成边缘区域和非边缘区域,然后对这两个区域采取不同的滤波降噪,再进行重新合成。与传统方法相比,本方法在保存图像细节上有一定的效果。

In the processing of noise reduction of medical image, the most important thing is to protect the detail information of the image. Usually, people perform a filtering computation on global of the image space, which although reduced the noise somewhat, but damaged the detail information at its edge. Therefore, the key is how to keep the detail information of the image in the processing of the noise reduction. In this paper, we introduced a method to solve the problem. This method simulates the processing of human vision. First, we extract the edge information from the image, and segment the image into the edge area and the non-edge area. Then, the different filtering computation on these two parts was carried out. At last, one can reconstruct the image from these two filtered parts of the image. Compared with the traditional method, this method makes the progress in keeping the detail information of the image.