获取医学灰度图像轮廓图的算法

A new boundary algorithm for acquiring medical gray-level image

目的:提出一种改进的形态学边缘检测算子,以获取医学灰度图像轮廓图,并保持边缘的平滑性,并与传统的边缘提取方法进行比较。方法:图像边缘检测通常是以类似于素描图的图像表达出物体的要素和特征,其任务是使图像边缘准确定位和抑制噪声。试验采用3×3的模板作为结构元素对原图像进行处理,利用基于数学形态学的方法,用形态运算膨胀、腐蚀、开、闭等变换以及它们的组合及灰度切片的方法获取质量好的医学轮廓图像,并与用Sobel算子方法和Roberts算子方法获得的轮廓图像进行比较。结果:试验结果表明,Sobel和Roberts等算子不能全面检测出边缘,且边缘模糊。采用复合型数学形态学算子与灰度切片结合的算法获得的轮廓图边缘连续和完整,断点少,且轮廓周围的灰度已进行合并具有更丰富的细节,相对于常用的微分算子和形态学边缘梯度算子更能有效地滤除噪声,这一方法对噪声医学图像边缘的提取效果更好。结论:采用复合型数学形态学算子与灰度切片结合的算法从原始图像获取医学轮廓图像效果好,优于传统的边缘提取方法。

AIM： To propose an improved edge detection using morphological operations to implement boundary acquisition of medical gray-level image, keep the detected edge smooth, and compare with the traditional method for edge extracting. METHODS： The image edge detection is usually subjected to express the features and outlines of objects using the images identical to sketch pictures. Its task is to locate image edges exactly and reduce its noise. The primary image was managed taking a 3×3 template as the pattern element. On the basis of mathematical morphology, image boundary image was acquired with dilation, erosion, closing, and opening, etc, and their combination, together with gray-level slice. Then the implemented boundary image was compared with the image acquired by Sobel and Roberts methods. RESULTS： The experimental results indicated that, Sobel and Roberts were unable to detect the boundary completely. The mathematical morphology algorithm and gray-level slice was better for implementing boundary image acquisition of medical gray-level image. This method got continuous boundary with few breaking points. Grey-level became combined and more detailed. Noise filtering was also superior to other different algorithms and gradient morphological algorithms. CONCLUSION： The medical boundary images can be efficiently processed by the mathematical morphology algorithm and gray-level slice, which is superior to the previous methods of edge extraction.