基于迭代技术的图像轮廓提取方案的研究与实现

The Study and Realization of Program on Image Contour Extraction Based on Iteration Technique

图像轮廓提取算法通过识别每个像素在其邻域内灰度的变化而检测出图像的边缘;但由于图像的复杂性和随机性,特别是自相关性强烈的图像,其微结构往往是十分复杂的。典型的轮廓提取算法已经无法直接用于其图像轮廓的提取,而且几乎无法从图像中提取到较为清晰的目标轮廓。对此,提出一种新的基于迭代技术的图像轮廓提取方案来改善;并通过VC++6.0软件进行验证。首先使用分形技术方法对图像进行了处理,噪声干扰因素得到了有效的降低;通过这样的办法净化了图像背景并增强了图像元素之间的差异度。然后对图像进行微分提取算法的轮廓提取。结果表明:传统图像边缘提取技术与分形迭代技术结合起来的新方案能够有效的区分图元颗粒的形态,可以从自相关性强烈的图像获得比较清晰的轮廓,从而更清晰地揭示图像组成元素的图元结构。该方法对图像处理范畴的拓展以及图元微结构的识别判定都有积极的指导意义。

The image contour extraction algorithm detect the image edge by identifying each change of pixel gray in the neighborhood, but since the complexity and randomness of images, especially the images of strong auto-cor- relation which microstructure is often complicated, the typical contour extraction algorithm is not suitable for direct- ly applied to micro-structure of the image processing, and almost can not be extracted the clearer target profile from the images. So the program of on image contour extraction based on iteration technique was proposed to improve the problem, and the verification was done by the VC＋＋ 6.0 software. Firstly, the fractal image method was used to treat the images, the noise disturb was effectively reduced, as a result, the background of image was purified as well as the obvious difference between the images by the way; and then profile extraction of image contour was done by the differential extraction algorithm. The results showed that ： the new program which combines of traditional im- age edge extraction technology and fractal iterative technique can effectively distinguish morphology of the particles in image elements, and also it can obtain the clear outline from the images of strong auto-correlation to reveal primi- tive structure of the image composed elements. The method is helpful to expense the image treatment area and judge the primitive micro-structure.