数字图像的边缘检测研究.docVIP

本科毕业论文 数字图像边缘检测研究 Research on edge detection of digital image 数字图像的边缘检测研究 摘要 边缘检测是数字图像处理技术中比较基础的一部分，其技术的本质上就是通过对图像特征中不连续的部分进行读取判断。在这门技术中有很多决定性的因素影响到数字图像处理的结果。边缘检测是属于图像分割技术的分支，也是数据图像处理中重要的一部分。边缘检测结果的好坏影响着接下来的图像分析理解过程，所以数字图像边缘检测在数字图像工程中有着十分重大的影响力。 本文结合实例对数字图像的边缘检测技术整体进行了介绍，重点的介绍了Roberts算子、Sobel算子、Laplacian算子和LOG算子几种梯度算子。详细的介绍了目前应用最普遍的Canny算法流程与原理。同时给出了各算子处理图像的效果图，直观的展示了各算子的检测边缘的效果。 经过实验对比得出结论，Roberts算子易于实现但检测效果不理想。Sobel算子对边缘的定位比较准确和完整，但是精确度下降。Laplacian算子边缘定位准确但受噪音影响较大。LOG算子受噪音影响较小但需要平衡滤波和精确定位边缘的问题。Canny算法能减少边缘的的中断，细化边缘，滤除噪声，检测效果强于梯度算子。 关键字：数字图像处理，边缘检测，高斯滤波，Canny算法 Research on edge detection of digital image Abstract Edge detection is the basic part of digital image processing technology.Essentially, edge detection technique is characterized by the image of the discontinuous section reads and judgment. Therefore, edge detection is part of image segmentation branch.Edge detection result is impact on the image understanding process,and the digital image edge detection has a very important influence in the digital image project. This paper introduces the edge detection of digital image with the specific example. We stress on introducing some gradient operator such as Roberts operator, Sobel operator, Laplacian operator and log operator.Besides, we will discuss the canny algorithm in detail about its process and principle. We will also show the effect picture of each operator in order to display the result of each operator visually. After the comparison of the experiment, the conclusion is drawn that the Roberts operator is easy to implement but the detection effect is not ideal.. The Sobel operator is more accurate and complete to the edge location, but the accuracy is decreased.The edge location of Laplacian operator is accurate, but it is affected by the noise. The LOG operator is less affected by noise but requires a balanced filtering and precise positioning of the edge problem.Canny algorithm can reduce the edge of the interruption, thinning edge, filtering noi