（毕业设计论文）《基于细胞神经网络的图像边缘提取算法研究》.doc

PAGE 天津职业技术师范大学本科生毕业设计 基于细胞神经网络的图像边缘提取算法研究 Research algorithm on edge extraction of images using cellular neural network method 专业班级： 学生姓名： 指导教师： 系 别： 2012年6月 摘 要 图像的边缘提取是图像分割、目标区域识别、区域形状提取等图像分析领域十分重要的基础，在工程应用中占有十分重要的地位。细胞神经网络(CNN）是一种并行处理器，细胞神经网络技术在图像模型识别上的应用使系统具有更高的识别率。 首先,详细说明了用 CNN 提取图像边缘的有关理论和分析，给出了所设计的二值图像算法的流程图，将其用于检测二值图像边缘。再在此基础上，改进了前人提出的分 8 层的算法，实现对灰度图像的边缘提取。 然后,将此方法与传统边缘提取方法roberts、sobel、prewitt、log和canny等相比较可知，该方法的有效性。并且由于细胞神经网络能够高速并行计算，处理速度与图像大小无关，同时便于硬件的实现，这使得它在图像实时处理方面还有很大的发展潜力可以发掘。 关键词：图像处理；边缘提取；CNN；算法 ABSTRACT Image edge extraction is image segmentation, the target area recognition, regional shape extraction from image analysis area very important basis in engineering application in an important position. Cellular neural network (CNN) is a kind of parallel processor and cellular neural network technology in the application of image model identification system are higher recognition rate. Firstly, detailed description of the image edge extraction with CNN the theory and analysis, then we give the binary image design flow chart of the algorithm for testing of the binary image edge. And on the basis of this, the improvement on previous proposed points of eight layer arithmetic, realize to the gray image edge extraction. Then, the method and the traditional edge extraction methods, for instance, Roberts, Sobel, Prewitt and log compared, etc, it is known that the effectiveness of the method. And because the cellular neural network can high-speed parallel computing, and the processing speed have nothing to do with the image size, and facilitate at the same time the realization of hardware, this makes it in image processing of real time have great potential for growth. Key Words：Image processing; Edge detection; CNN; Arithmetic 目 录 MACROBUTTON InsertCrossReference TOC \o 1-3 \h \z \u HYPERLINK \l _Toc326741884 1 引言 PAGEREF _Toc326741884 \h 1 HYPERLINK \l _Toc326741885 2 图像边缘提取的基本知识 PAGEREF _Toc326741885 \h 3 HYPERLINK \l _Toc326741886 3 基