基于blackfin的图像边缘检测算法-电路与系统专业论文.docxVIP

中文摘要边缘检测主要用于图像处理、机器视觉和模式识别，其检测结果的好坏对于 中文摘要 边缘检测主要用于图像处理、机器视觉和模式识别，其检测结果的好坏对于 进行图像高层次的特征描述、识别和理解有着重大影响。小波分析是近年来兴起 的数学分析理论，由于其时频局部化特性，在图像信号的分析与处理领域获得了 广泛的应用。特别是其在大尺度下利于抑制噪声、而小尺度下利于精确定位边缘 的性质，非常适用于图像的边缘检测。 本文首先分析yd,波理论及边缘检测算法发展概况，通过对现有边缘检测算 法的分析和讨论，提出了一种基于边缘融合的小波边缘检测算法。该算法利用了 平稳小波的相位不变性，将各尺度下的边缘检测结果进行融合，获得了更好的边 缘细节信息，并进一步结合PCNN开发了预处理算法。在此基础上，针对小波方 向性的问题，提出了利用非抽样Contourlet变换分解后，对各子带检测结果进行 融合的边缘检测算法。 其次，通过分析比较各种DSP性能，并且结合本设计具体情况，选择ADI公 司的Blackfm 531DSP作为图像处理系统的核心处理器。并以此为平台实现了本文 提出的基于边缘融合的小波边缘检测算法的硬件实现。仿真结果表明，该算法在 硬件平台上能很好地运行，且获得了较好的边缘检测结果。 最后，对本文工作进行了总结，给出了下一步的研究目标。 关键词：图像处理；小波变换；边缘检测；Blackfm；图像融合：脉冲耦合神经 网络 ABSTRACTEdge ABSTRACT Edge detection is used widely in image processing，machine vision and pattern recognition．It’S significant for feature extraction，model identification and image understanding．Wavelet is a powerful mathematical analysis tool for time-fif．equency, which are emerged in recent years．Image processing methods with wavelet have achieved good resuRs in image signal analysis．Due to the feature of muti-analysis， wavelet is also suitable for edge detection． After reviewing the existing edge detection algorithms and analysing the advantages respectively,this paper proposed a noval algorithm of edge detection． Expoilting the phase invariance of stationary wavelet，detect each sub-band’S edge with Sobel algorithm and integrate the information through image infusion technology． The experiments show that the algorithm can preserve more detail information；and exploring pulse coupled networks we propose a pre-processing alogorithm．And we test our scheme by non-sample contourlet again． By comparing performance with other DSPs and according to the application in this paper，Blackfm 53 1 DSP is choosed as the central processor of the image processing system．And the proposed wavelet edge detection algorithm is complemented 0n this platform．．n坞resuRs of simulation proved that the proposed algorithm worked well on the hardware platform,and better edge detection effects