节点文献
基于FPGA的图像处理算法的研究与设计
The Study and Design of Image Processing Based on FPGA
【作者】 冯柳;
【导师】 高晓蓉;
【作者基本信息】 西南交通大学 , 光学工程, 2008, 硕士
【摘要】 “机车车辆轮对动态检测装置”是一种动态非接触测量装置,对保证机车安全运行起到重要作用。本文就“机车车辆轮对动态检测装置”为背景,研究图像处理的算法,旨在提高图像处理速度和精度。本文的实验平台是红色飓风系列的FPGA(Field Programmable Gate Array)开发板;开发软件是Altera公司的QuartusⅡ6.0和第三方仿真工具Modelsim SE 6.1f;使用的语言是硬件设计语言Verilog HDL(Hard DesignLanguage)。本文涉及的图像处理包括三部分——视频采集及数字化、图像增强和曲线提取。视频采集部分详细介绍了视频信号的格式,完成了对视频解码芯片SAA7113的配置,并且根据处理需要提取了视频信号中的亮度信号,进行存储。图像增强和特征提取部分调研了灰度级修正、平滑降噪、边缘检测、轮廓提取中的典型算法,并对各个算法的特点加以分析总结。根据处理对象的特点和FPGA自身的特点对Canny算子和轮廓提取的算法予以改进。将直方图均衡化、加权均值滤波、传统中值滤波、多级中值滤波、Sobel算子、改进的Canny算子和改进的链码法在FPGA上实现。结果表明以上各种算法在FPGA上实现的可能性和正确性。最后根据实验结果,综合考虑算法的速度、处理精度和实现的复杂度等因素选择出适合本系统的处理方法,对提高装置性能起到一定作用。
【Abstract】 "The dynamic detection equipment of vehicle wheels" is a dynamic non-contact detection equipment,which plays an important role of the train running safety.To improve the processing speed and precision of the equipment, this paper researches on image processing based on the images which come from the equipment.In this paper,the FPGA board named Red Cyclone is the platform of hardware;QuartusⅡ6.0 of Altera Company and Modelsim_SE 6.1f of the third party are the exploited software;Verilog is used as hard design language.Three parts of image processing are concerned in this paper,video collection and digitization,image enhancement and boundary extraction。In the part of video collection and digitization,the formats of the video have been introduced particularly.The video decoding chip SAA7113 has been configured.According to the requirement of the design,the brightness included in the video signal has been picked-up and stored.In the part of image enhancement and boundary extraction,the paper introduces some typical methods of image filtering,edge detection,boundary tracking.It also analyses the characteristics of those methods.Taking the characteristics of FPGA and the object to be dealt with into consideration,the paper optimizes the Canny operator and boundary tracking method in more efficient ways.Histogram equalization,traditional median filter,multi-median filter,Sobel operator,improved Canny operator and improved edge extraction are implemented on FPGA.The results show that all of methods which could be implemented on FPGA are correct.In the end,considering the speed,precision and complexity of methods,the paper picks up the adaptive methods to be use in the equipment,which plays a positive role in improving the performance of the equipment.
【Key words】 FPGA; video collection; image filter; edge detection;
- 【网络出版投稿人】 西南交通大学 【网络出版年期】2009年 01期
- 【分类号】TP391.41
- 【被引频次】18
- 【下载频次】1347