实时图像增强研究及其系统设计

	实时图像增强研究及其系统设计
其他题名	The Research of Real-time Image Enhancement and ItsSystem Design
	张辽
学位类型	硕士
导师	赵金宇
	2015
学位授予单位	中国科学院大学
学位专业	机械工程
关键词	图像增强实时性抗噪性动态增强目标识别
摘要	图像处理技术发展至今，对信息时效性的要求也越来越高，图像处理正向着实时化的方向迅猛发展。制约着图像处理的实时性能的两个要素是处理器的性能和算法性能。近年来，DSP 技术的发展使得处理器的实时信息吞吐量得到大大提高，处理能力随之加强；图像增强算法的结构优化等方面也取得了较多突破，处理延时越来越短，实时处理技术正蓬勃发展。首先，本文设计了实时处理系统的硬件系统和软件系统。硬件系统以 TMS320DM642 型号的DSP 芯片为核心，包括了Cache、视频端口、EDMA 和 EMIF 控制器。功能模块上包括了采集模块、存储模块、电源模块和串口通讯。软件系统包括了图像采集、目标识别、目标大小计算、目标跟踪、目标增强等模块，从基础原理和代码两部分进行了阐述。其次，为了解决在工程应用中遇到的两个问题，提出了两种对应的改进算法：基于局部统计特性的对比度增强算法和改进的动态单目标识别算法，分别解决传统局部增强算法抗噪性能低和搜索模式下动态目标增强问题。此外，还提出了一种改进的高速二值连通域检测方法。从数学原理上对改进算法进行了阐述，列举了几种常用的传统算法，并用软件进行仿真以分析算法可行性。最后，做了3 组实验作为验证：基于局部统计特性的增强对比试验；改进的高速连通域检测验证实验；搜索模式下的动态的单一目标的实时增强实验。搭建了实验平台，设计了软件算法，得到了实验数据，并对实验结果进行了分析，对比之前的可行性分析，3 种改进算法均具有较好的增强效果和实时性，在工程应用中有重要意义。
其他摘要	Modern image processing technology is developing rapidly, so that the timeliness of transmission of information have become increasingly demanding, and real-time image processing is becoming a major development direction of image processing. Performance of CPU and the algorithms restrict the development of real time image processing. Recent years, the development of DSP technology makes the throughput of real-time information processor been greatly improved and its processing capability followed strengthened. There are also many breakthroughs of image enhancement algorithms in structural optimization, so that processing delay becomes shorter and shorter, which shows that real-time processing technology is booming. This paper describes the hardware design and software design of real-time image processing system firstly. Hardware system use TMS320DM642 DSP chip as the core, it also included the cache, video port, EDMA and EMIF. Function modules include the acquisition module, memory modules, power modules and serial communication modules. Software system includes image acquisition, target identification, target size calculation, target tracking, target enhancement and other modules, the basic principles and code of these parts are described. Secondly, in order to solve the two problems encountered in engineering applications, two corresponding improved algorithms are proposed: improved contrast enhancement algorithm based on local statistical characteristics and the dynamic single target recognition algorithm, which solve the problems of poor anti-noise performance of traditional enhancement algorithm and dynamic enhancement of single tagert. Furthermore, an improved binary connected domain detection algorithm is proposed. It lists several commonly used conventional algorithms, and use simulation software to analyze the feasibility of the algorithm. Finally, three groups of experiments have been finished to validate these improved algorithm proposed in this article. We build an experimental platform and design software algorithms to conducted experiments obtain experimental data, then the experimental data are analyzed. It can be concluded that three kinds of improved algorithms play well in image processing, they are of great significance in engineering applications.
语种	中文
文献类型	学位论文
条目标识符	http://ir.ciomp.ac.cn/handle/181722/49303
专题	中科院长春光机所知识产出
推荐引用方式 GB/T 7714	张辽. 实时图像增强研究及其系统设计[D]. 中国科学院大学,2015.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
张辽.pdf（3348KB）	学位论文		开放获取	CC BY-NC-SA	请求全文