基于FPGA的拉普拉斯图像增强算法设计与实现

发布时间：2018-03-01 00:14

本文关键词： 拉普拉斯增强 FPGA 图像滤波图像增强　出处：《东南大学》2017年硕士论文　论文类型：学位论文

【摘要】：拉普拉斯图像增强是一种空间域的二阶微分算法,可以突出模糊的细节提取图像的边缘信息,为人们提供更加清晰锐利的视觉效果。图像滤波是为了滤除图像中的噪声,削弱噪声对后续图像处理的影响。图像处理中增强和滤波都属于底层处理,需要进行大量的并行计算,因此适合在FPGA上进行。本文首先介绍了基于FPGA的图像增强及相关算法的国内外研究现状,然后分析了底层的图像滤波及增强算法的原理,并对各个算法进行了 matlab仿真,通过对比仿真结果总结了不同滤波和增强算法的优势及劣势。本文对部分算法进行了改进,提出了适合硬件执行的方法,设计了更适合在硬件上运行的中值滤波算法,这种算法大大减少了数据比较次数,同时为了减少噪声在增强过程中的影响,提出了受限的拉普拉斯增强算法。相比于四邻域和八邻域的普通拉普拉斯增强算法,受限的拉普拉斯增强不会对图像中的噪声进行放大,很好的保护了图像的原有信息。本论文还对设计的各个模块进行了行为级的仿真,并通过RAM中预存的数据模拟相机实时数据在FPGA上进行了验证,表明了算法设计的正确性和硬件实现的可行性。各个模块的设计和分析采用了自顶向下的方法,先对顶层架构进行分析,然后再设计实现各个底层模块。图像增强和滤波整体包含外围模块和处理模块两大主要部分,外围模块主要包括3×3窗口生成模块、数据输出控制模块以及HDMI显示配置模块。处理模块是针对不同的滤波和增强算法的硬件实现,最后输出串行的处理后图像数据。模块化的设计思想使不同算法的处理模块可以复用相同的外围模块,给设计带来了很大的便捷。最后论文指出了设计存在的不足和研究的进一步方向,并提出了改进意见。
[Abstract]:Laplacian image enhancement is a second-order differential algorithm in spatial domain, which can extract edge information of images by highlighting fuzzy details, and provide people with clearer and sharper visual effects. Image filtering is to remove noise from images. The effect of weakening noise on the subsequent image processing. In image processing, both enhancement and filtering belong to the underlying processing, which requires a large number of parallel computation. So it is suitable for FPGA. Firstly, this paper introduces the research status of image enhancement and related algorithms based on FPGA at home and abroad, then analyzes the principle of image filtering and enhancement algorithm, and makes matlab simulation of each algorithm. By comparing the simulation results, the advantages and disadvantages of different filtering and enhancement algorithms are summarized. In this paper, some algorithms are improved, a method suitable for hardware execution is proposed, and a median filtering algorithm is designed, which is more suitable for running on hardware. In order to reduce the influence of noise in the enhancement process, a constrained Laplace enhancement algorithm is proposed, which is compared with the ordinary Laplace enhancement algorithm with four and eight neighbors. The restricted Laplace enhancement does not amplify the noise in the image, which protects the original information of the image. The real-time data stored in RAM is verified on FPGA, which shows the correctness of the algorithm and the feasibility of hardware implementation. The top-down method is used in the design and analysis of each module. The image enhancement and filtering are composed of two main parts: peripheral module and processing module. The peripheral module mainly includes 3 脳 3 window generation module. Data output control module and HDMI display configuration module. Finally, the serial processed image data is outputted. The modular design idea enables the processing modules of different algorithms to reuse the same peripheral modules. Finally, the paper points out the shortcomings of the design and further research direction, and puts forward some suggestions for improvement.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TN791

【参考文献】