青 岛 科 技 大 学
本 科 毕 业 设 计 (论 文)
题 目 __________________________________
指导教师__________________________
辅导教师__________________________
学生姓名__________________________
学生学号__________________________
___________________________学院 ____________________________ 专业________________班
______年 ___月 ___日
基于FPGA的图像卷积设计与实现
摘 要
图像解卷积技术一直以来都是中外学者们浸心研究而又老而弥坚的话题,其深入研究蕴含着无限的发展前景。图像的数字化是一个二维的离散信号,对数字图像做卷积操作其实就是利用卷积核(卷积模板)在图像上滑动(划窗),将图像点上的像素灰度值与对应的卷积核上的数值相乘,然后将所有相乘后的值相加作为卷积核中间像素对应的图像上像素的灰度值,并最终滑动完所有图像的过程。图像处理中,平滑、模糊、去燥、锐化、边缘提取等等工作,其实都可以通过卷积操作来完成,不同的卷积核作用于图像,可以更清晰的获得图像的某种特征,如轮廓、颜色等。
FPGA利用其可编程性、流水操作和快速性,使其实时运行速度可以达到每秒处理数以千万条复杂指令的程序,它远远超过平常通用微处理器,是数字化电子世界中日益重要的电脑芯片。它的强大数据处理能力和高运行速度,是最值得称道的两大特色。由于它运算能力很强,速度很快,体积很小,而且采用软件编程具有高度的灵活性,因此为从事各种复杂的应用提供了一条有效途径。
本论文从FPGA进行图像解卷积着手,阐述了其重要意义及实现方法,重点专注于FPGA的运算能力强,速度快和灵活性高进行了深入的研究。在理论与实践研究的基础上,本文实现从matlab到FPGA的图像卷积算法。
关键词:图像卷积;FPGA;速度快;运算能力强;matlab;
Design and implementation of image convolution based on FPGA
ABSTRACT
Image deconvolution technology has always been an old and strong topic which is deeply studied by scholars both at home and abroad. Image digitization is a two-dimensional discrete signal. To Do convolution operation on digital image is to use convolution kernel (convolution template) to slide (window) on the image, the Pixel Gray value on the image point is multiplied by the value on the corresponding convolution kernel, and then all the multiplied values are added as the gray value of the Pixel on the image corresponding to the Middle Pixel of the convolution kernel, and eventually slide through all the images.In image processing, smoothing, and, drying, sharpening, edge extraction and so on, in fact, can be completed through convolution operation, different convolution core on the image, can more clearly obtain a certain feature of the image, such as outline, color, etc. .
With its programmability, pipelining, and rapidity, FPGA can run real-time programs that process tens of millions of complex instructions per second, much faster than ordinary microprocessors, is an increasingly important computer chip in the digital electronic world. Its powerful data processing capability and high running speed are the two most praiseworthy features. Because of its high computing power, high speed, small size, and high flexibility in software programming, it provides an effective way to engage in various complex applications.
This paper begins with image deconvolution based on FPGA, elaborates its significance and implementation method, and focuses on the deep research of FPGA for its strong operation ability, fast speed and high flexibility. On the basis of theoretical and practical research, this paper realizes the image convolution algorithm from Matlab to FPGA.
KEY WORDS:Image convolution; FPGA; high speed; strong computing ability; Matlab;
目录
图像解卷积技术一直以来都是中外学者们浸心研究而又老而弥坚的话题,图像卷积的研究在众多应用层面都蕴含着无限的发展空间。在经常用到的图像处理中,像平滑、模糊、去燥、锐化和边缘提取等等这些工作算法其实都可以通过卷积的相应操作来达到目的,对图像进行不同的卷积核操作,可以实现获得该图像的一些我们所需要的特征,例如如轮廓、颜色等等。
随着EDA技术的发展,FPGA一词出现在我们视野的频率越来越高。FPGA利用其自身的诸多优点,在科技高速发展的现时代它扮演着重要的角色。它具有可编程性能够为用户提供想要实现的多种逻辑容量、特性、速度和电压参数的标准成品部件。而且FPGA处理复杂指令的速度可以达到每秒计以千万条,而在常见的微处理器中与FPGA相比来说都会逊色很多。可以发现FPGA已经成为科技数字化电子生活中无比重要的数据处理芯片。FPGA最大的两个优点为:具备强大的数据运行处理能力和高