【作者】 王畅；

【导师】 李峰；

【作者基本信息】 长沙理工大学 ， 计算机应用技术， 2007， 硕士

【摘要】 随着数字图像采集技术和处理技术的飞速发展,图像已成为人们获取信息的重要途径,图像的边缘信息反映了图像中最有价值的信息,边缘检测是图像处理和计算机视觉中最重要、最经典的课题之一。小波变换在空域中分辨率随频率的大小而调节,低频粗疏,高频精密,在小尺度参数的边缘检测算子能够检测出灰度发生的细变化,而大尺度参数的边缘检测算子能够检测出灰度发生的粗变化,使用小波多尺度变换可以更好的检测图像的边缘和细节,能够很好的将信号与噪声分离,但在使用小波多尺度变换进行图像边缘检测的时候,虽然可以一定程度的去除噪声点的干扰,同时也会去除掉一些较弱的信号边缘,对模糊边缘的提取有一定的缺陷。如果同时利用模糊理论来进行边缘检测,就能更好的将弱边界图像从背景中分离出来。根据以上这些特点,本文提出了一种新的基于小波多尺度变换和模糊方法的图像边缘检测法,结合了小波多尺度和模糊方法,既能在很大程度上克服噪声的影响,又能保持丰富的边缘细节,还能更有效的将图像的弱边界从背景中分离出来。本文将小波多尺度变换理论和模糊边缘检测方法相结合应用于图像的边缘检测中,将图像通过滤波器分为高频和低频两部分分别处理,高通滤波器目的是去除或衰减低频带LL的小波系数,而保留高频带的小波系数,低通滤波器目的是保留低频带的小波系数,而衰减高频带的小波系数,去除图像中大量的边缘信息,对于高频部分使用小波多尺度变换的方法进行边缘检测,低频部分则利用竞争模糊边缘检测方法进行处理,最后对两种方法得到的边缘图像进行融合。在matlab7.0的编程环境下进行了实验比较,使用本文算法最终检测出来的图像边缘结合了多尺度边缘检测和模糊边缘检测的优点,克服两种边缘检测的局限性,与以往边缘算子检测出的边缘相比有很大的改善。更多还原

【Abstract】 Along with the rapid development of digital image gathering and processing technology, image has become an important way of people to gain information. Image edge information reflects the most valuable information of the image. The Edge detecting is one of the most important and classic topic in the image processing and computer vision.The resolution in wavelet transformation adjusts in the time zone along with the frequency size. Low frequency is careless. High frequency is precise. Using the edge by the small scale which can detect the small transformation of the gray scale, and using the edge by the great scale which can detect the great transformation of the gray scale, using the wavelet multi-scale transfomation can detect the edge and detail much better, it also can separate the signal and noise much better. When we use the wavelet multi-scale transformation to detect the edge of image, to a certain degree, the plan can take off the noise disturbance, it also will take off some weak signal of the edge. It has limitation to detect blur border. If we use the fuzzy theory to do the detection at the same time, we can separate the weak border image from the background much better. Base on such characteristic, the paper considered the drawback of the classic edge detection, present a new image edge detect method bases on the wavelet multi-scale transform and fuzzy method. The new method which was combine the multi-scale and fuzzy method would overcome the effect of noise efficiently, hold the richness details of the edge and can effectively separated the weak border of the image form the background.In this paper, wavelet Multi-scale transform and fuzzy edge detection algorithm was combined in the image edge detection. The method divided the image into two groups of high frequency and low frequency and deal with separately. The high pass filter rejector’s aim is wipe off the wavlet’s parameter of low frequency band and hold the wavelet’s parameter of high frequency band, the low pass filter rejector’s aim is hold the wavelet’s parameter of low frequency band and attenuate wavelet’s parameter of high frequency band. The high frequency of the image was used the wavelet multi-scale transform for the edge detection. The low frequency of the image was used the Compete Fuzzy edge detection algorithm for the edge detection. Then fusion the two images which come from the two methods. The experiment result show that the edge come from the new method combines the advantage of Multi-scale edge detection and Compete fuzzy edge detcion method, overcome the limitation of two methods and gets a significant improvement compared with other edge operators.更多还原