【作者】 谢艳玲；

【导师】 洪金益；

【作者基本信息】 中南大学 ， 地图学与地理信息系统， 2009， 硕士

【摘要】 边缘是图像中重要的特征之一,反映了图像中最有价值的信息之一。小波变换具有数学显微镜的特征,使用小波变换多尺度可以很好的检测图像的边缘和细节。模糊方法可以通过设置不同的隶属度函数来检测图像的边缘。本文将模糊方法和小波多尺度变换理论用于图像的边缘检测中,并对每种方法的边缘检测结果进行了详细的分析和比较,通过分析和比较,说明了模糊方法和小波变换与传统的边缘检测方法相比在边缘检测方面的优越性,并将模糊方法和小波变换的边缘检测效果进行了详细的比较。对模糊方法在原有的理论的基础上进行了改进,在快速模糊算法的基础上加入了阈值,使得边缘检测的效果更好,将直方图和模糊方法结合起来,运用多阈值的模糊检测方法,与双阈值的方法检测的边缘效果相比,提取的边缘信息更加丰富。为了检测模糊方法和小波变换的边缘检测的抗噪性能,在原有的图像中加入了椒盐噪声和不同类型的高斯噪声,对噪声图像的边缘检测结果进行了详细的分析和比较,并与传统的边缘检测方法的抗噪效果进行了分析和比较,显示了模糊方法和小波变换在抗噪方面的优越性。本文提出了一种新的多光谱图像的边缘检测方法,将高空间分辨率的全色图像的边缘检测结果应用于多光谱图像的边缘检测中,并用模糊方法和小波变换来检测提出的新方法的可行性,扩展了边缘检测的应用空间,也为遥感图像的解译提供了方便。更多还原

【Abstract】 Edge, one of the most important characteristic in the image, communicates very valuable information. Fuzzy method can detect image edge by setting different subject functions. And wavelet transform has the character of math microscope which makes the detection of the edge and details perform excellent in the image processing.This paper applied both wavelet Multi-scale transform and fuzzy method in the image detection. Then it analyzes and compares the results of both methods in detail. It shows that wavelet Multi-scale transform and fuzzy method are more advantageous than the traditional methods. This paper makes an improvement on traditional fuzzy method based on fuzzy theory by adding a restrain value in the fast fuzzy method. By combining the fuzzy method with histogram, it puts forward a multi-value fuzzy method which can detect more precise information of image edge compared with the two-value method. In order to test the new method’s performance of edge detection, some salt & pepper and Gaussian noise is added to the original image, and then analyze the results of edge detection of noisy images in detail. It shows the advantage of wavelet Multi-scale transform and fuzzy method in noise resistance, compared with the traditional methods. This paper propounds an original method to detect the Multi-spectrum image’s edge, which applies the edge detection result of the panchromatic image to the multi-spectrum image edge detection, and employs the wavelet transform and fuzzy method to test the new method’s feasibility. This method expands the range of the edge detection means, also promotes convenience for the remote image’s interpretation.更多还原