摘 要

随着科技发展和图像信息的大量产生和交流，目标图像的自动识别和分类应用越来越广泛，同时实际应用问题对目标识别的要求也越来越高，计算机自动识别技术远没有满足实际应用需求。要识别图像上的明显目标，就要用到图像特征提取，图像特征在图像研究领域中是一个重要的研究方向。图像特征提取不仅可以用于图像识别，还可以用于图像配准、图像分割、图像拼接等各个方面。 本文研究并实现了数字图像的点特征提取及Hough变换直线提取。

在图像的各种特征中，角点包含的信息数据量较小，即非常小的数据信息保存了图像的灰度变化的特征信息，同时外界影响因素对角点特征提取算法的影响也很小。研究图像的点特征提取方法，包括，Moravec算子、Harris算子以及SUSAN算子，编程实验得到了各自的检测结果，对算法的性能和适用性进行了对比分析。

在Hough变换中，本文主要给出Hough变换的基本原理，Hough能够检测直线、圆、椭圆和抛物线等众多解析图形。它是图像处理中的常用的检测算法，能够有效地在较大的噪声环境中提取图像中的特定信息。文中重点对Hough变换的直线检测做了分析研究。

 

关键词：图像特征；特征点提取算子；hough变换；Harris算子  

 

Abstract

 

      With a larger number of scientific and Technological Development and image information, automatic recognition and classification is applied more and more widely in the target image, at the same time, the practical application of target recognition are increasingly high requirements, the computer automatic identification technology does not meet the needs of practical application.To identify the image on the target, it is necessary to use the image feature extraction, image features in the image field is an important research direction. Image feature extraction not only can be used for image recognition, but also can be used for image registration, image segmentation, image mosaic and other aspects.

  In a variety of image features,angular point contains just a little of information, it means very small data information preserved gray change the image of the feature information .at the same time, the external factors have little effect on the corner feature extraction algorithms . The methods for extracting feature points of images include Moravec operator, Harris operator and SUSAN operator,and the programming experiment results were got respectively, finally,we analyized the performance of the algorithms and applicability. 

  In the Hough transform, this paper gives the basic principle of Hough transform, Hough transform is capable of detecting straight line, circle, ellipse and parabola and other analytic graphics. It is commonly used in the detection algorithm in image processing, which can effectively extract specific information in the image of the large noise. This paper focuses on the studying the Hough transform line detection.

 

Key words: Image feature, Feature extraction operator, hough transform，Harris operator