毕业设计（论文）自然场景图像局部不变特征检测与描述.docx

班级 1—0713学号 本科毕业设计论文 题 目 自然场景图像局部不变特征检测与描述 学 院 通信工程学院 专 业 通信工程专业 学生姓名 导师姓名 摘 要随着现今新的应用领域和研究对象的不断涌现，图像特征提取分析成为图像处理以及计算机视觉等领域的一个重要研究内容，是许多问题的研究基础，例如图像匹配、三维重建、数字水印、视频检索、目标识别、目标分类、视频数据挖掘等技术。如何提取图像的稳定特征并对其进行描述成为了相关领域的研究重点。 本文针对此图像的特征提取与匹配问题，首先简单介绍了Harris算法，SUSAN算法和SIFT算法等几种常见的经典特征提取与匹配算法，并重点介绍了在鲁棒性及计算效率上具有优势的SURF算法，其次，在文章中较为详细的阐述了SURF算法的原理和应用，通过在Microsoft Visual C++ 2008开发环境下的应用，对自然场景图像局部不变特征进行了测试与描述，最后对拍摄的自然场景的图像进行了特征点的标记，并且采用欧氏最短距离法，计算最相近的两个特征向量的距离来判断图像是否匹配。 实验表明，SURF算法可以运用于图像特征性的检测与描述。并且具有较高的计算效率，在未来的发展中有较好的应用前景。关键字： 图像匹配 特征点 检测SURFABSTRACT With the new application areas and the current study continue to emerge, image feature extraction and analysis as image processing and computer vision is an important research areas such as content, the research base of many problems, such as image matching, reconstruction, digital watermarking, video retrieval, object recognition, object classification, video, and data mining technology. How to extract the stable features of the image and its description as the related areas of research focus. ?In this paper, the image extraction and matching, the first introduces the Harris method, SUSAN algorithm and SIFT algorithm several common classical extraction and matching algorithms. And highlighted the robustness and computational efficiency advantage of the SURF algorithm, and secondly, in the article a more detailed description of the SURF algorithm theory and applications, through the Microsoft Visual C + + 2008 development environment applications, of natural scenes local invariant features of images and description of the test, the last scene of the shooting images of the natural feature points of the mark, and the shortest Euclidean distance method used to calculate the closest distance of two feature vectors to determine whether the image matching. ?Experiments show that, SURF algorithm can be applied to the image characteristic detection and description. And has high efficiency in