视频图像中车辆的车型识别算法研究及实现.doc

毕业设计说明书 作 者： 学 号： 系： 信息工程系 专业： 电子信息工程 题 目： 视频图像中车辆的车型识别 算法研究与实现 指导者： 评阅者： 2013年 6月 1日 毕业设计（论文）中文摘要 视频图像中车辆的车型识别算法研究与实现 摘要： 主成分分析法是一种常用的车型识别方法，该方法通过对视频进行分帧后，提取彩色图像进行预处理，主要包括灰度图像变换、图像平滑、图像分割、图像横向填充与纵向填充，再进行特征提取和特征识别（即判别车型）。预处理阶段的图像分割效果直接关系到特征提取和车型判别结果的准确性，该课题对图像进行分割采用了两种不同方法：灰度阈值分割法和图像背景差值法。在把待识别车辆抽象成“工”字型的前提下，提取待识别车辆的顶长比、顶高比、前后比来进行车型的判别。该课题应用两种方法对车型不同、背景不同和车辆位置不同的图像进行验证，经过分析比较发现，基于不同图像分割方法的两种车型识别技术各有优缺点，但在实际应用中与基于灰度阈值分割法的车型识别技术相比，基于图像背景差值法的车型识别技术要更加快捷、准确。 关键词： 灰度变换 灰度阈值分割 背景差值 特征提取 车型判别 毕业设计（论文）外文摘要 Title Video Images in the Research and Implementation of Vehicle Model Recognition Algorithm Abstract Principal component analysis (PCA) method is a commonly used model identification method, the method is based on the video framed, then extracting the color image for preprocessing, which includes gray image transformation, image smoothing, image segmentation, image transverse filling and vertical filling , at last extracting features and matching features(namely model identification). The effect of image segmentation is directly related to the accuracy of the result, the subject of image segmentation uses two different methods: gray-level threshold segmentation method and image background difference method. Stay in the premise of vehicle abstraction into the premise of work word, extracted to draw the top long than, peak height ratio, before and after than for model identification. The subject uses two methods to check models of different types, different backgrounds and different locations, after analysis and comparison, found that two model identification technique based on different kinds of image segmentation methods has its advantages and disadvantages, but in practical application, compared to model identification technique based on