基于图像处理的花卉识别技术的研究与实现

发布时间：2018-06-25 15:40

本文选题：图像识别 + 特征提取　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：图像识别属于模式识别技术,能够通过计算机技术和现代信息处理技术模拟人的视觉认知和理解过程,它的主要内容是根据图像处理对图像的特征进行提取,并以此为基础对图像进行识别和分类。互联网时代图像识别技术除了被广泛应用于指纹识别、人脸识别、医学诊断、卫星航空图片解释以及遥感等这些专用领域外,也更多的应用于人们的日常生活当中。随着网络技术、计算机技术的软件硬件技术的快速发展,多媒体技术也得到了迅速发展。人们日常生活中获取的信息手段更多来源于各种多媒体设备,如摄像机、照相机、音频处理器、视频处理器及手机等。这样虽然丰富了信息形式,但也带来了信息量的激增。在信息爆炸式增长的今天,如何从海量的信息中准确且快速的提取出感兴趣的信息成为最迫切的问题。基于图像识别的“以图搜图”成为研究的热点。人们希望通过拍摄到的图片获取更多的信息。本文正是在这种背景下提出的,主要工作如下:首先,介绍了图像识别在不同领域的研究现状及应用价值,对几种较为经典的图像识别算法,以及图像识别的相关技术进行了较为深入的分析。其次,针对所研究图像的内容,特别从颜色特征和纹理特征等几个方面对图像进行较为细致的描述。分析应用LIRe图像特征检索库,选取对花卉图像有较好效果的特征描述方法,对图像特征进行提取,构建图像特征库。在Borda算法的基础上,提出了融合多种特征的识别方法,并对该方法的识别效果及性能进行了对比分析。最后,本文在以Nodejs为开发语言,以Express为开发框架所构建的花篮子系统平台上,采用NoSQL数据库——MongoDB数据库,对所提出的多特征融合识别图像的方法,设计完成了花卉图像识别功能。
[Abstract]:Image recognition is a pattern recognition technology, which can simulate the process of human visual cognition and understanding through computer technology and modern information processing technology. Its main content is to extract the features of images according to image processing. And on this basis, the image recognition and classification. In the Internet era, image recognition technology is widely used in fingerprint recognition, face recognition, medical diagnosis, satellite aerial image interpretation, remote sensing and other specialized fields, but also more used in people's daily life. With the rapid development of network technology, software and hardware technology of computer technology, multimedia technology has also been developed rapidly. People get more information from various multimedia devices, such as cameras, audio processors, video processors and mobile phones. This enriches the form of information, but also brings a surge in the amount of information. With the explosive growth of information, how to extract the interesting information accurately and quickly from the massive information becomes the most urgent problem. Image recognition based on map search has become the focus of research. People hope to get more information by taking pictures. The main work of this paper is as follows: firstly, the research status and application value of image recognition in different fields are introduced, and several classical image recognition algorithms are introduced. And the related technology of image recognition is analyzed deeply. Secondly, the image is described in detail from several aspects, such as color feature and texture feature. Based on the analysis and application of Lire image feature retrieval library, the feature description method which has good effect on flower image is selected, and the image feature library is constructed. Based on the Borda algorithm, a new recognition method is proposed, and the recognition effect and performance of the method are compared and analyzed. Finally, on the platform of basket system based on Nodejs and Express, using NoSQL database, MongoDB database, the method of multi-feature fusion and image recognition is proposed. The function of flower image recognition is designed.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S68;TP391.41

【参考文献】