摘要植物识别是植物科学研究领域和农林业生产经营中重要的基础性工作,植物识别学是一项具有长远意义的基础性研究,其主要的识别依据是植物的外观特征,包括叶、花、枝干、树皮、果实等。因此,花卉识别是植物识别学的重要部分,利用计算机进行花卉种类识别具有重要意义。84460
本文主要介绍了花卉识别软件的主要设计原理和过程。在获取到花朵图片的基础上,提出了图片预处理的概念,并确定了处理步骤。通过对图像进行灰度化处理来减少计算量,用均衡化算法来对图像进行增强,再是用中值滤波算法对噪声进行消除,最后对图像进行归一化,让图像被处理在限定的范围内。本课题中特征提取使用的是纹理特征和颜色特征相结合,用SIFT算法提取纹理特征,用颜色直方图来提取颜色特征,并通过SVM分类器实现花卉识别。本文构建了基于安卓的花卉识别系统,并用十三种花卉对系统进行了测试,达到了的很高的准确率。实验结果表明,系统具有较高的识别准确率和稳定性。
毕业论文关键词: 实时识别;安卓;花朵识别
Abstract Plant recognition is groundwork in botanical studies, and in agro-forestry production and management。Plat taxonomy is a fundamental research field whose recognition criteria rely heavily on the appearances of plants such as leaves, flowers, branches, bark, and fruits。 The recognition of flower is a signification part of plant taxonomy。 Automatic recognition of flowers with high fidelity using computer technology is of great social benefits。
This paper mainly introduces the main design principles and process of flower recognition software。 The paper proposes the concept of image preprocessing based on getting flower pictures, and determines the processing steps。 By using gray processing to reduce the amount of computation。 Using equalization algorithm for image enhancement。 Then, using median filtering to eliminate the noise。 Finally, normalizing the image, let the image processed in a limited range。 The subject uses the combination of texture features and color features as feature extraction。 SIFT algorithm is used to extract texture features, color histogram is used to extract the color feature, and using the SVM classifier to realize the recognition of flowers。 The paper constructs a flower recognition system based on Android, and tests the system with thirteen kinds of flowers,meanwhile achieves a high accuracy rate。 The experiment result shows that the system has high recognition accuracy and stability。
Keywords: Real-time Recognition;Android;Flower Recognition
目录
第一章 绪论 1
1。1 开发该系统的依据及意义 1
1。3 课题研究内容及成果 2
1。3。1 课题研究内容 2
1。3。2 课题研究成果 3
第二章 图像预处理与特征提取 4
2。1图像预处理 4
2。1。1 灰度化 4
2。1。2 图像增强 5
2。1。3 噪声消除 6
2。1。4 归一化 7
2。2 特征提取 8
2。2。1 SIFT特征 8
2。2。2 颜色特征 11
第三章 基于sift特征和颜色直方图的图像识别 安卓SIFT算法花卉识别系统设计+源程序:http://www.youerw.com/jisuanji/lunwen_100226.html