基于自分类和颜色空间变换的图像特征自动检测  

Automatic detection of image features based on self classification and color space transformation

作　　者：郭高鹏[1] GUO Gaopeng(Shangluo University,Shangluo Shanxi 726000,China)

机构地区：[1]商洛学院,陕西商洛726000

出　　处：《自动化与仪器仪表》2020年第8期33-36,共4页Automation & Instrumentation

摘　　要：为解决传统方法容易受图像噪声、大小变化与旋转影响,不易提取准确的图像特征,图像特征检测结果准确率低,稳定性差的问题,提出一种新的基于自分类和颜色空间变换的图像特征自动检测方法。通过主成分分析法对图像进行降维处理,将高维数据映射至低维空间中,降低计算复杂度与计算难度。将RGB颜色空间转换至HSV颜色,把图像看作二维信号,对图像特征进行提取时,对其三个颜色分量进行小波分解与重构处理,获取小波变换公式与逆变换公式,描述图像特征,按照小波转换结果,利用支持向量机实现图像特征自动检测。实验结果表明,所提方法和Harris方法、SIFT方法相比,能够快速准确地检测出图像特征,且对图像特征的聚类性最好,检测稳定性最高。In order to solve the problems of traditional methods,which are easily affected by image noise,changes in and rotation,difficult to extract accurate image features,low accuracy of image feature detection results and poor stability,a new automatic image feature detection method based on self classification and color space transformation is proposed.Principal component analysis is used to reduce the dimension of image,map the high-dimensional data to the low-dimensional space,and reduce the computational complexity and difficulty.Convert the RGB color space to HSV color and treat the image as a two-dimensional signal.When the image features are extracted,the three color components are decomposed and reconstructed by wavelet,the wavelet transform formula and inverse transform formula are obtained,and the image features are described.According to the results of wavelet transform,the image features are automatically detected by support vector machine.The experimental results show that compared with Harris method and sift method,the proposed method can detect image features quickly and accurately,and has the best clustering and the highest detection stability.

关 键 词：自分类 颜色空间变换 图像特征 检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]