多分辨率图像模糊特征快速抽取方法仿真  

Multiresolution Image Fuzzy Features Rapid Extraction Method Simulation

作　　者：李响 

机构地区：[1]南京信息工程大学大气物理学院,江苏南京210044

出　　处：《计算机仿真》2017年第9期226-229,共4页Computer Simulation

摘　　要：为了有效地提升对多分辨率图像的识别效率,需要进行多分辨率图像模糊特征快速抽取。但是采用当前方法进行多分辨率图像模糊特征抽取时,无法区分不同区域的图像特征,存在图像特征抽取误差大的问题。现提出一种基于分块分色的多分辨率图像模糊特征快速抽取方法。上述方法先融合图像边缘检测理论采用幂函数形式的模糊隶属函数对多分辨率图像进行边缘增强,得到经过增强后的多分辨率图像像素级,给出图像模糊特征边缘矩阵,基于HSV颜色空间抽取每个子图像块的每维颜色空间模式值的方差特征和均值,对图像进行分块,利用最近邻法构造分类器对多分辨率图像模糊特征进行分类,进而有效地完成了对多分辨率图像模糊特征快速抽取。仿真证明,所提方法特征抽取精确率高,有效地提升了对多分辨率图像的识别效率。A rapid extraction method of hazy feature of multi-resolution image is proposed based on segmentation and color separation. A fuzzy membership function with form of power funetion is used to earry out edge enhancement for multi-resolution image integrated with image edge detection, and pixel level of multi-resolution image after en- hancement is obtained, then a matrix of hazy feature edge of image is provided. The variance feature and mean value of pattern value of each dimension color space in each sub-image block is extracted based on HSV color space, and segmentation for image is earried out. In addition, the nearest neighbor method is used to build classifier, and the hazy feature of multi-resolution image is classified. Thus, the rapid extraction is completed. Simulation proves that the method has high extraction precision. It improves recognition efficiency of multi-resolution image effectively.

关 键 词：多分辨率图像 模糊特征 抽取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]