特定虚拟图像局部多变特征识别仿真研究

采用当前方法识别特定虚拟图像中存在的局部多变特征时,识别多变特征所用的时间较长,得到的识别结果与实际不符,存在识别效率低和识别准确率低的问题。提出特定虚拟图像局部多变特征识别方法,在不改变原有样本协方差结构的基础上,采用PCA降维算法调整类间离散矩阵和类内离散矩阵,最大化类间聚类、最小化类内距离,去除特定虚拟图像中存在的无用数据和冗余信息。融合非局域全变分和结构张量构建图像去噪模型,利用图像去噪模型去除特定虚拟图像中存在的噪声。对预处理后的特定虚拟图像进行Contourlet变换,在不同方向、不同尺度上提取特定虚拟图像的变换系数,构建语言变量,通过模糊逻辑方法计算模糊区域在模糊特征空间中对应的激活强度值,获得特定虚拟图像局部多变特征向量,将特征向量输入支持向量机分类器中,实现特定虚拟图像局部多变特征的识别。仿真结果表明,所提方法的识别效率高、识别准确率高。

Simulation Research on Local Multivariate Feature Recognition of Specific Virtual Images

When the current method was used to recognize the local multi-variable features in specific virtual image,the recognition efficiency and the recognition accuracy were low.Therefore,a method to recognize the local multi-variable features of virtual image was proposed.Without changing the original sample covariance structure,PCA dimensionality reduction algorithm was used to adjust the inter-class discrete matrix and the intra-class discrete matrix,so as to maximize the intra-class clustering and minimize the inter-class distance.And then,the useless data and redundant information in specific virtual image were removed.The denoising model was constructed by combining non-local total variation and structural tensor,and then the image denoising model was used to remove the noise in specific virtual image.Moreover,Contourlet transform was performed on the pre-processed virtual image,and the transform coefficients of specific virtual image were extracted in different directions and different scales.In addition,the linguistic variables were constructed,and the fuzzy logic method was used to calculate the corresponding activation intensity of fuzzy regions in the fuzzy feature space.Finally,the local variable feature vector of specific virtual image was input into the classifier of support vector machine to recognize local multi-variation features of the specific virtual image.Simulation results prove that the proposed method has higher recognition efficiency and recognition accuracy.