摘要
针对传统的压缩图像特征提取方法,忽略了压缩过程对图像像素频率波动带来的失真,导致提取结果误差大的问题。提出采用基于统计特征融合的SAR压缩图像特征提取方法,依据嵌入式无线传感器中图像像素传递的波散射原理,真实地反映图像像素的灰度值,将图像灰度和纹理特征相结合,利用模糊C均值法对SAR图像像素特征进行聚类,将分类后的压缩图像特征进行提取。仿真结果表明,改进方法在进行嵌入式无线传感器中压缩图像特征提取时,提取数据量提高了约5.59T,误差率降低了约10.25%,具有良好的实用性。
In the paper, we proposed a method of SAR compressed image feature based on fused statistical features. According to the theory of wave scattering which can truly reflect the gray values of the image pixel, the image gradation and textural features were combined, and SAR image pixel features were clustered based on the method of fuzzy c-means. Then the classified compressed image features were extracted. Simulation results indicate that during extracting the features by improved method, the data size is increased by about 5.59T, and the error rate is reduced by about 10.25%.
出处
《计算机仿真》
北大核心
2017年第1期344-347,共4页
Computer Simulation
关键词
嵌入式
无线传感器
压缩图像
特征提取
Embedded
Wireless sensor
Compressed images
Feature extraction
