一种有效评估压缩对影像特征提取影响的方法  

A Method for Evaluate the Effects of Compression to Image Feature Extraction

作　　者：王昱[1,2] 杨秀策[2] 李晨钊 

机构地区：[1]武汉大学遥感信息工程学院 [2]西安测绘研究所 [3]解放军78155部队

出　　处：《光子学报》2013年第5期601-604,共4页Acta Photonica Sinica

基　　金：国防技术基础基金(No.J08302)资助

摘　　要：在分析压缩对于影像特征提取影响的基础上,提出了一种定量化评价压缩对于影像线特征提取的方法.该方法首先对影像进行特征提取,然后使用变化检测算法统计压缩对于特征的影响.利用JPEG2000算法对一幅SPOT5影像进行压缩的试验表明,该方法可以获得连贯的边缘曲线,且具有较强的去除噪声的能力,其线特征提取结果适用于进行特征变化的比较;能够较好地反映出不同倍数的压缩影像信息量的损失情况,线特征变化率越大,压缩影像信息损失量就越大.Based on analysis of the effects of compression to image feature extraction,a quantitive method is proposed to evaluate the effects of compression to image feature extraction.The feature of image is extracted,and change detection method is used to statist the effect of compression.The compression test used JPEG2000 to SPOT5 image showed that the proposed method can obtain the coherenceedge curve,and has strong ability to remove noise;line feature extraction results are applicable to feature change comparison;the method can also reflect the different image information loss of multiple compressed ratio,the higher of line feature change rate,and the more of compression image information loss.

关 键 词：影像压缩 线特征 CANNY算子 峰值信噪比 

分 类 号：TP751.2[自动化与计算机技术—检测技术与自动化装置]