无人机侦察图像压缩  

Compression for UAV reconnaissance images

作　　者：王源源[1] 黄大庆[2] 

机构地区：[1]南京航空航天大学电子信息工程学院,江苏南京210016 [2]南京航空航天大学无人机研究院,江苏南京210016

出　　处：《光学精密工程》2014年第5期1363-1370,共8页Optics and Precision Engineering

基　　金：国家自然科学基金资助项目(No.61106018);航空科学基金资助项目(No.20115552031)

摘　　要：为了减少图像数据的存储空间并高质量恢复侦察目标区域,提出了一种基于感兴趣区域分割压缩重构的新方法。首先,利用侦察目标具有规则性的特点,识别提取出感兴趣区域。然后,采用基于区域的分割算法将原图像分割成感兴趣区域(ROI)和背景区域(BG)。最后,选用基于小波变换的压缩方法,采用多级树集合分裂算法(SPIHT)嵌入式编码对分割开的ROI和BG用不同的编码比特率进行编码压缩。仿真试验证明,在同样环境下,采用本文提出的算法,感兴趣区的压缩效果比较好,恢复后图像更符合人眼视觉特性。和其他算法的处理结果比较,本文算法的图像峰值信噪比有所提高,很好地解决了高压缩比和目标图像质量之间的矛盾。To reduce the storage space of image data and to recover the high quality image of an interested region,a method of segmented compression reconfiguration based on the interested region was proposed.Firstly,the Region of Interest（ROI）was extracted from the whole image according to some regularity of surveillance targets.The original image was then divided into ROI and Region of Background（BG）by using the segmentation algorithm.Finally,with the compression method based on wavelet transform,the ROI and BG region were compressed with different compression ratios by using the embedded coding of Set Partitioning in Hierarchical Trees（SPIHT）.Simulation experiment shows that the method gives higher compression efficiency and the recovered image is more suitable for human eye under the same condition.Compared to other methods,this method increases the Peak Signal to Noise Ratio（PSNR）of image and solves the contradiction of the high compression ratio to high quality.

关 键 词：感兴趣区域(ROI) 小波变换 图像压缩 无人机 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]