网络环境下的遥感影像金字塔纹理压缩算法与实验

Research of Remote Sensing Pyramid Texture Compression in Network

建立不同分辨率的遥感影像金字塔模型是现阶段虚拟地球平台的主要技术手段,遥感影像数据的高效压缩是模型应用的基础。论文在分析多种图像压缩技术的基础上,提出以小波变换的JPEG2000标准来压缩网络环境中遥感影像金字塔的纹理瓦片数据。首先,介绍了小波变换的JPEG2000标准的基本原理和EBCOT算法特点,然后,结合实例实现了网络环境中遥感影像金字塔纹理压缩/解压缩的具体过程,对4幅影像图像进行5级离散小波变换,分别对在1024×1024、512×512、256×256、128×128、64×64不同分块大小和1∶15、1∶30、1∶60不同压缩倍率情况下进行图像压缩和重构耗时实验,通过改变分块大小参数和压缩倍率,对压缩性能进行了对比分析,最后,从主观视觉质量和客观辐射质量,对JPEG2000压缩、DXT压缩和JPEG压缩几种方法的压缩图像进行了评价与对比分析。实验结果表明,小波变换的JPEG2000具有高质量、高压缩率、良好的抗误码能力等特性,重构图像的视觉质量与原始图像相比人眼看不出失真,而峰值信噪比较好,是一种简单有效、易于快速实现的压缩方法,更适合于网络环境下空间遥感图像数据的近无损实时压缩。

To establish the pyramid model of remote sensing data with different resolution is the main means in Virtual Earth platform＇s construction, and the efficient compression of remote sensing data is the basis of the model application. To search for an efficient image compression method the paper analyzed a variety of image compression methods and proposed to use JPEG2000 compression standard which based on wavelet transform as the image tile＇s compression method. This paper introduced the basic framework and algorithm features of wavelet-based JPEG2000 standard, and it realized the compression and decom- pression process of image tiles based on J J2000 software under the network environment, and it implemen- ted a five-grade discrete wavelet transformation on four remote sensing mages. The paper did series of ex- periments on consumption time of the image compression and the image reconstruction under different block sizes and different compression ratios. The block sizes included 1024 × 1024, 512× 512, 256× 256, 128×128, 64×64, and 1 ： 15, 1 ： 30, 1 ： 60 were involved in the compression ratios, and it gave a per- formance comparison by changing the block size and the compression ratio. Finally, the article took the peak signal to noise ratio （PSNR） as the objective evaluation criteria of image compression quality and used standard PSNR with different compression methods to conduct an objective evaluation and a perform- ance comparison of image quality, and the compression methods included JPEG2000, DXT and JPEG. The results indicated that JPEG2000 can maintain high quality with high compression ratio and has many good characteristics, such as progressive transmission, random access stream, region of interest encoding and good capabilities of error-resilience. This study is helpful in providing a more efficient image compression method for the pyramid model application.