高光谱数字图像无失真实时压缩方法仿真
|
摘要
为了提升高光谱数字图像的质量,需要进行高光谱数字图像无失真实时压缩方法的研究。但是采用当前方法对图像进行实时压缩时,无法对降维后的光谱矢量进行分类,导致高光谱数字图像失真率较高的问题。为解决上述问题,提出一种基于DPCM的高光谱数字图像无失真实时压缩方法。上述方法先利用自适应波段选择方法对高光谱图像进行降维,并对降维后的光谱矢量分类,确定图像各个波段的预测顺序,计算出相邻波段的相关系数,并进行自适应波段分组,计算出类内部分像素,进行最优预测系数的训练,融合DPCM计算出不同波段的预测残差图像的标准差,设计基于均方差最小的线性预测器,利用上述预测器对图像各波段进行预测,对各波段预测残差图像进行SPIHT编码,消除同类像素的谱间相关性,并完成对高光谱数字图像无失真实时压缩。实验结果证明,所提方法压缩精度较高,可以较为全面的保留高光谱数字图像的纹理信息。
An undistorted real-time compression method for hyperspectral digital image is proposed based on DPCM. Firstly,the dimension of hyperspectral image is reduced by using selection method of adaptive band,and spectral vector of dimension reduction is classified,thus prediction order of each image band is determined,and correlation coefficients of adjacent bands are calculated. Then,the bands are adaptively divided into groups,the pixel within class is calculated to carry out training of optimal prediction coefficient,the DPCM is fused to calculate standard deviation of predicted residual image of different bands,and the minimum linear predictor is designed based on mean square deviation. Finally,each band of image is predicted by using the minimum linear predictor,and SPIHT encode for prediction residual image of each band is carried out,meanwhile spectral correlation among similar pixels is eliminated,thus undistorted real-time compression for hyperspectral digital image is achieved. The simulation results demonstrate that this method has high compression accuracy. It can entirely preserve texture information of hyperspectral digital images.
An undistorted real-time compression method for hyperspectral digital image is proposed based on DPCM. Firstly,the dimension of hyperspectral image is reduced by using selection method of adaptive band,and spectral vector of dimension reduction is classified,thus prediction order of each image band is determined,and correlation coefficients of adjacent bands are calculated. Then,the bands are adaptively divided into groups,the pixel within class is calculated to carry out training of optimal prediction coefficient,the DPCM is fused to calculate standard deviation of predicted residual image of different bands,and the minimum linear predictor is designed based on mean square deviation. Finally,each band of image is predicted by using the minimum linear predictor,and SPIHT encode for prediction residual image of each band is carried out,meanwhile spectral correlation among similar pixels is eliminated,thus undistorted real-time compression for hyperspectral digital image is achieved. The simulation results demonstrate that this method has high compression accuracy. It can entirely preserve texture information of hyperspectral digital images.
引文
[1]何彩莲,等.马铃薯光谱及数字图像特征参数对氮素水平的响应及其应用[2].光谱学与光谱分析,2016,36(9):2930-2936.
[2]杨新锋,胡旭诺,粘永健.基于分类的高光谱图像压缩算法[J].红外与激光工程,2016,45(2):263-266.
[3]李朝晖,等.基于独立分量分析的高光谱图像降维与压缩算法[J].计算机与数字工程,2014,42(8):1472-1475.
[4]徐大卫,张荣,吴倩.多尺度字典学习的高光谱图像压缩算法[J].遥感学报,2015,19(2):000263-272.
[5]韩勇,陈善学.基于小波域点到线模型的高光谱图像压缩算法[J].计算机应用研究,2014,31(7):2238-2240.
[6]冯慧,等.基于高光谱成像和主成分分析的水稻茎叶分割[J].激光生物学报,2015,24(1):31-37.
[7]王忠良,等.高光谱图像的分布式压缩感知成像与重构[J].光学精密工程,2015,23(4):1131-1137.
[8]吕腾达,刘成.探空火箭振动遥测数据压缩算法设计[J].电子设计工程,2016,19(24):28-30.
[9]莫云华,张文娟,张连蓬.基于PROSAIL模型的高光谱遥感图像模拟研究[J].红外,2016,37(9):1-7.
[10]宋家慧.基于压缩传感的SRM断面图像重建方法研究[J].计算机仿真,2016,33(2):132-137.
[2]杨新锋,胡旭诺,粘永健.基于分类的高光谱图像压缩算法[J].红外与激光工程,2016,45(2):263-266.
[3]李朝晖,等.基于独立分量分析的高光谱图像降维与压缩算法[J].计算机与数字工程,2014,42(8):1472-1475.
[4]徐大卫,张荣,吴倩.多尺度字典学习的高光谱图像压缩算法[J].遥感学报,2015,19(2):000263-272.
[5]韩勇,陈善学.基于小波域点到线模型的高光谱图像压缩算法[J].计算机应用研究,2014,31(7):2238-2240.
[6]冯慧,等.基于高光谱成像和主成分分析的水稻茎叶分割[J].激光生物学报,2015,24(1):31-37.
[7]王忠良,等.高光谱图像的分布式压缩感知成像与重构[J].光学精密工程,2015,23(4):1131-1137.
[8]吕腾达,刘成.探空火箭振动遥测数据压缩算法设计[J].电子设计工程,2016,19(24):28-30.
[9]莫云华,张文娟,张连蓬.基于PROSAIL模型的高光谱遥感图像模拟研究[J].红外,2016,37(9):1-7.
[10]宋家慧.基于压缩传感的SRM断面图像重建方法研究[J].计算机仿真,2016,33(2):132-137.