双线阵CCD错位采样高分辨成像

High-Resolution Imaging by Bi-Linear CCD Shifting Sampling

为了克服遥感相机分辨力受电荷耦合器件(CCD)像元尺寸的限制,利用系统超过奈奎斯特频率地物信息的成像能力,研制了在线阵方向和推扫方向同时错位半个像元的双线阵CCD相机高分辨成像系统,给出了该系统提高分辨力的原理。高分辨模式下,为避免均值法插值造成的拉链效应,提出了一种基于综合梯度因子判断插值方向的单向插值法;超分辨模式下,利用现场可编程阵列(FPGA)片上随机存取存储器(RAM)对图像进行交错重组实时输出高分辨率图像。利用Wiener滤波的方法对高分辨率图像进行复原,使两种成像模式输出图像的灰度平均梯度(GMG)分别提升62.5%和78.3%。实验表明,所提出插值算法使图像边缘清晰完整,插值效果优于均值法,且易于硬件实现;两种成像模式均可降低图像频率混叠现象,分别提高1.16倍和1.6倍系统分辨力,错位采样技术实现了系统高频信息成像能力的有效利用及高分辨成像。

The resolution of remote sensing camera is restricted by the charge coupled device（CCD） pixel size.To solve the problem, imaging capability of optical system in frequency bandwidth exceeding Nyquist frequency is used to develop a camera which consists of two linear CCDs. These two CCDs are stitched together with a halfpixel width displacement in both directions along and perpendicular to the CCD axis. The principle of improving the resolution of the proposed system is presented. In hipermode, one way interpolation algorithm based on integrated gradients to determine the direction of interpolation is proposed to avoid zipper effect caused by average method. In supermode, high resolution image is acquired in real time by using random access memory（RAM） blocks inside field programmable gate array（FPGA） to regroup image stagger. Gray mean gradients（GMG） of the two high resolution images are increased by factors of 62.5% and 78.3% respectively after using Wiener filter. Results show that the proposed algorithm can make image edge clear and intact. The effect of the algorithm is better than average method and the hardware implementation is easy. The resolution is 1.16 and 1.6 times greater than that of conventional system. Two imaging method can also reduce the image frequency aliasing. Sampling phase staggered technology can take advantage of imaging capability of optical system in high-frequency bandwidth and perform high-resolution imaging.