CCD Subpixel Imaging System And High-resolution Image Reconstruction Technique

Charge-coupled device(CCD) is widely used in aviation and aerospace remote sensing imaging, and plays an important role in various fields, such as environmental monitoring, resource surveys, topographic mapping and military reconnaissance. With development of remote sensing application technology, higher spatial resolution of space camera is continuously requested. There are two traditional methods to improve spatial resolution of remote sensing camera, increasing the focal length of optical system and reducing the CCD pixel size. Improving spatial resolution of remote sensing camera using traditional methods will result in volume, weight, and costs of space camera increasing significantly, and system MTF, SNR, and image quality decreasing. Especially in infrared imaging field, due to the special nature of the infrared spectrum and limitations of the infrared photosensitive material, compared to the visible spectrum detectors, pixel size and pixel pitch of infrared detector are generally much larger, so spatial resolution of infrared imaging system is lower with the same optical system. Therefore, volume and weight of space camera will increase quickly using traditional methods to improve the spatial resolution, and optics and mechanism of space camera is difficult to realized and requirements of the satellite platform is more rigorous.Research on how to improve the spatial resolution and enhance image quality of the imaging system with the existing system has become a hot issue. CCD subpixel imaging technology has been researched broadly in recent years. CCD subpixel imaging technology is effective to enhance the spatial resolution without changing the focal length of optics and the CCD pixel size. In the case of having the same or nearly same spatial resolution, it can reduce the focal length of optics, and volume, weight of space camera decrease consequently. With development of space camera toward high resolution and miniaturization, it is significant to research and explore CCD subpixel imaging technology.For CCD subpixel imaging technology, how to obtain the low-resolution images with sub-pixel displacement and how to reconstruct the high-resolution image are two key problems. Based on that, realization of CCD subpixel imaging system and high-resolution image reconstruction technique has been mainly researched in this paper. In this paper, the main research contents and achievements can be summarized as follows.First, quantitative evaluation of image quality of subpixel imaging based on MTF is performed. A mathematical model for simulating CCD subpixel imaging with different CCD pixel fill factor is established, and simulation experiment using a sub-image of ISO12233 standard resolution card is performed on Matlab. Quantitative evaluation based on MTF and simulation on Matlab indicates that CCD subpixel imaging technology can improve MTF and enhance image quality. Considering image quality, the amount of data, SNR and system complexity of optical remote sensing imaging system in the round, the 1/2 pixel subpixel imaging mode using two linear CCD and four point subpixel imaging mode is most reasonable, the proposed method has some reference value to the design of CCD subpixel imaging system.Second, an area CCD subpixel imaging system based on high precision two-axis translation table is realized using self-designed KAI-1020 imaging system. Multiple low-resolution images which shift half pixel with each other are achieved according diagonal subpixel imaging mode and four point subpixel imaging mode. Registration by using SIFT method is performed to achieved multiple low-resolution images and result show that the average shift error is 0.015 pixels and the maximum shift error is 0.05 pixels. The method proposed to implement subpixel imaging reduce shift precision requirement and is easy to realize, overcoming some disadvantages in existing subpixel imaging realization ways, such as complex structure, difficult to implement, high cost, high subpixel precision and so on.Third, a cubic B-spline high-resolution reconstruction algorithm based on image gradient is proposed. Two low-resolution images shift half pixel with each other that obtained in diagonal subpixel imaging mode are interpolated and reconstructed using proposed algorithm. Compared to traditional reconstruction algorithm, image quality evaluation parameters of reconstructed image are improved using proposed algorithm. The proposed algorithm also reduces the distortion of reconstructed image edge detail and improves the image quality of reconstructed image.Forth, image restoration combined with CCD subpixel imaging to enhance image quality together is proposed. An improved Wiener filtering algorithm based on MTF is proposed and performed to compensate MTF of reconstructed high-resolution image. Experimental results demonstrate that CCD subpixel imaging can improve image resolution to 1.4 times at least, reduce image aliasing and enhance image quality. Image restoration using proposed improved Wiener filtering algorithm not only can enhance image detail, but can suppress noise and improve SNR of reconstructed image. Image restoration can effectively compensate MTF of reconstruction images of subpixel imaging, which solve the problem of resolution enhancement but MTF reduction in subpixel imaging.Fifth, an application and implementation plan using CCD subpixel imaging is proposed. CCD diagonal subpixel imaging system based on prism is implemented by using two same CCD sensors. The two CCD are fixed manually which shift half pixel in two-dimension direction. The CCD imaging system is designed based on FPGA. The two images obtained by two CCDs shift half pixel in two-dimension direction with each other. Reconstruction and image restoration are realized real-time on hardware based on FPGA controller. Final high-resolution image that spatial resolution is improved and MTF is compensated is output and obtained by CCD subpixel imaging system. CCD subpixel imaging system based on prism is implemented on hardware using two CCD ICX415 ALs of Sony Corporation as image sensor and Altera’s FPGA EP4CE30F484 as controller.