Research On Driving And Information Processing Technology Of High-frame-rate Backside-illuminated CCD In Space

Hyperspectral imaging technology has become one of the most important means of observation in space-borne remote sensing for its outstanding recognition capability. With the growth of needs on spatial and spectral resolution for spaceborne hyperspectral imagers, the requirement for sensitivity of the information acquisition system is becoming higher and higher, where high-performance detector and its application have become a key technology for further breakthroughs in system performance, and back-illuminated CCDs remain most preferred as detectors for the visible and near infrared spectral band in spaceborne hyperspectral imager with their significant advantages on quantum efficiency. This thesis conducted in-depth study in aspects of CCD driving technology and information acquisition and processing technology on the application of high-frame-rate back-illuminated CCD in spaceborne hyperspectral imaging system.For the aspect of CCD driving technology, the effects of smear of frame transfer CCD on the spectral radiation accuracy and signal to noise ratio were studied. A low line transfer time of 100 ns was achieved by means of optimization on both the circuitry and the internal structure of the CCD, making up for the greatest disadvantage for CCDs in high-frame-rate applications. The influence of dark current on system performance was studied, and a method was proposed to suppress dark current by adaptively adjusting the timing and biases of CCD drivers according to the signal of dark pixels, stabilizing dark current within a low level at different temperatures. Moreover, a method of on-orbit adjustment of bias conditions for drivers was proposed to compensate for the impact of ionizing radiation damage on surface dark current and full well charge capacity.For the CCD information acquisition technology, the influences of the driving signal wave shaping, analog front-end selection, clock jitter and grounding on system noise were studied, and a low noise information acquisition system was designed accordingly. A technique of spectrum dependent gain was proposed to enhance the SNR in violet and near-infrared spectrum bands. The issue of image crosstalk between different channels in multi-port CCDs was studied, and significantly improved by optimizing the internal structure of the CCDs. To compensate for the ionizing radiation effects on the gain and operating point of readout amplifiers, an approach of on-orbit adjustment of biases and gain compensation with analog front-end chip was proposed to make up for the impact of radiation damages to a certain extent.For the aspect of CCD information processing technology, the cause and influencing factors of the etalon effect of back-illuminated CCDs was analyzed. An approach to obtaining a reference spectrum curve through averaging in spatial dimension and smoothing in spectral dimension was proposed, with which the correction coefficients of etalon effect were calculated, leading to good results in the spectral curve correction experiments for a halogen lamp. For the issue of smear correction, an algorithm combining image data array and data of dark rows was proposed for estimating the light intensity and spectral distribution at the time of frame transfer to improve the algorithm’s ability to adapt to changes in the scene, and the effectiveness of the method was verified by experiments. In addition, a method to estimate real-time dark current based on the signal of dark pixels or the sensor temperature came out with good accuracy, so that the dark current correction process for both ground test and on-orbit operation could be greatly simplified with better temperature stabilization.This thesis conducted the design of specification requirements of a customized CCD according to the characteristics of hyperspectral imaging systems. The implementation of the system was completed, with an equivalent input referenced noise of as low as 23e- for the information circuit in high sensitivity mode, a noise level of only 70.8e- in normal operating mode at a frame rate of 500 Hz, and a dynamic range of greater than 78 d B. Under typical test conditions, the SNR of the system managed to reach above 300 within the spectrum range of 410~900nm, and the images obtained by the hyperspectral imaging experiment were of good quality, while the spectral information was also accurate.Finally, this thesis put forward some suggestions on some issues to be resolved and methods for further enhance system performance.