In the view of space remote sensing，such as satellite detection，space debris detection etc.，visible band is usually used，in order to have the all-weather detection capability, long wavelength infrared (LWIR) detection is also an important supplement. However, in the tow wave band, the earth can be a very strong interference source, especially in the dim target detecting. When the target is close to the earth, especially the LEO target, the background radiation of the earth will also enter into the baffle, and became the stray light through reflection, the stray light can reduce the signal to clutter ratio (SCR) of the target and make it difficult to be detected. In the visible band, the solar albedo by the earth is the main clutter source while in the LWIR band the radiation of the earth is the main clutter source. So, in this paper, we establish the energy transformation from the earth background radiation to the detection system to assess the effects of the stray light. Firstly, we discretize the surface of the earth to different unit, and using MODTRAN to calculate the radiation of the discrete point in different light and climate conditions, then, we integral all the radiation which can reach the baffle in the same observation angles to get the energy distribution, finally, according the target energy and the non-uniformity of the detector, we can calculate the design requirement of the system stray light suppression, which provides the design basis for the optical system.
Point spread function (PSF) is a very important indicator of the imaging system; it can describe the filtering characteristics of the imaging system. The image is fuzzy when the PSF is not very well and vice versa. In the remote
sensing image process, the image could be restored by using the PSF of the image system to get a clearer picture. So, to measure the PSF of the system is very necessary. Usually we can use the knife edge method, line spread function (LSF) method and streak plate method to get the modulation transfer function (MTF), and then use the relationship of the
parameters to calculate the PSF of the system. In the knife edge method, the non-uniformity (NU) of the detector would
lead an unstable precision of the edge angle; using the streak plate could get a more stable MTF, but it is only at one
frequency point in one direction, so it is not very helpful to get a high-precision PSF. In this paper, we used the image of the point target directly and combined with the energy concentration to calculate the PSF. First we make a point matrix target board and make sure the point can image to a sub-pixel position at the detector array; then we use the center of
gravity to locate the point targets image to get the energy concentration; then we fusion the targets image together by
using the characteristics of sub-pixel and get a stable PSF of the system. Finally we use the simulation results to confirm
the accuracy of the method.
The performance of small and dim IR target detection is mostly affected by the signal to noise ratio(SNR) and signal to clutter ratio(SCR), for the MWIR especially LWIR array detector, because of the background radiation and the optical system radiation, the SCR cannot be unlimited increased by using a longer integral time, so the frame rate of the detector was mainly limited by the data readout time especially in a large-scale infrared detector, in this paper a new MWIR array detector with windowing technique was used to do the experiment, which can get a faster frame rate around the target by using the windowing mode, so the redundant information could be ignore, and the background subtraction was used to remove the fixed pattern noise and adjust the dynamic range of the target, then a local NUC(non uniformity correction) technique was proposed to improve the SCR of the target, the advantage between local NUC and global NUC was analyzed in detail, finally the multi local window frame accumulation was adopted to enhance the target further, and the SNR of the target was improved. The experiment showed the SCR of the target can improved from 1.3 to 36 at 30 frames accumulation, which make the target detection and tracking become very easily by using the new method.
It is always affected by the influence of limb atmosphere when the space-based remote sensing systems detect spatial targets using limb observation mode. In this paper, the characteristics of the limb atmosphere and the impact of limb atmosphere to target observation are theoretical modeled. Based on the model, we propose an algorithm to compute the vertical structure of atmosphere radiance through the image of limb atmosphere as well as the star image. Realization of atmosphere radiance under similar situation can then be computed based on inversion algorithm proposed in the paper. The stellar images of different areas including areas over Antarctic and Equator are captured by in-orbit space borne camera. The method of how to inverse from the gray image to atmosphere limb radiance in engineering applications is described in detail and statistical analysis of the result of inversion to limb atmosphere radiance is conducted whose trend is consistent with simulation result of MODTRAN which increases at lower altitude to a peak value then drop to zero slowly while there are two peaks in the statistical radiance distribution curves illustrating the polar light over Antarctic.
AS infrared CMOS Digital TDI (Time Delay and integrate) has a simple structure, excellent performance and flexible operation, it has been used in more and more applications. Because of the limitation of the Production process level, the plane array of the infrared detector has a large NU (non-uniformity) and a certain blind pixel rate. Both of the two will raise the noise and lead to the TDI works not very well. In this paper, for the impact of the system performance, the most important elements are analyzed, which are the NU of the optical system, the NU of the Plane array and the blind pixel in the Plane array. Here a reasonable algorithm which considers the background removal and the linear response model of the infrared detector is used to do the NUC (Non-uniformity correction) process, when the infrared detector array is used as a Digital TDI. In order to eliminate the impact of the blind pixel, the concept of surplus pixel method is introduced in, through the method, the SNR (signal to noise ratio) can be improved and the spatial and temporal resolution will not be changed. Finally we use a MWIR (Medium Ware Infrared) detector to do the experiment and the result proves the effectiveness of the method.
Proc. SPIE. 8923, Micro/Nano Materials, Devices, and Systems
KEYWORDS: Signal to noise ratio, CMOS sensors, Aerospace engineering, Imaging systems, Sensors, Linear filtering, Spectral resolution, Integrating spheres, Modulation transfer functions, Analog electronics
With the requirements of high time resolution, high spatial and high spectral resolution development in geostationary orbit, photodetector pixel size has gradually become the bottleneck of the space exploration technology. Shanghai Institute of Technical Physics of Chinese Academy of Science has made a new breakthrough in CMOS image sensor area. The scale of its new CMOS image sensor achieves 2.5K×2.5K, and then use 24 detectors to achieve a detector whose scale is 150 million. The detector has been successfully imaging on the ground. In the application process, presents a systematic test and measurement methods to deal with the time noise, dark current, fixed pattern noise, MTF and other parameters of the detector. The test results are below. The MTF of the detector is 0.565 which is measured at 57.21/mm Nyquist frequency. The number of saturated electrons reaches 8.9×104. The total number of transient noise electrons is smaller than 16. The signal to noise ratio is 58.02dB. Through comprehensive analysis and measurement, it shows that the overall performance of the 2.5K×2.5K detector among the same types of products is in the leading position currently.
Geosynchronous satellite has obvious limitations for the weight and the scale of payloads, and large aperture optical system is not permitted. The optical diffraction limit of small aperture optical system has an adverse impact on the resolution of the acquired images. Therefore, how to get high resolution images using super-resolution technique with the acquired low resolution images becomes a popular problem investigated by researchers. Here, we present a novel scheme to acquire low resolution images and process them to achieve a high resolution image. Firstly, to acquire low resolution images, we adopt a special arrangement pattern of four CCD staggered arrays on the focal plane in the remote sensing satellite framework .These four CCD linear arrays are parallelized with a 0.25√2 pixel shift along the CCD direction and a 1.25 pixel shift along the scanning direction. The rotation angle between the two directions is 45 degree. The tilting sampling mode and the special arrangement pattern allow the sensor to acquire images with a smaller sampling interval which can give the resolution a greater enhancement. Secondly, to reconstruct a high resolution image of pretty good quality with a magnification factor 4, we propose a novel algorithm based on the iterative-interpolation super resolution algorithm (IISR) and the new edge-directed interpolation algorithm (NEDI). The new algorithm makes a critical improvement to NEDI and introduces it into the multi-frame interpolation in IISR. The algorithm can preserve the edges well and requires a relatively small number of low-resolution images to achieve better reconstruction accuracy .In the last part of the paper, we carry out a simulation experiment, and use MSE as the quality measure. The results demonstrate that our new scheme substantially improves the image resolution with both better quantitative quality and visual quality compared with some previous normal methods.