The Research Of Image Terminal Quality Assessment For Spaceborne Optical Remote Sensor

Spaceborne optical remote sensor is a precision scientific instrument carried onsatellites or shuttles, which can obtain the optical attributes and related informationabout the target. It has the following advantages due to working on various satelliteplatform: working in vacuum environment, far away from atmosphere environment,which will benefit to clear image; the imaging task is not subject to regionalrestriction, which can achieve to observe the interested target in global or outer space;the working time is flexible, which can work in any of the scheduled time for remotesensing; the data obtained by the remote sensor are widely used in many fields such asmilitary reconnaissance, environmental resources assessment, natural disasterassessment, and weather forecast.Imagery quality of spaceborne optical remote sensor is a key test item, whetherremote sensor can produce high quality remote sensing imagery is directly linked tothe success or failure of the remote sensing tasks. So the imaging quality assessmentof spaceborne remote sensor has a vital significance for a long time. The work,assessing the spaceborne optical remote sensing imagery quality, not only canquantitatively characterize the imaging quality of spaceborne remote sensor, but alsocan provide assistance for ground assembly and promote the technology of groundassembly. In the meantime, through the work of monitoring the remote sensor imaging quality on orbit for the long time, the factors of impact the remote sensorimaging quality and lifetime can be analyzed, it also can help promoting andimproving the design technology of remote sensor.The methods of imaging quality assessment were researched in this dissertation.The spaceborne optical remote sensor we researched is working in the visible lightwaveband with the function of detailed observation. There are many traditionalmethods for remote sensor imaging quality assessment, such as resolution test patternmethod, using rectangular spatial frequency target plates to measure modulationtransfer function, and so on. But these methods face some new challenges with theoptical remote sensor research technology development and the improvement of thetechnical requirements. Nowadays the optical remote sensors working in visible lightwaveband with the function of detailed observation have some new trends: the opticaldiameter is increasing, the focal length is getting longer, the ground resolution isgetting higher, and the volume and weight is growing larger. The new trends make thetraditional methods of imaging quality assessment limit by the test site, testenvironment and test equipment. For example, the temperature and vibration level ofthe test site meeting the imaging quality assessment for large diameter and long focallength optical remote sensor is difficult to control. And the collimator devices, thehigh spatial frequency rectangular target and the integrating sphere apparatus aredifficult to be manufactured.In this dissertation, the novel idea, applying the spaceborne optical remote sensorimage terminal quality to assess the imaging performance of remote sensor, wasproposed. It can realize to assess spaceborne optical remote sensor imagingperformance through the quality of image terminal. The imagery quality cancharacterize not only the imaging performance of the spaceborne remote sensor, butalso the imaging technology. Therefore, the imagery quality we assessed synthesizesthe whole factors which degenerate the imagery in imaging chain, such as obtainingoptical information, compensation of image motion, photoelectric conversion, digitalsignal processing, storage, compression and so on. Assessment the imaging quality ofremote sensor by this method can realize monitoring the imaging quality of remote sensor on orbit, and it is not constrained by the test equipment, test site and testenvironment.In order to assess the imagery quality of spaceborne optical remote sensor, theresearch methods of imagery quality assessment taken from two different perspectiveswere proposed.One is using the optical modulation transfer function as imagery quality index,the MTF of remote imagery at Nyquist frequency was estimated by the improved BPneural network. The input parameters of the neural network are extracted from theimagery obtained by the remote sensor. This paper discusses in detail as follows: thesimulation method of remote sensing imagery with controlled MTF, the extractionmethod of parameters which can characterize the MTF level of remote sensingimagery, the model designing and parameters determination method of the neuralnetwork, and the accuracy of imagery quality assessed by neural network.The other one is using the distance of parameters model between ideal naturalimagery and degraded imagery as quality metrics. The method is based on the theoryof statistical natural scenes, which is a no reference image quality assessment method.By the theory of statistical natural scenes, it is believed that ideal natural scenesimagery have some unchanged statistical model parameters, and the parameters willbe changed along with the image degrading by noise, compression, blur and so on.The follows were discussed in detail: the select method of image characteristicparameters, the design method of statistical model, and how to measure the distanceof the statistical model parameters, in the end the correlation between this method andsubjective image quality assessment method was researched.The experimental results show that the ideas of two remote sensing image qualityassessment methods are feasible, the assessment results have a reasonable accuracy,and the results can better reflect the subjective feelings of the image quality. Thisthesis laid the foundation for the further research about remote sensor imagingperformance characterized by the remote image quality.