Research On Remote Image Quality Assessment

Remote sensing image quality evaluation plays an important role in remote sensingapplication. Remote sensing image can be used to interpretate,find target and achievemilitary intelligence. In surveying and mapping applications,remote image are appliedto match, stereo position for establish digital geographic information systems. Remotesensing image quality directly affects the accuracy and precision of the surveying andmapping. And image quality reflects the performance of the sensor. In addition, imagequality assessment can evaluate the performance of the image processing algorithms.This article main research content is as follows:(1) Remote sensing image quality influence factors analysis. Factors influenced theremote image quality will be classified noise, aliasing, block effect and ringing byanalysising remote sense chain.(2) Blur, noise, aliasing, block and ringing characterization methods are proposed.Resolution, signal-to-noise ratio, signal-to-aliase ratio, block and ringing used to be thecharacterization parameters respectively to represent blur, noise, aliasing, block effect,and ringing. And experiment verify the parameters can characterize the image factorseffectly.(3) Characterization parameters for experimental analysis.Analysis results showthat the correlation between the parameters, Principal component analysis is to be usedto decorrelate that the correlation between the parameters. And gain an image qualitycomprehensive evaluation method.(4) Experiments validate the image quality evaluation method.The experiment isdivided into three parts. The first part, select a typical scenario for each scenario to joindifferent degrees of blur, noise, aliasing, block effect,and ringing to gain the simulationimages respectivly. Use the comprehensive quality evaluation method to evalutesimulation image quality. The second part, select typical scenario, for each add differentdegree of all kinds of qualitative factors to get the simulation images, assess imagequality of each simulation. The third part, selecting typical scenario, add blur andnoiset to each scenario,and use direct inverse filter and wiener filtering for simulationimage restoration.and evalute the restored image quality.Compare the two kinds ofalgorithm performance with the image quality.