Remote Sensing Image Restoration Based On Spatial Optics Remote Sensor MTF Compensation

The paper takes the visible light panchromatic wave band remote sensing imageas the object of study, takes the domestic and foreign correlation domain researchresults as the foundation, has studied the remote sensing image modulation transferfunction (MTF) modelling and the restoration algorithm. The purpose is to enhancespatial optics remote sensor system MTF as far as possible, and improve the remotesensing images quality, makes recovery images more meet people's eye visualcharacteristics, more highlight features target information, and solute the excessivezoom of noise problem in recovery process. And lay the foundation for subsequentprocessing for remote sensing image registration and object extraction.The remote sensing image restoration quality mainly is decided by theestablishment degenerated model accuracy and the restoration algorithm validity.Through the analysis of different methods to get space optics remote sensor MTF, andchooses the laboratory measurement method to establish system MTF model, thenaccording to the restoration algorithm performance, the order of complexity and so on,proposed different restoration algorithms.In order to obtain the precise remote sensing image degeneration model, whensatellite condition normal condition, laboratory measurement remote sensor MTF with in-orbit movement the basic maintenance is consistent, but the optical system centerfield of view and edge field of view correspondence MTF has the obvious difference,thus proposed uses the laboratory measurement method, obtain remote sensor ownvarious part to synthesize MTF precisely, and establish the entire field of view MTFmodel, then carries on the varying degree separately to the different field of viewremote sensing image the MTF compensation, causes the remote sensor entire field ofview image restoration result to be more precise.In order to the more prominent remote sensing image edge characteristic,sharpens the target identification ability, first carries on the atmospheric radiationadjustment and the denoising pretreatment to the visible light panchromatic waveband remote sensing image. Then analyzed the frequency domain restraint to besmallest two rides the filter method as well as passes the filter operator high based onthe smooth measure the selection method, proposed can the better prominentrestoration image edge characteristic restraint filter remote sensing image restorationalgorithm. And proposed the algorithm applies in remote sensing image entire field ofview restoration processing.In order to maintain the restoration image edge detail information at the sametime, well the noise elimination, according to the remote sensing image in the edgedetail and the noise different characteristic, as well as the fringing field and the flatsite different characteristic, proposes based on the aeolotropic operator auto-adaptedregularization remote sensing image restoration algorithm.The experimental result indicates that carries on the result using the laboratorymeasurement method establishment remote sensor MTF model which the imagerestoration obtains to be more precise, in the restoration image does not have theripple phenomenon. In guarantee signal-to-noise ratio certain situation, enhancesremote sensor system MTF large scale, improved the remote sensing image overallquality. Based on restrains the filter the remote sensing image restoration algorithm tobe able well the prominent restoration image edge characteristic. Based on eachopposite sex regularization operator auto-adapted remote sensing image restoration algorithm could during maintenance image edge, suppress the noise well, improvedthe restoration image quality.