The suspending liquid consists of various concentrations of particles which are either0.22μm or0.494μm in diameter and were mixed into filtered,distilled water,and serve as the scattering medium forstudying the polarization property of radiation side-scattered by the particles.The experiment results showthat the de-polarization of radiation side-scattered by the particles is very sensitive to the variation of theparticles’diameters: the horizontal polarization degree of the side-scattered radiation of the particles withsmaller diameters is much higher than that with larger diameters,but the vertical polarization degree of theside-scattered radiation of smaller particles is much lower than that of larger particles.

Lead zirconate titanate thin films were successfully prepared on 5-inch TiO~~x/Pt/Ti/SiO~~2/Si substrates by RF-Magnetron Sputtering method.The experimental results show that the orientation of PZT thin films can be changed from(111) to(100) by precisely controlling the substrate temperature.The (111)-oriented films with thickness of 500nm have the remanent polarization of 20μC/cm~2,dielectric constant of 370,dielectric loss of 1.5%,coercive field of 130kV/cm and pyroelectric coefficient of 1.1×10~(-8)C/cm~2K.The films can be used for fabricating uncooled infrared focal plane detector arrays.

The design and fabrication of one-layer bimaterial micro-cantilever array made of silicon nitride(SiNx) and gold(Au) without Si substrate for uncooled optical readout infrared(IR) imaging system were presented.One-layer structure simplifies the fabrication process.IR thermal images can be obtained when the cantilever array is placed in the atmosphere.The experimental results with the micro-cantilever array of 140×98 elements and 12-bit CCD indicate that the objects at the temperature of higher than 120℃ can be detected and the noise-equivalent temperature difference(NETD) of the device is ~7K.The experimental results are well accordant with the thermomechanical design of microcantilever array.

With a well-developed infrared radiation transfer model KCARTA,the effects of spectral resolution,spectral response function shape,spectral calibration accuracy,propagation path and surface emissivity on the detected brightness temperature for hyperspectral sounding were simulated.The results from forward simulation show: 1) improvement of spectral resolution will increase detectable brightness temperature 2) any uncertainties from internal and external factors will lead up to much larger brightness temperature difference for hyperspectral sensor than for low spectral sensor.Results can be taken as the referrence in designing future hyperspectral IR sensors and in developing hyperspectral retrieval algorithm.

A 3mm band integrated oscillator is researched,which is consisted of a IMPATT diode,a microstrip resonator,and a fin-line transition.With the theory analysis and research,China's first 3mm band integrated oscillator using the W-band CW IMPATT diode,was made.It delivers a output power of 7mW at 94.78GHz.

On the basis of developing 3mm Dicke radiometer,a set of 3mm radiometric imaging system controlled by PC-MCU mode was presented.To solve the contradiction of sensitivity and spatial resolution for the motion platform imaging system,the software integrator was designed and the cascade system integration formula was derived.Furthermore a morphological composite filtering was used to improve the imaging resolution.The experimental results show that 3mm radiometric images spatial resolution is superior to 8mm images' and the processed image can wipe off noise effectively.In addition,it is verified that the software integrator can take place of the hard integrator in some degree.The radiometric parameters are given below: the center frequency 94GHz,the sensitivity 0.2K,the linearity 0.999,the system bandwidth 2GHz,the cassegrain antenna 3dB beam width 0.5°.

An experiment research of large scale neural networks classification was carried out by applying GIS technique synthetically,which took the Spot/vegetation NDVI image in Inner Mongolia as basic information source.A highresolution image database of many years time series was build by integrating GIS software with remote sensing visual interpretation and selecting "pure" sample data in the range of the whole municipality.Land cover's BP artificial neural networks classification of the area was accomplished by combining DTM data and grid climate data with the database.It turns out that with the support of GIS technique,BP artificial neural networks classification of spot/vegetation NDVI image at large scale can reach a higher classification precision.

A new object feature,i.e.,tree descriptor,was used to represent a skeleton tree.And the maximal isomorphic subtree was obtained by searching the longest same substring in the tree descriptors,then a novel method of objects similarity measuring based on matching skeleton tree was presented.During the matching process,a new model was used to measure the similarity between two skeleton branches,furthermore,the shape feature and topology feature were associated perfectly.The problems of cyclic skeleton and noise on the skeleton were discussed.The similarity computing with our method is invariant under rotation,scaling,translation,and the time complexity of the algorithm is o(n~3),n is the number of the skeleton branches in the skeleton tree,and the experiments on a variety of objects get satisfying results.

Compared with traditional remote sensing technologies, hyperspectral remote sensing has a great number of bands from visible to thermal infrared range,it presents a combination of image and spectrum.Normally,ground information of historical relics is not distinct.Contrasted to traditional remote sensing,hyperspectral remote sensing has more advantages in identifying the unclear ground information of historical relics and quantitative research.But at present,hyperspectral remote sensing still has not got effective utilization in archaeology according to our literature review.Taking Qinshuhuang's Mausoleum as an example,this stuey systematically explains these innovative application research results of hyperspectral remote sensing technology in archaeology.These results can be as a successful trial research achievement of hyperspectral remote sensing in archaeology and other new breakthroughs in Archaeology of Qinshuhuang's Mausoleum area by remote sensing such as the discoveries of high thermal anomalies of th

In the process of reflecting,scattering and transmiting electromagnetic wave,earth targets can produce polarized characteristics which related to the nature themselves,therefore,detecting the objects' polarization information becomes a new method in remote sensing.Since polarized reflectance always goes with bidirectional reflectance,the polarized three-dimensional spatial distribution of the targets can be obtained by a polarimeter during detecting its bidirectional reflectance.From the point of view of multi-angle polarized remote sensing mechanism,the quantitative relation between polarized BRDF and BRDF was studied.And the experiment testified the equality of the reflectance of bidirectional reflect,polarized reflect of 45 degree and polarized reflect of mean value in the corresponding azimuth angle,zenith angle,and detection angle and detection channels in 2π space.

Magneto-optical(MO) storage is an attractive technology for next generation storage.In order to increase the capacity of storage in a single disk,one of the approaches is applying the multi-valued recording and multi-wavelength readout technique,i.e.,so called 3-dimensional(3-D) MO storage.Lasers of different wavelengths are used to read or write data on different recording layers.The difference in the temperature profiles of this kind of MO disk caused by different lasers is expected to be large enough for improving the signal to noise ratio(SNR).In this study,optical matrix method and finite element method were applied to analyze the thermal and magneto-optical responses for a 3-D MO disk.The results show that the MO disk with GaP as interlayers has a potential in improving the thermal and optical performance at two blue wavelengths compared with the MO disk with SiN interlayers.

Background clutter plays a very important role in the area of human-in-the-loop target acquisition.To model the target acquisition capability of an electro-optic system operated by a human,one should take into account how the clutter in the scene affects the human performance.Based on the classic probability of edge(POE) metrics for quantifying background clutter,and the up-to-date research findings in human visual perception,a new method for modifying NVESD's classic human target acquisition performance was proposed.And the modified model was preliminarily validated.

By calcu lating the areas of the lateral aberration curves accord ing to Tchebycheff num erical integration m ethod,an im age-quality index based on these areas was set up in order to simu late the im age-quality m erit function that always is used in damped least square(DLS).And then,it is used in the adaptive optical autom atic design program as an optim iza-tion index,thus realize the composite of these two optim ization ideas.Th is composite can improve fish-eye lens’in itial structure rap id ly and procure the auto-design process of fish eye lens convergence qu ickly.Added to th is,the virtual reality techn ique,successive mu lti-link-c ircu lar“morb id ity”process m ethod is also put forward in th is paper.Here,virtual lens can conquer“ray-overflow,” break away from“morb id ity”and seek the simp lest optical structure;And virtual variab le stops and the“morb id ity”processm ethod can conquer the d ifficu lties during optim ization of IR fish-eye lens,such as slowly convergence，divergence,oscillation or no optimization solution,etc.and make convergence procedure quickly and stably.A example is given in the end.

The application of laser beam induced current(LBIC)for the technology detecting of Mercury-Cadmium-Telluride(HgCdTe)infrared two-color detector was presented.By LBIC test,it ws found that the area of n-type region derived from p-type HgCdTe having been implanted by Boron ion was larger than the area to be implanted,and the precise transversal distribution of this n region was also obtained.At the same time,the transversal distributions of plasma etching induced n-type doping HgCdTe from p-type material among the side wall of etched mesa was obtained,and the relation between the transversal width of this n-type region and the energy of etching plasma was also achieved.

A new scheme for the detection and classification of subpixel spectral signatures in remote sensing images was presented.By minimizing the energy function with two special constraints,the mixed pixels in multichannel remote sensing images can be decomposed more precisely.In this study,experiments on the proposed scheme and BP neural network with artificial and real-world data were performed.The experiments show that the proposed scheme can get more precise results and is obviously more robust than BP neural network.

The problem of noise reducing and resolution enhancement of SAR image was discussed.A unified regularization variation model of partial differential equation(PDE) was established,which has the noise reducing merit of PDE model and the resolution enhancement merit of regularization model.On the background of the image,the PDE model was used to reduce the noise of background,and at the targets region of the image,a background equation was used to sharp the targets firstly,then the regularization model was used to enhance the resolution of the targets.Hence,the new model can get better result on the whole image.The diffusivity based on the background of SAR was also constructed,and a forward and backward diffusion equation based on the amplitude of SAR image was obtained,which could reduce the noise of background and sharp the edges of targets.Experimental results show that the new model can enhance the strong scatter points and suppress the speckle of the image effectively.

Based on the statistics characteristics of contourlet coefficients,a new multi-scale image segmentation method(CHMTseg)combining Contourlet domain hidden Markov trees model with multiscale Bayesian approaches was presented.A novel weighted neighborhood model was given for preserving more inner-scale information in Contourlet domain.The pixel level segmentation based on Gauss mixture model and the multiscale fusion method based on the new contextual model were provided.In experiments,synthetic mosaic image,aerial image and SAR image were selected to evaluate the performance of the method,and the segmentation results were compared with wavelet domain HMTseg method.For synthetic mosaic texture image,miss classed probability was given as the evaluation of segmentation results.Experiment results show that the method not only has better performance in edges and anisotropy information detection but has lower missed classed probability,and it can achieve satisfied segmentation results for real images.

Because the traditional binary phase-only matched filter(BPOF) has a great loss of information and difficulty in recognition in hybrid optronics scene matching system,ternary amplitude-phase filter(TAPF) was proposed here to try to solve these problems.The computer simulation of infrared scene matching was carried out and the performances of TAPF,BPOF and ACMF were compared.Experiment results show that TAPF can improve SNR and PCE of the correlation pin.So it can greatly enhance the anti-noise performance of the system,and then improve the probability of scene matching effectively.