Research On Theory And Application Of Image Translation Between SAR And Optical Image

With the rapid development of space remote sensing technology,more and more remote sensing satellites were launched,it shows alarge difference not only in the principle of imaging but also in the effect of display while using different kinds of sensors.This difference would produce barriers when sharing information,exchanging information and converting information between the two kindsof images.So,we propose a new concept which is called image translation,which aims to interconversion between each other.In the train of this research thought,the main research works of this paper are described as below.(1)On the basis of previous studies,the concept of image translation is defined as following:"for the purpose of image expression,based on the image understanding,the image translation is a process to change the form of display to anther one".We can get three key points of image translation from upper definition,which are called comprehension,transformation,and expression of the images.In this article,further explanations were written to describe the difference and connection among image translation,image understanding,image transformation and image fusion.(2)Explaining the aims and effects of image translation:by using the method of image translation,we can exchange the way of display between optical and synthetic aperture radar(SAR,for short)image,which means making up the deficiency of SAR image without color information naturally.We can also amend or identify the areas of optical image which are obscured by clouds.Translating SAR image to optical image will also improve the identifiability and visual effect compared with the raw images,which will play a positive role on emergency disaster monitoring,image mapping,map revision and information sharing.(3)Constructing the framework of image translation:giving the detailed process of how to translate.According to the definition of image translation,a flow line process about image understanding,transformation and expression was build up,that is to say,we firstly make classification of the SAR image,so we get the land using classification maps,and then use the method of image fusion to combine the optical image's colorful components with SAR image in the same regions.In this section,we also roughly discussed the methods of directly translation between SAR image and optical image.(4)Summary the methods of SAR image classification.There are two kinds of SAR data,namely single polarization SAR and polarimetric SAR(PolSAR,for short)image.Because the single polarization SAR only record the backscattering coefficient index(?o)of radar objects,people still use the optical-based methods to classify this kind SAR image.Concerning the factors of pooler texture information and the influence of speckle noise,the classification accuracy towards these images is lower than optical ones obviously,whatever we use the supervised classification method or unsupervised ones.PolSAR image,otherwise,not only record the ground target's backscattering coefficient,but also get the information of microwave's polarization and phase,all those information are saved in particular arrays which is called Sinclair's matrixes.By using the method of H/?+Wishart,we can derive a highly classification precision towards PolSAR image.(5)There are two main patterns in current remote sensing system,optical imaging pattern and microwave ones.So we put the emphasis on the exchange between SAR image and multispectral image,respectively take the single polarization TerraSAR and UAVSAR in full polarization as examples,do some simulated translation tests by using the upper mentioned process.(6)Taking advantage of the theory of image translation,we make a useful application to rediscover the areas obscured by clouds in multispectral images.In the mountains or on the sea,the optical images are always polluted by clouds in large scale,so far as to almost completely covered by clouds.Usually,people will avoid using those optical images covered by clouds,but in the specific time and space conditions we must take such cloud-contaminated data.If we can translate SAR to optical image,we can reappear these invisible regions.