Research And Implementation Of Image Quality Assessment For Spatial Images

The image quality of spatial image has been widely concerned in the space measurement and control index.It is also a key step in the study of space target detection and tracking.With the rapid development of sensor technologies,digital signal processing technologies,and communication technologies,the quality assessment of massive spatial images by humans has become more and more difficult.Therefore,the technology of using computers to evaluate the quality of spatial images has emerged.Image quality assessment(IQA)technology has always been an important research direction in the field of digital image processing.However,the existing image quality assessment algorithm has insufficient quality resolution capability since the imaging target of the spatial image is a dim point target that is easily confused with noise.In addition,due to the effects of many other factors such as the spatial particle radiation,diffuse scattering of light,high-speed operation of the observation platform and vibration,the background of the image with low temperature and deep space,etc,the rare noise types and uneven noise distribution in the image are often emerged.Hence,the description capability of the extracted image features is inadequate.Therefore,the quality evaluation of the space surveillance images still faces many challenges.To overcome the above-mentioned problems in the spatial image quality assessment research,we propose a spatial image quality assessment method based on information fusion.Based on the design of spatial image quality index system(SIQI),some spatial image quality assessment methods are studied.Due to the lack of an image quality database containing spatial images,this paper constructed a spatial image quality database.The algorithms are verified on this database and a spatial image quality assessment system is established.The main works of this paper are as follows:(1)We build a Spatial Image Quality Database(SIQA).The existing image quality databases do not include spatial images.We collect real spatial images containing various complicated conditions in the past three years,such as:different monitoring times,different shooting positions,and different seasons of spatial images.To achieve a standardized quality data labeling process and ensure the sustainability of research,we embed it into the space surveillance image quality assessment system.(2)IQA methods based on Spatial image Quality Index System(SIQI)are studied.By studying the degraded characteristics of image quality in space surveillance,a set of quality index system(SIQI)suitable for spatial surveillance image quality assessment are proposed for the first time.Based on this SIQI,we use many methods such as the support vector regression,back propagation(BP)neural network,and linear regression algorithm to assess the image quality.(3)We propose the IQA method for spatial image based on information fusion.This method uses different feature extraction operators to perform feature extraction on spatial images,and evaluate the image quality based the image-based fusion methods.Six different fusion methods which are performed on the feature level and the model level,further improve the performance of the quality assessment models.The proposed methods achieve the purpose of spatial image quality assessment.In addition,we design and implement a space image quality assessment system based on C#and MATLAB for the application of this research.