Research On High Dynamic Range Imaging Technology For Spaceborne Area Array CMOS Camera

Improving the imaging quality of remote sensing cameras is one of the most important tasks in the field of aerospace remote sensing detection.High Dynamic Range(HDR)imaging technology,as an important image processing technology,can effectively improve the imaging quality of aerospace remote sensing cameras,its research value is huge.However,the dynamic range of the current area array CMOS remote sensing camera is much smaller than the dynamic range of nature,which makes the remote sensing camera cannot capture the actual brightness change of the real scene well;Affected by the attitude stability of the spaceborne platform,the image pixels cannot be well aligned;In addition,there is relative motion in satellite-to-ground imaging.In order to prevent the smearing phenomenon of remote sensing camera imaging,the integration time is set very short,and the target area cannot be well exposed,resulting in dim image imaging and loss of details.To address these issues,this study was primarily conducted as follows:(1)An image registration method of area array CMOS camera based on combination feature and gray scale was proposed to solve the problem that the target before and after the image frame was offset due to the influence of the attitude stability of the satellite platform in the high-speed continuous shooting process of area array CMOS camera under high dynamic scene.This registration method combines the feature registration algorithm with the gray level registration algorithm,extracts the image features through the feature extraction algorithm,and combines a fast matching strategy to carry out the rough image registration.The rough registration results were taken as the input data of the gray registration algorithm,and the image pyramid was combined with Open MP block parallel acceleration to calculate the gray level correlation number,and the image registration was carried out layer by layer quickly,and finally an accurate image registration was obtained.The registration method can not only solve the registration problem caused by large-scale rotation and translation of remote sensing images,but also realize high-precision registration of large-scale remote sensing images.Through experimental verification and analysis,the image registration method of area array CMOS camera based on combined features and gray level proposed in this paper is better than feature registration algorithm and gray level registration algorithm.It has the characteristics of high efficiency,high precision and high robustness,and contains richer and easier to identify information.It provides a new solution to the problem of low registration accuracy of large-scale and high-resolution remote sensing images,at the same time,it provides strong support conditions for high dynamic range imaging of spaceborne area array CMOS camera.(2)A multi-exposure image fusion method of area array CMOS camera based on virtual digital time delay integration(TDI)technology is proposed to solve the high dynamic range imaging problem of area array CMOS remote sensing camera.This method consists of image registration and digital TDI technology.Firstly,eight frames of original low illumination images taken continuously by the area array CMOS camera at high speed are registered to select the region of interest(ROI),and then the ROI region is generated by multi exposure sequence imitating the digital TDI technology.This method can effectively solve the problem of limited exposure time of area array CMOS camera,and there is no need to reform the hardware of the camera.In order to generate the input data required for image fusion,an exposure image synthesis strategy is proposed.By setting three exposure levels,three exposure images collected and generated by virtual digital TDI technology can be obtained.Finally,in order to solve the problem of image blur caused by virtual digital TDI technology,the image restoration method is used to deblur the image.From a global perspective,by improving the multi-exposure fusion weight evaluation factor,an HDR remote sensing image with rich details can be generated.Through experimental verification and analysis,the multi exposure image fusion method of area array CMOS camera based on Virtual Digital TDI technology proposed in this paper is better than the comparison algorithm,effectively solves the problem of the limitation of dynamic range of area array CMOS camera,and provides a new solution for remote sensing satellite to obtain high-quality images.(3)In order to realize the spaceborne integrated HDR imaging technology,the method in this paper is implemented and deployed in the Texas Instruments digital signal processor(DSP)TMS320C6678using a multi-core parallel strategy,and applied to the H863 project of the CX Special Zone of the Military Science and Technology Commission "XX Information Fusion and Extraction"(Phase II)".By comparing the operating efficiency of single-core and multi-core using remote sensing image data of different scales,the effectiveness of the multi-core parallel strategy of this method is proved.