Design Of High Dynamic Imaging System Based On Dual-exposure Camera

The contrast between light and dark in the natural environment is as high as10¹⁴:1,and the contrast between light and dark perceived by human eyes at the same time is about10⁵:1.The ratio of light and dark contrast corresponds to the concept of dynamic range in the imaging field.The dynamic range of images obtained by traditional image sensors generally does not exceed10³:1,and a large amount of image details are lost in overexposed and underexposed areas,which cannot meet people’s growing demand for image quality.High Dynamic Range（HDR）imaging technology can significantly improves the dynamic range of visual perception,significantly increasing the image details of overexposed and underexposed areas,which is widely used in surveillance security,TV broadcasting,satellite remote sensing,backlight shooting,etc.It is difficult for traditional HDR imaging equipment to take high dynamic range,real-time and anti-ghosting effects into account at the same time:Some imaging equipment with a large dynamic range can only take pictures,which require computers to process pictures,lacking real-time performance;HDR images synthesized by some HDR video imaging equipment have insufficient visual dynamic range or poor ghost removal effect.In view of the above shortcomings of traditional HDR imaging equipment,this thesis adopts the architecture of FPGA+GPU heterogeneous platform to design an HDR imaging system based on a dual-exposure camera.The system can alternately acquires two Low Dynamic Range（LDR）images with different brightness,synthesizing HDR video with more detailed images in overexposed and underexposed areas,and achieves clear imaging of moving targets by removing ghost images.This system can be used in many fields such as monitoring security,video capture of target recognition equipment,HDR video shooting in backlight scenes,etc,and has important engineering practical significance.The main work of this thesis is as follows:（1）A dual-exposure gradient domain HDR imaging algorithm is designed.By solving the poisson equation based on the image gradient structure tensor,the algorithm synthesizes HDR images with significantly increased image details in overexposed and underexposed areas.At the same time,the algorithm combines the information entropy method to detect ghosts,and removes ghosts caused by moving objects by using the consistency measure of gradient direction changes between two LDR images.After applying this algorithm,compared with the traditional HDR video imaging equipment,the dynamic range increase of the output image of this system is improved to a certain extent,and the clear imaging of the moving targets is realized by removing the ghost images.（2）The hardware of the dual-exposure HDR video imaging system is designed.Through the analysis of the development background and existing problems of HDR imaging equipment,the overall system architecture is designed and indicators such as frame rate,exposure time,are proposed.Based on the characteristics of dual-exposure,a camera module that can alternately capture LDR images of different brightness is designed.In view of the shortcomings of traditional single-system architecture imaging equipment to achieve HDR fusion and de-ghosting algorithms,there are certain difficulties,limitations,and high power consumption.Heterogeneous computing platform based on FPGA+GPU can efficiently utilizes the hardware resources of two processors and has the advantages of high performance and low power consumption.（3）System testing and analysis.Based on the implementation of the algorithm designed in this thesis on the hardware platform,the key data communication nodes of the system are tested,and the system data transmission process is accurate and stable.The HDR imaging effect and ghosting effect has been evaluated many times.The results show that with a certain resolution,the image information richness is increased by about 25%,which is higher than the 23%increase in image information of two other similar systems.The ghost removal effect of the system is remarkable,and the imaging of moving targets is clear.The dual-exposure HDR imaging system designed in this thesis can stably generates HDR video,and the visual dynamic range of the output images are significantly increased.Under the premise of a certain resolution,the richness of image details is significantly improved.At the same time,the system achieves clear imaging of moving targets by removing ghost images,which has important engineering practical significance.