Research On Distributed Imaging System Platform Design And Image Compression Algorithm

A distributed imaging system is a novel optoelectronic imaging regime that utilizes multiple small aperture optical detectors,which are easy to manufacture,to effectively function as a large aperture optical detector.The small aperture detectors can be constructed using different dimensional detectors,such as polarization and spectral detectors,to achieve collaborative detection and imaging in a division-of-labor manner.The development is faced with two main challenges,namely,the system control processing platform needs to possess parallel processing capabilities and meet the requirements of rapid iteration and refactoring.The massive data challenge arising from the dual-multichannel configuration requires targeted simplification of information to ensure system flexibility and functionality,without compromising imaging and detection performance.Therefore,platform design and information compression are currently the key and challenging aspects of building distributed imaging systems.From a single dimensional multi-aperture imaging platform based on distributed imaging system,a system parallel control processing platform is designed around programmable device FPGA and 4 sets of CMOS sensing modules,and the corresponding design framework and indicators are presented.Combining the framework indicators,the system working principle and specific implementation plan are determined,and the core devices of the platform such as the main control,sensors,and physical layer chips are selected based on the working principle.The platform structure design is implemented to achieve parallel processing of information and rapid reconstruction.The parallel data flow logic of the multi-aperture platform was studied with a focus on the core devices.The AXI4 bus protocol;the RGMII timing protocol;and the Mac data packet structure were theoretically analyzed.Based on the analysis,the digital logic design of each submodule in the framework was described using the Verilog HDL language,and the logic design of the key modules was verified using RTL behavioral-level and timing simulation,enabling the design of the digital flow logic for multi-aperture parallel acquisition,storage,and transmission on the platform.Furthermore,the communication between the upper computer and the platform was established.A biologically inspired selective attention imaging(SOI)mechanism for compressing regions of interest in images was proposed based on the analysis of the physiological basis of visual perception,in order to achieve information simplification in the system.First,the NLM algorithm was used to remove image noise.Then,the ITTI bottom-up visual model was introduced to extract the SOI regions in the image.Finally,Max Shift coding was used to implement region-based compression.To determine the Mask region,the outer rectangle of the SOI connected domain was obtained using a greedy algorithm,and small SOI regions were filtered out by a window to avoid unnecessary calculations.The redundant information in the detection image was removed while preserving the quality of the SOI regions.The design platform was subjected to board-level experimentation,and the results indicate that the platform is capable of synchronously and in real-time outputting four sets of stable1280×720p×60fps CMOS sensor images.Furthermore,the platform is capable of reliably transmitting sensor detection data via Gigabit Ethernet.Simultaneously,field experiments were designed,where the platform was moved to the field to capture real-life images of two groups A and B and processed them using the compression algorithm mechanism.The experimental results showed that both the objective and subjective evaluation metrics of the region-based compression algorithm were superior to those of traditional global compression,achieving a maximum compression ratio of 9.3:1.After processing the data from the 4 CMOS sensors,the average size of the system was only 10.75% of the original data size.The research has shown that a control processing platform designed around programmable devices meets the basic requirements of a distributed imaging system.Moreover,the platform eliminates redundancies and reduces the size of image data while maintaining image quality and detection performance.As a result,the system’s transmission bandwidth is saved,indirectly enhancing its carrying capacity.This addresses the problem raised in the first paragraph.The research results have certain application value in the field of optoelectronic imaging detection and provide guidance for the advancement of new optoelectronic imaging systems.