Design And Implementation Of Infrared Image Processing And Transferring System Based On ZYNQ

In relations of on-rail foreign matter detection,in order to acquire information of scene under insufficient lighting,we designed an on-chip infrared image processing and transferring system is of high speed,low consumption and high integration.Because the surface temperature of the objects in the scene is relatively close bring about infrared image collected by sensor has low contrast that is solved as mainly work in this paper.The specific work is as follows:To improve contrast of infrared,firstly,in this paper,we explored why infrared image is of low contrast from origin of data,and contrast of infrared image is improved by intelligent selecting digital quantization range optimization combined with the actual scenes.Secondly,in order to further improve low contrast of infrared image,multi-scale Retinex(MSR)is utilized for enhancing in this paper.Given that complexity of design and implementation of MSR algorithm embedded on FPGA,a MSR algorithm of improving calculation process(MSR-RMLC)is proposed.The nova algorithm can obtain outcomes only using minimum scale kernel as alike as multi-scale computation,which effectively reduces the complexity of design and resource occupancy rate.Finally,a pseudo-color is studied to improve display effects of infrared image in this paper.Interference signal had been existed led to the data calculation in the image processing module is not synchronized to output irreparable error image.Hence a nova image data acquisition module,which can provide Synchronous Reset(SRST)signal for successive module,and the signal is set high before beginning of any frame image,which can clear all erroneous data from buffer to avert outputs of error data why robustness of the system is improved.Secondly,MSR-RMLC processing module is designed on RTL(Register Transfer Level),and the module can efficiently enhance low contrast and brightness of image,and the latency is 6 HSYNC and 20pixel-clock periods.Additionally,we designed a multi-color IP core with AXI4-Lite,including multiple display mode.Finally,HDMI and Ethernet two video output modules are designed and implemented.The testing shows the system can improve low contrast infrared image efficiently.Additionally,hardware utilization rate less than 25% meeting requirement of scalability.Total on-chip power consumption is 2.249 W,1.601 W on PS and0.648 W on PL,reducing consumption efficiently.Timing report shows the system can run stably.The design in this paper can meet requirements of on-rail foreign detection relations.