low-light image process
Low-light image real-time processing system design document
Zhu Meng Meng, Zhang Xing again, Super Wang
With the advancement of science and technology and the development of society, people's activities at night are more and more diverse, forcing a large proportion of night time activities . This makes the information reflected in low illumination, low light environment and even nighttime more and more important , and the corresponding requirements for acquiring image information at night are constantly increasing.
In recent years, FPG A has developed rapidly, and its high flexibility, strong parallel processing capability, and pipeline processing have greatly improved the speed of image processing. It is widely used in the field of image processing because of its small size , low power consumption , and short development cycle. In line with the demand of low-light night vision performance and the advantages of FPGA in today's society , we designed a real-time processing system for low-light video images based on Zynq-7000 SoC , combined with excellent image enhancement methods and FPGA processing capabilities, using the open source framework of PYNQ . Real-time enhancement of low-light images can be easily achieved . FPGA hardwareacceleration FPGA parallel processing capabilities and saves computing time.
Real-time video image processing system platform based PYNQ-Z2 Xilinx Z YNQ -7000 FPGA development platform. The PYNQ-Z2 platform enables high-performance embedded applications with parallel hardware execution, high frame rate video processing, hardware acceleration algorithms, real-time signal processing, high bandwidth IO , and low latency control.
Image information is captured by the camera. In real-time processing, we display the enhanced grayscale video image due to the frame rate . For color images, we first extract the RGBvalues of each frame image separately, and then perform the histogram equalization after 8 frames of integration processing, and synthesize the enhanced RGB images into enhanced color pictures. The output image is displayed on the display.
ZYNQ use of juypter end real-time processing of video screen image using the Python programming.
Histogram equalization is performed on the R , G , and B values of all the pixels of each frame image to avoid color space conversion, realizing real-time enhancement of color video images. And the enhancement effect on color images is obvious.
The method of inter-frame integration is used to remove noise. The denoising effect is obvious, and the noise generated during data acquisition and transmission can be effectively removed.
Added face recognition in low light conditions.
System composition and working principle block diagram
Figure 1 hardware block diagram
Figure 2 system overall block diagram
Source Code Github Link