Study And Design Of Low-power And Low-noise Uncooled Infrared Core Assembly

Infrared imaging is a visual imaging technology which is on the purpose of the detection of heat radiation. Compared with visible light, infrared imaging technology can be adapted to a more complex application environment, and it is an important technical support in the field of defense and security. In terms of both military and civilian, there are great prospects and market potential. Uncooled infrared imaging technology developed earlier in Western countries, China started late, and there is a some gap with foreign technology.The key of the uncooled infrared imaging system is uncooled infrared core assembly. Contents of this paper is the uncooled infrared core components expanded, mainly from the hardware system and image algorithm. The paper work focus on low power consumption and low noise design.On the hardware system, the core assembly of this paper uses FPGA as data processing platform, signal processing schemes and power solutions is the design points. We design efficient switch power solutions with the total efficiency more than 90% to achieve lower power consumption and select the linear power chips with noise lower than 50 μV to achieve low noise power supply. In signal processing terms, we use bias module with reference added operational amplifier and thermostat module to ensure low noise environment. We ensure the security of the system through the power protection modules ans power-on and power-down protection module. We reduce the DDR rate and replace analog PID feedback with digital PID in thermostat module to reduce power consumption.On the image algorithms, this paper detailed studys the infrared image processing algorithm commonly used. Infrared image processing including nonuniformity correction, image enhancement, and image de-noising. In the non-uniformity correction, the paper introduces the traditional fixed-point correction and scenario-based non-uniformity correction. We analyzes the shortcomings of the time drift of correction coefficient in fixed correction coefficient, and presented a non-uniformity correction method for correcting image column stripe. In terms of image enhancement, we compare the traditional linear stretch and histogram stretch, and presented an improved algorithm based on histogram equalization. This algorithm integrated the advantages of linear stretch and histogram stretch, can fully stretch the image, and will not lead to over-stretch the image, and can adapt to different scenarios.Thereafter, we study the details enhancement algorithm based on the guide filter; In the image denoising, this paper compares the traditional averaging few frames, and proposes an concise algorithm based on RC filter circuit model, with Better adjustability.In summary, this paper focus on hardware system and image algorithm,we design the more rational and efficient hardware system and image algorithm to achieve low-power low-noise high-performance target, the test results show that our work achieve a breakthrough in reducing power consumption and noise, with the power consumption less than 1.5W, RMS noise of 0.9m V, noise equivalent temperature difference less than 70 m K.