Research On Refrigeration-type Infrared Thermal Image Processing Technology

Infrared thermal imaging system has the advantages of strong anti-interference ability and high concealment,which makes it has a broad application prospect in the fields of warship investigation,medical diagnosis and industrial monitoring.However,with the development of science and technology,infrared images with high resolution and high sensitivity is favored by many fields.Although the refrigerated large area array infrared detector is the key to improving the infrared image sensitivity and resolution,it is easily affected by the semiconductor process and the uneven doping,leading to a large errors in output pixels,which seriously affect the infrared thermal image the quality of.Therefore,the blind pixel compensation,non-uniformity correction and pseudo-color enhancement are required for the refrigeration type infrared thermal image.First of all,the paper builds a refrigeration type infrared thermal imaging system based on FPGA+DSP+MCU architecture,in which the FPGA mainly collects and transfers the images,the DSP mainly completes the image processing,the MCU ensures that the working temperature of the infrared focal plane array is always at 77K-80K.Then,the paper researches the four aspects of blind pixel compensation,non-uniformity correction,pseudo-color enhancement and infrared image temperature measurement.For blind compensation,the paper analyzes the domain replacement algorithm,and concludes that it cannot compensate for blind pixel blocks,Therefore,a blind weight compensation algorithm based on weighted adaptive window is proposed.For non-uniformity correction,the paper analyzes the advantages and disadvantages of traditional two-point non-uniformity correction and neural network non-uniformity correction,and innovatively proposes an adaptive two-point non-uniformity correction algorithm.For pseudo-color enhancement,the paper analyzes the traditional fixed-threshold pseudo color enhancement algorithm,and concludes that this algorithm can not improve the infrared image's layering and visual effects.Therefore,the paper proposes an adaptive threshold segmentation pseudo-color enhancement algorithm.For infrared image temperature measurement,the paper analyzes ambient temperature,measurement time and measurement distance on temperature measurement,and innovatively presents a high-precision infrared image temperature measurement algorithm based on drift compensation.In the end,the paper based on TMS320DM6437 also studies the front-end and back-end of video processing,and innovatively designs the ping-pong cycling storage structure,realized the pipeline working mode of this system.,and analyzes loading method and optimization scheme on the infrared image processing algorithm.On the basis of the system platform and theoretical algorithm research,the paper collects and processes infrared images,combing with MATLAB simulation,which shows the algorithm proposed in this paper reduces the system's remaining blind pixel rate to 8.8×10^-5,and the residual non-uniformity is lower than 5%,the temperature error was reduced from 8.37%to 1.03%.Additionally,the pseudo-color algorithm proposed in this paper improves the visual effect and quality of infrared images.