Long-wave Infrared Camera System Based On DM3730

Infrared thermol imaging technology has the advantage of passsive imaging, night imaging and easy concealment, so it is widely used in military and civil fields. One kind of thermal imaging device is embeded infrared camera. It is with small size and weight, easy to deploy in the wild, so is popular in the market.Conventional infrared cameras adopt general DSP or FPGA as the main processor. these two solutions have the common problem of poor user-friendly interface due to simple software design with no operating system or basic input-outputsystem(BIOS).Given to problems above, the project chooses a mobile video processor DM3730 and ports the Linux operating system. TI's DM3730 integrates the image signal processor(ISP) for video capture, an ARM Cortex-A8 core, a C64X+ DSP core, and the display sub-system(DSS) for LCD and TV display. High integration reduces peripheral chips and power control module makes it power efficient.Based on this platform, first the original data was extracted from an imported infrared camera, and the data format was converted to 12 bits parallel sync format byan CPLD in order to match DM3730 ISP interface. Then the Linux operating system was ported to the camera and the V4l2 video driver was researched so as to design our own camera driver and ISP driver. Thirdly, the ARM and DSP interaction mechanism, Codec Engine,was researched and DSP was called to process the image data with appropriate algorithm. Finally, the DSS display function was researched and 0.5x to 8x video display was realized.The hardware design needsfewer chips while the dual core architecture makes it high performance. The drivers make the camera more flexible and the Linux operating system makes it more suitable for application development. The adoption of Codec Engine can integrate different algorithms to satisfy different needs.Experiments have proved that the camera can capture 12 bits graycale image, call the DSP for image processing and use DSS for scale display. Besides, images after nonuniformity correction have showed great quality improvement, which proves the algorithm is effective.