| The infrared imaging technique transforms the infrared radiation distributionsinto visual images. The image details are obtained passively by detecting the thermalradiation diversities on each superficial part of the scene. The infrared radiation canpenetrate smoke, distinguish the camouflages and be detected remotely and round theclock.Thermal imaging systems involve multiple subjects and have been widely appliedto military and civilian fields. The recently developed uncooled type based onmicrobolometer has more application prospects, as its unique advantages with highportability. Some developed countries have researched for a long time on thetechniques of the infrared imaging. They monopolize the key technologies, such asthe manufacture of core infrared detector, the image processing algorithms, the designof high-performance circuit system.This project is based on an imported microbolometer with response peakwavelength of8-14μm. First, its output signal format was explored by using anoscilloscope and developing an circuit board. Then, the pre-and post-processingalgorithms, including nonuniformity correction, grayscale transformationenhancement and interpolation enlargement, were researched and verified throughexperiments. Further, a high-performance far infrared imaging system was developed.(1) Regarding the output signal characteristics exploration:According to the serial output signals, a portable imaging system board wasdeveloped. The driving for the focal plane arrays was integrated into a CPLD, so thebit-level adjustment could be done. After a large quantity of experiments, the exactinfrared video driving was obtained. Then its noise characteristic was analysed.(2) Regarding the research on pre-processing and post-processing algorithms:A novel real-time nonuniformity correction solution for infrared focal planearrays based on S-curve model was proposed. To compensate the detector’s drift, acorrection algorithm of considering the drift of infrared focal plane arrays wasproposed. Experimental results prove both algorithms have high correction precisionin a large dynamic range. In allusion to the real scene images captured by the systemboard, a dynamic inter-frame nonuniformity noise filtering algorithm was alsopresented. The algorithm is especially suitable to the high-noise-ratio infrared images.A grayscale transformation algorithm based on histogram threshold adaptiveextraction was proposed. It can compress the dynamic grayscale of14bits original infrared image to8bits image to be displayed, suppress the large dynamic rangewhich includes little pixels occupying large quantity of grayscales and preserve thedetails. It is adaptive with low computation and applicable in engineering.By studying typical interpolation algorithms, an adaptive edge-preservinginterpolation algorithm for infrared image was improved. Experiments show that it iseffective for twice enlargement, but is time-consuming. Therefore, a fast imageinterpolation algorithm based on gradient was proposed, which is suitable toembedded system application.(3) Regarding the research on infrared imaging circuit system:Beforehand, an infrared imaging system with small size and low power wasdeveloped. Via system solutions comparison, a high-performance far infrared imagingsystem was developed. The hardware processor is TMS320C6416T, which has highspeed, large on chip RAM and double sets of EMIF. High speed FIFO mechanism offchip was adopted to buffer images. The software used interrupt function and the datatransfer employed EDMA techniques. These measures greatly decreased the datatransfer time and ensured processing time. The system was also designedmeticulously to minimize it. Experimental results show that, the system can capture,process and display the infrared image sequence in real time. The image is able tooutput standard PAL video at a frame rate of25Hz with a resolution of640×480. |
PDF Full Text Request