Research On The Framework Of The Infrared Imaging System And Key Technologies Of Image Processing

The infrared imaging system effectively expands the visible spectrum range of human eyes.In recent years,it has been widely used in various fields,such as military reconnaissance,security monitoring,forest fire-prevention,disease diagnosis,automatic driving,etc.However,due to the complexity of the infrared imaging system and the limitation of the manufacturing process of the infrared focal plane detector,the raw infrared images of the infrared detector usually suffer from severe non-uniformity,relatively concentrated gray levels,low contrast and dynamic range,thereby seriously reducing the visual quality and target discrimination of infrared images.Based on this,this article focuses on the infrared imaging system framework design,non-uniformity correction algorithm,contrast enhancement algorithm,and high dynamic range infrared imaging.Aiming to improve the performance of the infrared imaging system and the visual quality of infrared images,the main contents of this article include:1)Take one uncooled infrared focal plane detector of the ULIS from France as an example.Based on the analysis of the characteristics of the infrared detector driving signal and output video signal.The field-programmable gate array(FPGA)based hardware framework and the Qsys platform and Avalon-bus based software framework are designed.The studied hardware and software framework designs can effectively shorten the development cycle and reduce the costs of infrared imaging systems.2)In view of the poor ambient adaptability and the repeated calibration problem of the calibration-based non-uniformity correction algorithm,an adaptive shutter-less non-uniformity correction algorithm is proposed.The algorithm collects a series of raw infrared images of the detector at different ambient temperature conditions with a black body as the uniform radiation source.Then the optimal non-uniformity correction parameters are computed based on the selected raw infrared images and the actual image via the least square method.Thus the correction parameters are more stable to the ambient temperature.Experimental results show that the proposed algorithm performs well on non-uniformity correction under different ambient temperature conditions,and there is no non-uniformity residue.For the low convergence speed and the unsatisfactory correction result of the scene-based non-uniformity correction algorithm,a fixed pattern noise(FPN)estimation based non-uniformation correction algorithm is proposed.The algorithm assumes that the non-uniformity in the infrared imaging system is mainly determined by the FPN.Then based on the time-domain statistical characteristics of the adjacent pixel difference,the one order lag filter and mean filter are utilized to estimate the adjacent difference of FPN.Lastly,the non-uniformity correction parameters are computed based on the intra-frame recursion.Experimental results indicate that the proposed algorithm gets better performance on non-uniformity correction and faster convergence speed.3)In order to solve the problem of low contrast and poor visual quality of infrared images,the global and local histogram specification(GLHS)and neighborhood conditional histogram equalization(NCHE)based infrared image contrast enhancement algorithms are proposed.The GLHS based algorithm utilizes the 2D histogram and the histogram specification to compute the global and local mapping functions,which are then used to obtain the global and local contrast enhancement results.Lastly,the final optimized enhanced image is computed by solving an optimization equation based on the global and local enhanced images.The NCHE based algorithm replaces the clip-redistributed histogram of the contrast-limited adaptive histogram equalization(CLAHE)algorithm with the neighborhood conditional histogram.In order to avoid the over-enhancement of the homogeneous regions and the block artifacts between sub-blocks,the local mapping functions are adaptively updated based on the global mapping function.Lastly,global and local enhancement results are combined to get the final enhancement result.Experimental results show that the proposed algorithms outperform other algorithms on local contrast enhancement,avoiding over-enhancement,and block artifacts.4)A scheme for acquiring high dynamic range infrared images is proposed,that is firstly applying non-uniformity correction and contrast enhancement to the raw infrared images that captured at different integration time.Then the high dynamic range infrared image is acquired by fusing the enhanced images via the proposed multiple exposure fusion algorithm.Compared with the calibration-based methods,the fussy calibration process is not required to estimate the response function of the infrared camera.The proposed multiple exposure fusion algorithm is based on the multi-scale strategy,and the details of the fused pyramid and the weighted pyramid are compensated to reduce the loss of details during the pyramid fusion process.Thus,details of the original images are better preserved in the fused image.Experimental results indicate that the proposed algorithm outperforms other multi-exposure fusion algorithms in terms of the visual quality and the evaluation index MEF-SSIM.