Measurement Of High-temperature Objects Based On Dual Infrared Cameras And Accuracy Analysis

The size of high-temperature objects is measured in a non-contact way such asimage measurement. The visible imaging result will be greatly deviated by the lightdeflection in high-temperature field. And the infrared imaging mechanism is toconvert the object’s thermal radiation into electric signals, so there are someadvantages in dealing with the challenge compared with visible camera.It is the target of the research to measure the High temperature objects’ sizecombing with the advantage of thermal imaging vision technology. The research isfocused on these main problems about the high temperature objects’ infrared binocularmeasuring scheme. The main work is as follows:1. The thesis studies the present situation of the domestic and foreigndevelopment of the measurement of the high-temperature objects’ size, analyses thedevelopment and the application prospect, and summaries the Binocular stereo visiontechnology. The research direction and meaning are then determined.2. The changes in temperature field around the High temperature objects of900°C are analyzed in the method of finite element analysis. The refractive indexdistribution is calculated using the light propagation theory. The deviation of imagecaused by the refractive inde x change is analyzed in comparing with0.40μ m,0.76μ m,8.00μ m,14.00μ mtypical wavelengths combining with dispersion mode l.The expe riment shows the de viation between the High temperature objects’ image ofvisible imaging and infrared imaging, and the results show that the visible imagingdeviation is about4times of infrared imaging deviation when the High temperatureobject’s temperature is900℃. An infrared imaging model is established, and thegeometrical relationship transformation between the coordinate systems has beenanalyzed, then the influe nce of the system structure parameters on the visual accuracymeasurement is studied. Finally, we design the overall scheme of the infraredbinocular measuring system by selecting the hardware device according to the actualmeasurement condition, and we set up an experimental platform.3. The causes and features of infrared image heterogene ity phenomenon isanalyzed in the paper, and it discusses the global non-uniformity correction method,the weighted average filtering correction method and the BP neural networkcorrection method and we have improved the BP neural correction method, introd uce the eight neighborhood pixels correlation, redesign the hidden layer considering thenonlinear correlation factors. The experimental results show that infrared imageinhomogeneous de gree have decreased to4.986%with the improved BP neuralnetwork correction method, which can obtain better corrected results than otheralgorithms. Through analyzing the characteristics of infrared image, this paperdiscusses the platform histogram infrared reinforcement method, high-pass filteringinfrared image enhancement method and weighted average infrared imageenhancement method combining the infrared image quality evaluation inde x withdoing some experiments.4. We have designed the infrared visual ball targets especially according to theinfrared imaging mecha nism combining with the features of infrared image and putforward the new infrared binocular measurement system calibrating scheme throughhigh precision electronic three-dimensional translational platform. Through exploringthe comparison between the thermal image calibrating methods, the calibration of themeasurement system is completed using radial consistent constraint criterioncombining with the LM parameter optimization algorithm.5. Through studying the existing image edge extraction algorithm, combiningwith the advantages of neural network and fuzzy reasoning method towards theinfrared image characteristics, the left and right correlation gradient is introd uced andinfrared image edge extraction method based on eight pixel point fuzzy neuralnetwork is proposed. This method can make the edge single-pixel and improve theimmunity ability. Infrared binocular vision feature matching is analyzed in this pape r,and the size measurement is completed with Hough transformation using SURFoperator feature matching. The experimental results show that the system sizemeasurement relative error is about1%, and the system has good stability.