Using inverse filtering to increase the resolution IR images

The aim of the work is to increase the resolution of IR images obtained as a result of monitoring thermal objects. Methodology. It is known, an optical system сan not date a CCD point picture thermal object. Instead picture diffraction spots, have significant deterioration in quality thermal image which has a reduced effect with defraction. This is an urgent system task useful in the formative and processing signal. To achieve this goal, it is suggested to use the method of reverse filtration, which allows, knowing the function of point dispersion (FPD) of the optical system caused by the phenomenon of diffraction, to significantly reduce its effect on the quality of the resulting image. In the optics-computer system, the optical image is projected onto a CCD matrix, de FPD which is already presented in digital form. This method based on reverse filtration [Rabiner et al., 1978]. It is believed that blur is an irreversible operation and information is irretrievably lost, because every pixel turns into a spot and everything mixes. Results. In fact, all information is simply redistributed in accordance with the FPD and can be uniquely restored with some reservations. The proposed technique for using the reverse filter algorithm allows us to overcome the limitations imposed by the optical system. Scientific novelty. The authors propose to use the specially developed digital measures and the program of two-convolution convolution (convolution) of these images with FPD to determine the effect of the FPD values on the resolution of the monitoring system. Practical significance. The developed algorithm of reverse filtration (deconvolution) together with other methods (for example, sub-pixel processing) can be successfully used for processing IR images obtained as a result of remote monitoring of thermal objects. The deconvolution method allows overcoming the limitations on the resolution that are imposed by the optical system in the infrared range. This leads, in the absence of noise, to an accurate reproduction of the input image of the thermal object, regardless of the diameter of the world spot. The difference in the values of the FPD optics and the FPD model, which were used in the implementation of convolution and deconvolution programs, is of decisive importance. Especially important are the results of the operation of the method of reverse filtration under the conditions of noise on the thermal image and in the data transmission channel. The value of the signal-to-noise ratio at which the distortions are considered as insignificant is determined. At the same time, as the studies show, the magnitude of the photoluminescence spot of the objective is important. The question of application of the proposed method of reverse filtration in the case of uncertainty of lens data, which was used in thermal monitoring, is often encountered in the practice of processing thermal imagery available to the user. All the results obtained are verified on imitation models, which is the additional novelty and practical significance of the results obtained.