New methods of tomography and thermal images filtering

Two new thermal imaging methods, based on the original methods of heating and algorithms for solving the inverse problem of heat conduction in the test object. In addition, proposed three methods of thermal imaging, which are based on the account of the dynamics of thermographic images and allows you to filter artifacts related to the heterogeneity of the surface emissivity, significantly increase the number of virtual pixels and image quality, allows maximum use of a priori information in the solution of ill-posed inverse problems of reconstruction. Examples of application of these methods to the problem of the cooling channels in turbine blades control. The projection dynamic thermal tomography (PDTT) method is applicable for control of objects with irregular internal structure. It allows reliably detecting the foreign inclusion-type defects and controlling with high accuracy the geometric parameters of the internal structure of objects. This method is designed for using modern thermal imager with a standard set of parameters and does not require expensive additional equipment. Preliminary numerical analysis of capabilities of the proposed method for specific objects and the parameters of expected defects is possible, indicating the requirements to the parameters of thermal imager and heating. The PDTT method is superior in terms of information value to all alternative DTT methods and can also replace a number of more expensive control procedures using other methods (ultrasonic, x-ray, SHF). The generally accepted parameters of evaluating the quality of cooling turbine blades are missing. A wide range of sizes and theirs cooling mechanisms significantly complicate the use of such parameters as absolute values. We propose to use a universal relative parameter. It corresponds to the virtual heat transfer coefficient at a predetermined depth of the metal layer. It illustrates the possible use of this parameter for various methods of thermal control. It is shown that an integrated approach to the construction of the control system and the integration of different modes of cooling can significantly reduce the uncertainty of the quality of the cooling parameter. He is a local and could eventually be used to evaluate the residual life of the blade. In case of random nature of heating the measured thermal emission flow depends not only on the real surface temperature, but also on its emissivity factor. Unlike the known DTT methods, the filtration of artifacts requires taking account of the dynamics of the whole process. Such processing is performed in each pixel of the thermal image and the dynamics of measurements in adjacent pixels is also taken into account. The accuracy of the filtration is determined both by the number of adjacent pixels taken into account and the completeness of the obtained measurement information. The problems encountered when observing objects, whose dimensions do not exceed several pixels of the matrix of recorders. It is shown that the motion of the object relative to the matrix, and the use of multiple consecutive frames can significantly improve the fidelity of the shape and true size of the object. This measurement information can be represented as integrals along the trajectories on a three dimensional lattice of space-time pixels. The reconstructions of а heterogeneity function of pixels through the use of the controlled motion of the test object and methods of computer tomography. The possibility of determining the relative velocity of the object in each of the frames using the "blind" method. The algorithm of reconstruction of images with large non-uniform increment, as in passive, and when they are managed relative motion. An algorithmic approach to solving inverse problems of reconstruction of the structure of the scattering medium can significantly improve its informativeness in cases where the lack of prior information and a high noise level does not allow to use standard approaches. Thus, unlike statistical processing methods, constructing a set of hypotheses is not required. The unknown parameters of the processing procedure can be obtained using the "blind method" in the reconstruction process. This automatically eliminates the possible redundancy of model. The choice of language to describe the array of input data and the search algorithm of the model optimal parameters is determined not by the features of the device and the object of observation, but by the expected view of the final presentation of the results. The described technique is universal and can be used for solving different tasks of reconstruction as a whole, and in any intermediate signal processing.