Method of Group Coding of Infrared Images in Spectral-Wave Space

The main features of the construction of intellectualized services for the exchange of infrared images (IRI) are considered. Namely, the possibility of obtaining informative meta-information. Such information may include: class and state of objects of interest, identification of image fragments by their level of informativeness. The use of meta-information as a result of intellectualization is possible in a wide range of applied tasks. In particular, those that are solved using unmanned on-board systems. Some of these tasks are: monitoring of dynamic objects of interest, autonomous guidance of on-board complexes. In the process of performing the specified tasks by on-board complexes, decision-making is possible in the following modes: manual, autonomous, automated. However, there are conditions in which a hybrid variant of decision-making is possible. In this mode, at some stages of the decision, the on-board complex can be in an autonomous state, and at others in an automated state. Among the conditions that have an impact on the choice of mode: the presence of information conflict, crisis situations, the use of swarm technologies. Taking into account the modes of use and tasks performed by on-board complexes, the requirements for the completeness of information are increasing (increasing the number of frames, limiting distortions, increasing the number of pixels for describing objects). The consequence of their compliance will be an increase in the information load on information and communication systems. Therefore, a contradiction appears between the requirements for quality characteristics: speed of information delivery, integrity of IR frames. Therefore, a scientific and applied task is relevant, which concerns the improvement of the quality characteristics of the provision of intellectualized information services based on the sources of IRI in applied tasks using on-board complexes. As a result of the analysis of modern technologies for solving the given problem, such as PNG and JPEG 2000, the following shortcomings were revealed: high computational complexity, introduction of significant distortions to the semantics, low efficiency in segments with a high number of heterogeneous objects. Therefore, the objective of the article's research is justified: the development of a method of group coding of data in the spectral-wave space. The article describes the stages of method development, which begin with the decomposition of the IRI into a hierarchical structure of segments and minisegments. This allows to localize the homogeneous areas of the image. The resulting segments are further transformed into the spectral domain and spectral-group coding is applied, which allows to reduce the bit volume. An experimental evaluation of the developed method was carried out on the basis of the Open Turbulent Image Set (OTIS), which includes PNG images with different levels of informativeness. It was possible to reduce the bit volume of images by an average of 37%. In addition, a comparative analysis of the coding of 16-bit images with the existing coding method, which is used in the modern JPEG 2000 format, was conducted. Where it was shown that the developed method has an advantage in the compression ratio by 25%.