Abramov S. Methods of signal and image processing based on myriad estimation in remote sensing systems

The object is images and signals formed by systems of remote sensing (RS) complexes. The goal is to improve image and signal processing methods effectiveness and robustness in information-measuring, radar and optical RS systems in conditions of presence of mixed noise (the mixture of additive and impulse or multiplicative and impulse) in cases of restricted a priori information about their characteristics due to application of myriad estimation and filtering methods. The methods come from general probability theory and mathematical statistics, numerical simulation, digital non-adaptive and adaptive nonlinear filtering, estimation theory and numerical solution of nonlinear equations theory. The main novelty is the applying of myriad estimation to shift parameter determination for essentially nonsymmetrical data sample distributions in RS data processing systems. The blind noise variance evaluation methods for signal and image processing based on myriad estimation, the generalized adaptive myriad 1-D filtering method and principles of its parameters selection and the two-component locally-adaptive image filter based on myriad estimation for processing images corrupted by impulse noise with large probability are designed and tested. The proposed methods are implemented as algorithms and software and are used in several branch organizations. The sphere of application is RS data processing.