In this paper a non-linear parabolic PDE-based diffusion model is applied to enhance thermal images. The proposed smoothing technique is a non-linear time-variant system, in which the diffusivity field depends on the local characteristics of the image. The non-linear diffusion is controlled by human knowledge based linguistic rules. Some examples of intensity dependent diffusion, of edge preserving filters and their aggregates are presented. The image details in the interested regions are preserved while in other regions the variations of pixel intensities are attenuated. With the proposed fuzzy-technique a strongly non-linear diffusivity function can be defined in a user-friendly way.
This paper presents an IR-imaging technique to visualize a turbulent steam flow. Instantaneous, time-averaged, Gaussian-smoothed and non-linear filtered IR-images are compared. Wavelet threshold is applied to separate the coherent part of the instantaneous intensity field from the incoherent part. The averaged image has been smoothed with a non-linear filter that preserves the energy content of the thermal image better than the Gaussian low-pass filtering. High frequency wavelet components of the processed IR-images highlight the differences of the applied filtering techniques. Reconstruction the flow pattern from these components shows the directional orientation of the instantaneous and averaged flows.
In this paper a new, phenomenological, temperature dependent hysteresis model of the vapor-liquid first-order phase transitions is presented in order to improve the existing equation-of-state type diffuse interface methods implicated in homogeneous two phase flow models. The proposed hysteresis approach describes not only equilibrium but also non-equilibrium phase transitions. A saturation temperature dependent upper limit of the allowable supersaturation is proposed, this limit ensures that unstable conditions could be always avoided. To overcome the flowing system caused memory handling problems of classical Preisach-type hysteresis models, a partial differential equation based hysteresis operator is proposed to describe the statistics of a model fluid consists of bistable, constant-mass clusters of molecules.