Time-frequency distribution and image processing: an application to the combustion field
Abstract
The aim of this study is to provide a better understanding of the turbulent flow field inside an internal combustion engine (Floch, 1990) by measuring the "integral length scale", with a known optical method and essentially with new theoretical tools in both image processing and signal processing.
A photographic plate located at the viewing screen of a transparent combustion room, records density gradient which depends of the medium encountered by the light (Schlieren method). Such a method displays density gradients in terms of light intensity. Image processing methods are used to point out turbulence structures. In fact, a cross-correlation function is investigated to shape a new image with pseudo periodicity which contains the size information. Eventually, time-frequency distribution is introduced to determine integral length scales. The Wigner-Ville distribution is chosen to allow the analysis of "ghost terms", also named "interference fringes" (Flandrin and Escudié, 1985), giving a reliable estimation of structures size, with a convenient accuracy.
In addition, comparisons with laser Doppler velocimetry (L.D.V.) measurements are given (Fraser and Bracco, 1989). Furthermore, the dispersion of engine cycles can be estimated by taking into account both the spark gap adjustment of the sparking plug and the average of integral length scales.