Comparison of Fringe Thinning Algorithms

An implementation of the three thinning algorithms to interferometric  fringes reveals the following. The automatic thinning approach is unambiguous and repeatable, in the sense that the thinned fringes do not depend on the starting point specified by the user. The curve fitting and paintbrush approaches are sensitive to the user input. On the other hand, they do not require elaborate code preparation time and can be adopted if thinning is only sporadically required.

Figure 4.24: Thinned images, paintbrush drawing method for fringe thinning.

A preliminary examination of the thinned images (Fig.4.20,4.22, and 4.25) shown that the three methods produce qualitatively similar results. An independent assessment of these techniques is taken up in the present section. The first criterion adopted is the comparison of the width-averaged temperature profile for each projection. Under nominally steady conditions, i.e., after sufficient time is allowed to elapse, the width-averaged temperature profile in the fluid plotted as a function of the vertical coordinate can be shown to be independent of the projection angle. This is because the total energy transferred across any horizontal  plane in the fluid layers is a constant. These profiles are shown in figure 4.26, for all the three algorithms studied. The S-shaped curve, characteristic of buoyancy–driven convection can be seen in this figure. Within experimental limits, a close match between the zero and 90 degree profiles is seen to be realized when the first fringing thinning approach is employed. This match confirms the accuracy of fringe thinning, temperature allocationto fringes, and the absence of interpolation errorss while transferring the data to a rectangular grid. Figure 4.26 also shown that the fringe lost during thinning near the top wall does not degrade energy balance. The agreement in case of the curve-fitting method is partial, while it is unsatisfactory with the paintbrush method.

Figure 4.25: Superimposed thinned images (paintbrush method) with original, images