ISTEC 2011 International Science and Technology Conference - Istanbul University - Turkey December 7-9 , 2011 Image Registration for Pressure Sensitive Paint based on Improved Feature Detection Criteria Heider K. Ali Department of Systems and Computer Eng, Carleton University,1125 Colonel By Drive, Ottawa, ON,K1S 5B6, Canada, hkali@connect.carleton.ca Abstract: Image registration of a Pressure Sensitive Paint (PSP) is an important part of studying a model in a wind tunnel. To increase the accuracy of registration process, the image pairs of both wind-off and wind-on states have to be aligned precisely to study the effects of wind pressure on the model effectively. The features must be extracted from the images, then to match these features, and finally applying the required transformation to complete the registration process. Harris corner detection algorithm used to extract the corners from both wind-off and wind-on image, edges detected by Canny-detection algorithm, and Hough- transform applied to detect lines of the images, matching each neighborhood pair of corners to get the most matched corners. The experiments done on many sets of images showed the effectiveness of this detection algorithm and gave the possibility to modify and enhance the process. Keywords: Pressure sensitive paint, Harris-corner detection, Hough-transform, image registration, wind tunnel 1. Introduction The image registration has been carried out a lot of research work, a variety of methods or techniques were put forward. Registration based on corner detection as feature extracted has the merits of the rotation invariant and almost immunity to illumination conditions [1]. Image registration is a crucial part in studying the wind pressure effects using the pressure sensitive paint (PSP) technology. The pressure sensitive paint (PSP) is relatively a new measurement technique for global surface pressure measurements in aerodynamic systems. The fundamental operating principle of PSP is oxygen quenching of luminescence from the paint. Light intensity emitted by the paint is measured by a photodetector [2]. This technique is used in the wind tunnel applications to get the required data for image registration of wind-off and wind-on images. One image is taken at a known “wind-off” reference pressure while the other is taken at the test condition” wind-on”, to factor out the effects different factors affecting the wing. In order to well-recognize these effects like model deformation, nonuniform illumination, paint thickness, and wind pressure effects, the markers are used according to certain criteria as input tools which located on the model. The pressure data can be extracted only as a ratio between the “wind-off” and a “wind-on” image in case of exactly aligned images so as the pixel-by-pixel comparison of the registered images shows the wind pressure effects on the model through the studying of intensity changes on the model images. As the wind-on/wind-off image pairs have been captured they required processing. The brightness (intensity) of the PSP coating is recorded by a camera as gray scale values. The ratio of the wind-off image intensity (I 0 ) which is in static case (no air flow) is made with wind-on image (I) under air pressure effect, so, the variation in the ratio-image intensity can be related to pressure change, as described by the Stern-Volmer equation[2] I 0 p ---- = A+B ---- (1) I p 0 Where p and p 0 are the wind-on and wind-off pressures, respectively, and A and B are coefficients specific to PSP chemistry. The image rationing requires that the wind-off and wind- on images are exactly in the same position. But the studies showed that these two images did not still aligned correctly because of the model deformation and displacement due to wind pressure. This requires repositioning of the two images prior to measure ratios of equation (1) 2. System Arrangement The model under test is fixed in the wind tunnel which is regularly illuminated by the illumination sources to enable the camera to catching up both wind-off and wind on images in exactly the same illumination conditions. Also, it is important to keep camera and illumination sources out of wind pressure effects to guarantee that the factors of equation (1) are only the factors of concern. Artificial markers are fixed in the model to be good features can be relied on them in the next processing steps. The following figure shows the arrangement of model test [3]: