Adil RACHEK
Department of Industrial Processes Engineering, Ecole Nationale Supérieure des Mines de Rabat Rabat, Morocco
Colette PADET
Groupe de Recherche en Sciences Pour l’Ingénieur/ Thermomécanique UFR Sciences Exactes et Naturelles Reims, France
Download PDFIn this paper, we propose a contour extraction method where an index of non-linearity is introduced to select the best window size for the analysis. The method is based on one dimension set of data; so, we apply the treatment in four directions: the horizontal one, the vertical one and the two diagonals directions. The intersections of these new set of data with the original image can give the objects contours. But some noisy points can appear in the resulting image; so, there is a necessity to eliminate these points. In order to resolve this problem, an improvement of the previous method is proposed. The resulting image is swept line by line and the neighbor of every pixel is explored to detect the number of surrounding points and their locations to decide if this point can be a part of an object contour or not. In the case of an isolated point, it is considered as an noisy one and must be eliminated. The method is applied to a real photo image of non-isothermal jets, obtained by laser tomography. The good results obtained by this method confirm its effectiveness to eliminate a lot of noisy points and to give a better drawing of the contour. Moreover, we compare our results with many other existing methods, such as the Sobel, Laplacian and Canny detectors. Here again, an improvement of the extraction and the quality of the contours is observed.
Keywords: contour extraction; noise reduction; image processing; visualization; non-isothermal jet;
Disclaimer: All papers published in IJRST will be indexed on Google Search Engine as per their policy.
