Application of image processing in estimation of area of heat affected zone

Abstract

A filtering scheme based on morphology, thresholding and edge detection is investigated and implemented on metallograhic image to analyze the heat affected zone (HAZ). To establish the potential of the new filtering scheme, experiments were conducted on the image of a weldment. Quality of image of a weldment from the point of view of analysis of HAZ depends mainly on polishing during surface preparation and concentration of etchant to differentiate HAZ from other area of the weldment. Signal to noise ratio of the image acquisition system also influence quality of an image. To see effectiveness of the filtering scheme, study is made on a poor quality image of weldment.Experimental results revealed that the new filtering scheme is very much effective in automatic evaluation of area of HAZ. For the poor quality image, the difference in measurement of HAZ area was 9% when compared with the manual evaluation. Manual evaluation of HAZ area contributes larger fraction of the error as it is based on approximation of irregular shape of actual HAZ area.