Image processing method and implementation for a laser automatic tracking extensometer

Abstract

A laser automatic tracking extensometer for material deformation measurement based on CCD is proposed. The image processing methods of the laser mark localization and the automatic tracking of the mark line on the specimen are studied for the extensometer. First, geometrical mean filter (GMF), harmonic mean filter (HMF) and inverse harmonic mean filter (IHMF) using for the image processing are compared in order to select a suitable mean filter for removing noises from the specimen images, and then the GMF is adopted for the de-noising of the images. Second, Sobel operator is introduced to detect the edges of the specimen images. At last, the specimen images are reduced to eliminate unwanted background information by pruning. Hough transformation of pretreated specimen images is also studied and linking images algorithm is proposed based on the image gray distribution and the connectivity principle. The laser mark localization and the automatic tracking of the mark line on the specimen are then implemented. The experimental results show that the linking image algorithm is prior to Hough transformation on both recognition effect and recognition efficiency.