A dual threshold calculating method for fiber´s edge extraction

In the fiber cross-sectional image analysis system, correctly detecting fiber´s edge is critical for fiber geometric feature extraction and further fiber identification. In this paper, a dual threshold calculating method is proposed to obtain accurate and continuous fiber edge, as well as to control the image noise. After eliminating the non-uniform illumination in the fiber image, the low threshold is calculated based on the statistics histogram of the standard deviation of pixels´ gray level in the image. And then the high threshold is computed by bisection algorithm combined with the low threshold. After tracing and denoising algorithm, the fiber wall that can correctly describe fiber´s contour is detected. The pseudo edges and dual edges detected by traditional algorithms are eliminated, and image noise is well reduced. The experimental result shows that the proposed dual threshold selection algorithm is simple, convenient and efficient to extract fiber edge from fiber cross-sectional image.