A skeleton extracting algorithm for dorsal hand vein pattern

Extracting vein skeleton with little distortion from the vein image is very important for improving the identification rate. In the paper, an algorithm for segmenting the dorsal hand vein image and extracting the vein skeleton is presented. Firstly, after gray and size normalizing, Gaussian lowpass filter and median filter are used to eliminate the speck noise and the horizontal strip scanning noise respectively. Then, an improved NiBlack algorithm segments the vein pattern and an area thresholding algorithm removes the noise blocks from the vein pattern. Subsequently, opening, closing and median filter are used to smooth the vein boundary. After that, the vein pattern is thinned by Kejun Wang´s improved conditional thinning. Lastly, a pruning algorithm is presented to trim the spurs and the vein pattern is skeletonized. Experiment shows the algorithm acquires more real skeleton.