A parallel thinning method based on image marking

A new parallel thinning method based on image marking is proposed. Firstly the distance from each pixel to the nearest outline is calculated, based on which a marking image is then created as the input to the thinning method. The experimental results demonstrate that via the comprehensive utilization of the marking information, the proposed method performs better in terms of robustness and ability in terms of remaining geometric specialities, an important basis for further shape recognition.