Multi-scale Image Transition Region Extraction and Segmentation Based on Directional Data Fitting Information

摘要

This paper presents multi-scale image transition region extraction and segmentation algorithm based on directional data fitting information. Using direction-data fitting information measure, image transition region will be gained by modulating the wavelet scale. An optimal segmentation threshold will be obtained from the transition region histogram. This method has gone beyond the limitation that traditional methods of image transition region extraction are sensitive to noises and depend on clip limits L_(low) and L_(high) Analyses and experiments show that the proposed method is characterized by speed, high robustness and good anti-noise performance.