A Self-adaptive EM-HMRF Image Fusion Algorithm Based on Non-homogeneous Class and Direction
Aim at classification-oriented feature-level fusion of remote sensing image, MRF is firstly introduced in this paper to build prior probability models for multiple object classes, and EM-HMRF scheme is introduced into image fusion by taking advantage of the equivalence relation between EM-HMRF and fuzzy classification algorithms. Secondly, focusing on how to select model parameter b with self-adaptive character, a new selecting method is dirived. Then a self-adaptive EM-HMRF fusion algorithm is proposed basing non-homogeneous class and direction, which includes 2 centric and distributed-based fusion schemes. Theory analysis and experiment results show that our proposed algorithms with 2 fusion schemes can not only improve the classification accuracy but also enhance the ability to anti-interference, and they have the different advantages in various fusion systems for different applications, and thus improve the effectiveness of classification basing feature-level fusion of remote sensing image.
Guiqing HE Yan ZHU
Northwestern Polytechnical University, Xi’an
国际会议
2011亚太信号与信息处理协会年度峰会(APSIPAASC 2011)
西安
英文
1-10
2011-10-18（万方平台首次上网日期，不代表论文的发表时间）