Gaussian Mixture Model Segmentation Algorithm for Remote Sensing Image
- Resource Type
- Conference
- Authors
- Hou, Yimin; Sun, Xiaoli; Lun, Xiangmin; Lan, Jianjun
- Source
- 2010 International Conference on Machine Vision and Human-machine Interface Machine Vision and Human-Machine Interface (MVHI), 2010 International Conference on. :275-278 Apr, 2010
- Subject
- Computing and Processing
Communication, Networking and Broadcast Technologies
Bioengineering
General Topics for Engineers
Robotics and Control Systems
Image segmentation
Remote sensing
Pixel
Weather forecasting
Classification tree analysis
Machine vision
Man machine systems
Sun
Automation
Simulated annealing
Markov Random Field
Guassian Mixture Model
Remote Sensing Image
Maximum A Posteriori
- Language
The paper proposed a novel method for remote sensing image segmentation based on mixture model. The remote sensing image data would be considered as Gaussian mixture model. The image segmentation result was corresponding to the image label field which was a Markov Random Field(MRF). So, the image segmentation procedure was transformed to a Maximum A Posteriori(MAP) problem by Beyesian theorem. The intensity difference and the spatial distance between the two pixels in the same clique were employed in the potential function. The Iterative Conditional Model(ICM) is employed to solve MAP. In the experiments, the method is compared with the traditional MRF segmentation method using ICM and simulate annealing(SA). The experiments proved that this algorithm was more efficient than the traditional MRF one.