A fuzzy pixel decomposition approach in the classification of remotely sensed data | |
| Author | Pan, Rui |
| Call Number | AIT Thesis no. CS-91-25 |
| Subject(s) | Land use--Remote sensing Land use--Remote sensing |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | In conventional remote sensing classification, most classification results are represented in a one-pixel-one-class mapping in which a pixel is an indivisible minimum unit no matter whether it is class mixture pixel or not, which greatly and always limits the improvement of classification and boundary accuracies. This work describes a new classification approach of using fuzzy pixel decomposition algorithm to deal with the class mixture problem, where each pixel is represented as a fuzzy pixel and divided into non-fuzzy sub-pixels. The algorithm consists of two major steps: Pixel membership grade analysis, and Pixel decomposition, in which fuzzy set theory is applied for representing land cover membership grade for each pixel, neural network and statistical approaches are used for analyzing pixel class membership grade. Afterwards, each pixel is decomposed into sub-pixels, and the membership grades of neighboring pixels are used for identifying the-position of each sub-pixel. An idea of distance in neural network is presented, and a method of selecting minimum training data is also described in this work. Evaluations were made among conventional, neural network and pixel decomposition classifiers. The results of classifying TM and MOS-1 images in the areas of Thailand were presented and their accuracies were analyzed which showed the improvement in both classification and boundary accuracies as well as thematic map. The studies show that class mixture in a pixel can be identified, and the fuzzy pixel decomposition ·algorithm is a useful tool in improving final classification output of remotely sensed data. |
| Year | 1991 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Other Field of Studies (No Department) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Hosomura, Tsukasa |
| Examination Committee(s) | Vilas Wuwongse ;Zhao, Ming |
| Scholarship Donor(s) | Government of Japan |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1991 |