Home 
Knowledge Based Systems Analysis of Convergent evidence in an evidential reasoning knowledge-based classification
Knowledge Based Systems Analysis of Convergent evidence in an evidential reasoning knowledge-based classification | Analysis of Convergent evidence in an evidential reasoning knowledge-based classification |
|
|
Knowledge-based systems (KBS) are based on evidential reasoning for land-cover mapping based on remotely-sensed images are becoming widely held. In recent years, KBS utilizing Dempster-Shafer Theory of Evidence (D-S ToE) were found most successful in wide range of remote sensing applications. One important feature of the D-S ToE is that it provides a measure for the evidential support (belief) accumulated for each object class at each pixel. Despite their major rule in classification decisions, analysis of the cumulative belief values (CBVs) did not get much if any attention in the literature. In this study relationships between CBVs of the different classes and their corresponding classification accuracy/reliability were investigated. The CBVs were found good indicators showing level of classification complexity in both pixel and class scales. In addition to that, levels of two class properties could be analyzed through the CBVs distribution of each class: heterogeneity and uniqueness. Low correlations were found with the two properties and KBS classification efficiency, suggesting that the KBS facilitates identification of a class beyond its inter-variability (heterogeneity) and its similarity with other classes (lack of uniqueness). Indeed, low CBVs indicate for complex situations or difficulties that are a result of insufficient indicators and/or evidences in the knowledge-base. The expert may locate these pixels, examine their relevant supportive and negative evidences and refine the knowledge-base to improve the already resolved cases and to resolve more of the unresolved conflicts.  High (CBV > 75%) Medium (50<CBV<75%) Low (CBV < 50%)
|
| < Prev | Next > |
|---|
