|
|
|
Granular Computing Approach to Assess Landuse/Landcover Classification
V K Panchal
Defence Terrain Research Lab,
India
Email: vkpans@ieee.org
In the real world scenario, stringent requirement is collating multi-faceted geospatial information from various sources and assessing the accuracy in a time-constraint environment like disaster management, relief operations.
Granular Computing has emerged as a coherent and comprehensive framework for representing and processing heterogeneous pieces of information. It facilitates the human-centric pursuits, since much of the knowledge is perception based and granulation plays a key role in human cognition. Information granules are regarded as a direct manifestation of the fundamental mechanisms of abstraction and its practical realization lie in the heart of processes of knowledge acquisition and management. We discuss how Granular Computing helps realize their management and reuse. In this study, we briefly review the main features of Granular Computing, elaborate on the underlying formalisms of information granules (which include constructs such as fuzzy sets, rough sets, interval analysis, and shadowed sets, just to name the most visible representatives)
We will present how the mechanisms of collaboration and integration between sources of information can be established given their level of granular compatibility and the existing representation differences.
A simple and more concrete granular computing model is developed using the notion of information tables. Studies along this line have been carried out in the theory of rough sets. Within the proposed model, this paper reviews the pertinent existing results and presents their generalizations and applications pertaining to the assessment of landuse / Landcover classification. This is specially chosen for this study as it is a well recognized important task in landscape ecology. One of the tasks of concept formation (like the thematic categories) may be viewed as the representation, characterization, description, and interpretation of granules representing certain concepts. The motivation is to provide an efficient mechanism for time-constraints situations to not only resolve imprecision-based uncertainty but concept description based on the granular structure of the available information and content expansion of the granules to incorporate instance-based knowledge.
|
|
|