Maliheh Sadat Madanian; Alireza Sofianian
Volume 21, Issue 82 , September 2012, , Pages 44-49
Abstract
Change detection is the process of identifying changes in an object or phenomenon by observing it in different time intervals. Careful and timely detection of changes in land forms and reliefs provides a better basis for understanding relations and the interactions between human and natural phenomena. ...
Read More
Change detection is the process of identifying changes in an object or phenomenon by observing it in different time intervals. Careful and timely detection of changes in land forms and reliefs provides a better basis for understanding relations and the interactions between human and natural phenomena. In this way, it makes managing and exploiting resources possible. Remote sensing data is a wonderful resource for different applications in detecting changes, due to its temporal magnification, spectral and radiometric variety, appropriate digital format and integrated view. Many methods have been developed to detect changes, all of which have advantages and disadvantages. According to the studies, these methods show different results in the same environment. Generally, change detection methods are classified into 3 different classes: pre-classification comparison, post- classification comparison, advanced methods. The present article analyzes some of these methods like image subtraction, image division, main components analysis, detection of controlled changes, and detection of uncontrolled changes, hybrid, artificial neural networks, vegetation-impermeable surfaces-soil model and geographic information systems. Pre-classification methods detect changes caused by multi-temporal data without producing classified vegetation and land-use maps. Yet, post-classification methods provide a precise matrix of changes and they usually need input analysis. There are diverse advanced methods which are usually developed in response to specific studies. Studies indicate that image subtraction, main components analysis and post-classification methods are the most popular methods used for change detection. However in recent years, artificial neural networks and combinations of remote sensing and geographic information systems are regarded as important techniques.
