Home Articles Automatic Feature Extraction: A Solution for Extracting the Features from High Resolution...

Automatic Feature Extraction: A Solution for Extracting the Features from High Resolution Satellite Images.

Juber Ahmad
Solution Architect
DSM Soft,
Email: [email protected]

Mr. Murali CK
Sr. Technical Manager
DSM Soft
Email: [email protected]

Modern digital technology has made it possible to manipulate multi-dimensional signals (Images) with systems (computers) that range from simple digital circuits to advanced parallel computers. Advancement in digital technology has broadened the horizon of application of satellite images as well has redefined the way of digital data processing. This research paper is an effort to derive a solution for Automatic Feature Extraction from High resolution satellite Images (IKONOS & Quick Bird). Automatic Feature extraction has been always an interesting subject for researchers. In recent years various efforts have been made in this direction in recent past, but the fact still remains stand still “the result of the efforts are not reliable” therefore most of the data conversion exercise (feature extraction) is being done manually” Manual feature extraction takes a lot of time and holds majority of the share of solution implementation cost. Our study of existing solution/application shows that most of the solution developed for Automatic feature extraction either were designed using either radiometric or pattern recognition concept. Some of the works have been done using the both approach to get the optimum result. We have adopted a very simple but most practical approach for designing our solution. Basic idea behind our approach is quite simple that we should interpret the feature before extraction. Interpretation of image has been done by using the standard Image Interpretation keys. To get only the meaning full feature (fully controlled by user) we have also used the geometry identifier in our application which reduces the time in post data conversion process (editing).Above concept has been realized by using the Artificial Intelligence (Neural Network) in OOPS technology environment.