Abstract | Full Paper | PDF | Printer friendly format

Page 1 of 3
| Next |

1. Introduction
Consider studying a satellite remote-sensing image of bush fire or forest fire. The image is colored-coded and one is interested to know the percentage of the area on fire in relation to a larger surrounded area. The shape of the area is usually irregular. In the absence of any proper tool, the two areas can only be estimated by crude eye inspection. Of course, a more accurate method of calculating the areas can be done by manually tracing the required regions on a piece of graph paper and then approximating the areas by counting all the small whole squares plus all the fractional portions. This will take up a lot of time. Moreover, a small estimation error can mean a large actual error because of the vastness of the landscape.

Our objective here is to present novel techniques which can compute the percentage accurately using a PC. The image under study must first be scanned into the PC, or is captured in a digital form which can be stored in the PC. The proposed algorithms can count the number of pixels of an enclosed region on the computer screen in order to obtain the area.