Area Estimation

Application of the technique

After having checked the results given by the technique, we attempt to use it to determine the area of an irregular shape. Note that the technique is easily used only for convex shapes (i.e. no ‘holes’ or ‘caves’ in the shape). I chose my old school, Xavier School – Greenhills in San Juan City.

Searching my old school in Google maps, I am again reminded of how much things have changed. Huhu. Anyway, the view from Google maps can be found below. Beside it is the B&W binary image which can be done using Paint, Photoshop or Gimp.

B&W binary image of Xavier School

View of my old school from google maps.

Note that the method above calculates the area in pixel units. To convert pixel units to physical units, ratio and proportions may be used.

For the sake of comparison and to check the results, I used the built-in measurement tool of Google maps to know what area I should be getting if my method is correct. This is how the tool works.

The very awesome Google Maps.

1. Right click on an edge point in google maps.
2. Select Measure distance.
3. Select enough points around the desired location so as to get a smooth outline.
4. Finish tracing by selecting your first point.
5. The total area and total distance should pop up automatically.
6. Go Google!

For more comparison, I also used the built-in function of Matlab to get the area. The results are the following.

Area in m²

As in the case of the triangle, the calculated area is again larger than expected. Since the scaling factor I used for the simulated and built-in function is the same, this means that there is something amiss with the method I used.

ImageJ

A well-known free software for image processing is ImageJ, which is often used in medical applications. In this section, I explore its basic function to compute area.

I scanned my very weathered UP ID along with a triangle with rulings of known length. The ruler is important to know the necessary scaling factor to convert pixel units to physical units. Using that data, ImageJ reported my image to have an area of 44.95 cm². This is compared to the actual area of 46.17 cm², which is quite close. The actual area may be larger because I approximated the area to that of a rectangle, while it’s corners are actually rounded. Also, the ID is at the top of the scan, and some parts may have been cropped. There is also some uncertainty as I only used a ruler to measure the sides of my ID.