A Count of Palm Trees from Satellite Image

TL;DR

This paper presents a method to count palm trees from high-resolution satellite images by exploiting their distinctive top-view shape and shadows, enabling accurate agricultural and environmental assessments.

Contribution

The study introduces a novel image processing technique that accurately identifies and counts palm trees using their unique shape and shadow features in satellite imagery.

Findings

Effective in high-resolution images like QuickBird (0.6m)

Less accurate with lower resolution images like SPOT (10m)

Applicable to aerial images and free satellite data from Google Earth

Abstract

In this research the number of palm trees was calculated from the satellite image programmatically, taking advantage of the accuracy of the spatial resolution of satellite image, the abilities of software recognition, and characteristics of the palm tree, which give it a systematic top view can be distinguished from the satellite image and the manner of cultivation and vertical growth and stability form for long periods of time. While other trees are irregular in shape mostly because of their twisted branches. Palm trees consist of a long stem, a large head, and a large flare that is almost circular and consists of large tufts. The palms have large self-shadows other than ordinary leaves. The large shadows and the circular shape of the upper view give it a special feature that we could use to design a program that distinguishes the shape of the palm without all the trees. Then it counts the number of palms in any field shown in the satellite image. This method is useful in counting the number of palm trees for commercial, agricultural or environmental purposes. It is also can be applied to high-resolution satellite imagery such as QuickBird because the resolution of the images is 0.6 meters. Less accurate images such as the 10-meter SPOT do not show the interior shadows of the top view of the palm enough, nor the accurate satellites (5 meters), while the interior shadows appear in high-resolution images only (0.6 meters) or below. It can also be applied to aerial images of less capacity because they are more accurate of course. Satellite images can be obtained free from Google Earth explorer, which can be downloaded free from the Google website. It connects the user to a global database of high-resolution images for all regions of the world.