Question

Training dataset is used to identify desired features in remote sensed images in

• A. Supervised classification method
• B. Self-learning image analysis method
• C. Classification using manually
• D. Un-supervised classification method

Hint:

To remember:
- Supervised = User guides the software with "training sites" first, then the software classifies.
- Unsupervised = Software groups pixels into "natural clusters" first, then the user identifies the classes.

Answer 1

The Correct Option is A

Step 1: Understanding the Concept:
Image classification is the process of grouping pixels in a multispectral remote sensing image into land cover classes (such as forest, agriculture, water, or urban area).
The two primary methods used in remote sensing are supervised classification and unsupervised classification.

Step 3: Detailed Explanation:
Let us analyze the differences between these two classification methods:
1. Supervised Classification:
In this approach, the user selects representative sample sites of known land cover types, called "training areas" or "training datasets."
The image processing software analyzes the spectral signatures of these training pixels to build a statistical model (such as Maximum Likelihood, Minimum Distance, or Support Vector Machine).
The software then uses this model to classify every remaining pixel in the image based on which training class its spectral properties match most closely.
Thus, the use of a training dataset is the defining feature of supervised classification.
2. Unsupervised Classification:
In this approach, the classification is purely data-driven.
The software groups pixels into spectral clusters based on natural groupings of pixel values, without any prior training dataset.
The user then labels these resulting spectral clusters with land cover names afterward.
Thus, Option (D) is incorrect.

Step 4: Final Answer:
Training datasets are used to identify desired features in supervised classification.
Therefore, the correct option is (A).