GGR 315 Environmental Remote Sensing, 2024

Assignment 5: Image Classification and color display

Questions

(100 marks, 10% of the final grade)

1. (20 marks) The following are the training field spectral statistics for three forest types:

Show the likelihood curves for these three cover types and classify the following image using the maximum likelihood rule. In the classified image use the following code: BS  for black spruce; PB for paper birch; and QA for quaking aspen. (5 marks for each line; 5 marks for classification results)

Original image

Classified image

2. (20 marks, 5 for each) From the confusion matrix shown below,

a. Calculate the producer’s accuracy for Type 2;

b. Calculate the user’s accuracy for Type 1;

c. Calculate the overall accuracy and the KHAT value of this classification (show intermediate calculations);

d. Explain why KHAT better represents the classification accuracy than the overall accuracy.

3. (20 marks)  Image classification. Following Lab 5 exercise, produce classified images using unsupervised and supervised methods with the following classes: vegetation, water and built area (bare soil). Attach final land cover maps in color in screenshots. (10 marks for each)

4. (20 marks)  Explain the main difference and similarity between supervised and unsupervised classification. (5 marks) Provide an error matrix for the supervised classification of the image in Lab 5. (10 marks) Discuss the accuracy assessment on the supervised classification. (5 marks)

5. (10 marks) Explain the following color display techniques: (a) true color, (b) false color, (c) greyscale and (d) pseudo color.

6. (10 marks) What are the three additive primaries? (2 marks) In a 6-bit display device, fill in the blank spaces in the three display channels (red, green, and blue) in the following table with digital numbers that produce the colors specified in the last column with the highest intensities. (8 marks)