I am doing a land-use change detection analysis with high resolution (1.5m) image? . Conventional with such an exercise, accuracy assessments (AC) are required (especially for low resolution images e.g. Landsat). But I have a serious feeling that AC may not be necessary for a high resolution image analysis, given that all features are visible. I would greatly appreciate your thoughts and ideas on this or possibly any useful literature. Thank you!
Sure. An accuracy assessment is necessary. You will be carrying out an image segmentation so an object oriented classification instead of a pixel based one. Your accuracy will base on the number of objects or point correctly identified and delineated instead of pixels. Ground truthing will be very important in this exercise.
Dear all, many thanks for your kind responses. @ Mehdi, thank you for the link. I got the paper and its very useful. @Joseph, thank you for clarifying that. Indeed, I performed image segmentation and then I supplemented this with ground-truth data before classification.
Any work and results carried out using remote sensing data need field observation and accuracy assessment ..both will strength your work and result
We did a large area in Haiti and, via high resolution images, were able to identify land uses and crops. the site visit confirmed the assumptions and helped better define the visuals for later use. Ground truthing is important, by virtue of oblique imaging, verification with on-site observers, and ground based images for local resource people.
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking