• To study the image segmentation and object-based classification methodologies.
• To develop the image segmentation and object-based classification software modules, integrarted in ENVI as its extenssion modules.
• To apply them on SPOT images acquired in 1995 and 2010 for monitoring the extenssion of Ha Noi urban land in that period.

Theoretical results:

• Study to master the principle of object-based classification methods. Fully collect and describe in detail the typical image segmentation methods, including:

a. Region-based segmentation methods:
- Seeded region growing;
- Unseeded region growing;
- Region Splitting and Merging;
- JSEG;
- Fast scanning algorithm;
b. Data clustering:
- Hierarchical agglomerative algorithm;
- Hierarchical divisive algorithm;
- K Mean;
- ISODATA;
c. Edge-based segmentation:
- Watershed algorithm;
- Fast watershed algorithm;

• Collect and describe some morphology indexes may be used in object classification;

• Analyze to select the algorithm, successfully develop two software modules:  

a. Image segmentation module, applying the modified for multispectral images fast scanning algorithm, allowing the user to choose the segment minimum size threshold;
b. Object classification module, using on the input the segmented image received from above image segmentation module, allowing the user to optionally choose one of 7 classification methods, including: parallelepiped, minimum distance, maximum likelihood, spectral angle mapper, spectral information divergence, Mahalanobis and binary encoding;

Applied results: The result of applying the developed software modules on SPOT images acquired in 1995, 2010 for monitoring the expansion of Ha Noi urban land in that period showed that the software modules fully meet the requirements and are ready for use.

Study to master the principle of object-based classification methods, analyze to select the algorithm, successfully develop two software modules: image segmentation and object classification. That allows to expand the use of a new, innovative classification method in Viet Nam.

The Remote Sensing Agencies in variety of Ministries and Universities.