An enhanced markov random field (MRF) based approach for image segregation
The main objective of this thesis is to present an alternative technique to enhance the depth segregation of a 2D monocular image using Markov Random Field (MRF). Depth segregation considered challenging task due to an extensive segment of nonrigidity and textural contrasts among objects. Object...
Saved in:
| Format: | Thesis |
|---|---|
| Language: | English |
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/76657/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/76657/2/Full%20text.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/76657/3/Declaration%20Form.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The main objective of this thesis is to present an alternative technique to enhance
the depth segregation of a 2D monocular image using Markov Random Field (MRF).
Depth segregation considered challenging task due to an extensive segment of nonrigidity
and textural contrasts among objects. Object appearance in various shape and
location in the scene, depth segregation is likewise made difficult due to extra components,
occluded objects, which can be either visible or totally invisible from the scene,
variation in light distribution in image can give rise to a significant change in the aspect
of the objects in the image most likely to increase the difficulty of the process.
Here is what gives the rise to the problem of developing image segregation tools that
can deal with this variation in images, this tool required to be flexible and robust to
successfully segregate objects in to layers from a 2D monocular image. This thesis
shows that monocular depth segregation can be successfully used in edge region based
image segmentation. Depth image segregation enhanced technique has been proposed
to search for depth cues in the image regions, the image regions corresponding to a
specified level of depth cue in occluding and occluded objects such as T-junction and
L-junction can be identified and classified. The proposed technique initially executes
image segmentation; to identify region in the image, morphological operation; to eliminate
any error pixel in the region, then edge detection; to identify the boundaries of the
regions, and use these information to perform depth segregation processes to identify
and label the objects in depth labeled order. The experimental result shows that the
technique has successfully segregate the regions in depth order from a 2D monocular
image. The segmented image has been combined with the edge detection image to
perform the depth segregation process and the result was efficient in terms of object labeling
order. |
|---|