Fuzzy mamdani inference system skin detection
Skin detection is well known to detect the appearance of human and human parts within an image. However, there are several limitations exist in skin detection when using skin colour as cue to detect skin appearance. These limitations include problems such as illumination, skin-like pixels and camera...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Book Section |
| Published: |
IEEE
2009
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/14626/ http://dx.doi.org/10.1109/HIS.2009.224 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Skin detection is well known to detect the appearance of human and human parts within an image. However, there are several limitations exist in skin detection when using skin colour as cue to detect skin appearance. These limitations include problems such as illumination, skin-like pixels and camera characteristic. In this paper, a set of modified fuzzy rules has been introduced to deal with the skin-lie pixels problem. These modified fuzzy rules were integrated with skin modelling method in order to discriminate skin pixel and non-skin pixel. The experiment conducted in this paper is classification of human skin image and animal images. The experimental result is then compared with explicitly defined skin region and fuzzy sugeno classification method. From the experiments, we have found that the proposed fuzzy rules are applicable if the RGB value of pixel does not close to low value. |
|---|