Submit Manuscript

Information Technology Journal

Year: 2008 | Volume: 7 | Issue: 4 | Page No.: 560-569
Facebook Twitter Digg Reddit Linkedin StumbleUpon E-mail

Article Trend

Total views 361

Search

Authors

Mingxin Zhang

Junyi Shen

Zhaowei Shang

Keywords

Research Article

Color Image Fuzzy Classification Algorithm with Salient Regions

Mingxin Zhang, Junyi Shen and Zhaowei Shang
For the classical image segmentation algorithms, a larger number of implicit regions will be achieved, which leads to the lose of effectivity for these region-based image classification schemes. These individual regions are generally meaningful and unexpected results would be caused by these classification strategies, although some salient regions are extracted to reduce the complexity of region-based image classifications. Furthermore, these schemes are the arbitrary classification schemes and could not indicate the information about other implicit objects and the accurate semantic segmentation is still out of implicit semantic object regions in one image. In this study, we proposed one salient region based fuzzy classification (SrFC) method. Firstly, we give a salient region extraction algorithm based dominant colors and Gabor texture features. Then, according to the proposed classification scheme the salient regions from all the candidate images are clustered into disjoint categories. These categories are further used as for those for candidate images and for each candidate image its number of salient regions belonging to the same category would be used to evaluated the degree of the images belonging to the category. Experiments showed the propose SrFC scheme can achieve better performance than that of current common schemes. With comparison to the HCBA scheme for individual salient regions, the classification accuracy is improved about 16% by the proposed SrFC scheme.
PDF Fulltext XML References Citation

Leave a Comment

Your email address will not be published. Required fields are marked *