Research
Texture Image Segmentation using Morphology in Wavelet Transforms
One of the essential and crucial steps for image understanding, interpretation, analysis and recognition is the image segmentation. This paper advocates a new segme- ntation scheme using morphology on wavelet decomposed images. The present paper provides a good segmentation on natural images and textures by dividing an image into non overlapping regions, which are homogenous in terms of certain features such as texture, spatial coordinates etc. using simple morphological operations. Morphological enhancement technique based on Top Hat transforms enhances the local contrast in this paper. The morphological treatment and followed by Otsu’s threshold overcomes the problem of noise and thin gaps, and also smooth the final regions. The experimental results on four different databases demonstrate the success of the proposed method, compared to many other methods.
Image Retrieval based on Macro Regions
Various image retrieval methods are derived using local features, and among them the local binary pattern (LBP) approach is very famous. The basic disadvantage of these methods is they completely fail in representing features derived from large or macro structures or regions, which are very much essential to represent natural images. To address this multi block LBP are proposed in the literature. The other disadvantage of LBP and LTP based methods are they derive a coded image which ranges 0 to 255 and 0 to 3561 respectively. If one wants to integrate the structural texture features by deriving grey level co-occurrence matrix (GLCM), then GLCM ranges from 256 x 256 and 3562 x 3562 in case of LBP and LTP respectively. The present paper proposes a new scheme called multi region quantized LBP (MR-QLBP) to overcome the above disadvantages by quantizing the LBP codes on a multi-region, thus to derive more precisely and comprehensively the texture features to provide a better retrieval rate. The proposed method is experimented on Corel database and the experimental results indicate the efficiency of the proposed method over the other methods.
