Image Classification Based on Effective Probabilistic Latent Semantic Analysis Model

D. Antony Pandiarajan; S. N. Nisharani

This article proposes a new method for classification of rock images using Tamura features and an effective topic generation model called probabilistic latent semantic analysis (PLSA). The rock textures can be very well represented by the six Tamura features known as coarseness, contrast, directionality, line likeness, regularity and roughness. A topic model is generated by applying Tamura features to PLSA. The Sum of Square Difference (SSD) classifier is employed for the classification process. The SSD classifier is applied over the topic model to classify the rock texture. This classification is compared with GLCM, color co occurrence and Tamura features methods. This method gives the accuracy of 74.33%.

Select Volume / Issues:

• Volume 4 Issue 3, May. 2013

Year:

2013

Type of Publication:

Article

Keywords:

Rock Images; PLSA; Tamura Features; GLCM; Color Co Occurrence CCM; SSD Classifier; Topic Model

Journal:

IJECCE

Volume:

4

Number:

3

Pages:

833-839

Month:

May