Deploying multi layer extraction and complex pattern in fabric pattern identification

• Published in: Multimedia tools and applications Vol. 79; no. 15-16; pp. 10427 - 10443
• Main Authors: Kumar, K. Sharath, Bai, M. Rama
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.04.2020; Springer Nature B.V
• Subjects: Algorithms; Cameras; Classifiers; Color; Computer Communication Networks; Computer Science; Computer vision; Correlation analysis; Data Structures and Information Theory; Design; Equalization; Fault tolerance; Feature extraction; Histograms; Identification; Identification methods; Image enhancement; Image retrieval; Luminous intensity; Multimedia; Multimedia Information Systems; Neural networks; Object recognition; Pattern recognition; Robots; Special Purpose and Application-Based Systems; Support vector machines; Textiles; Texture recognition; Visual impairment; Complex pattern detection; HMAX model; Fabric weave patterns; SVM classifier
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-019-7421-y

Computer aided vision models are convincing so they can be correlated with the traditional classifier models. The prototype obtained serves as a vision for visually impaired by obtaining an outstanding interface platform for user in need. The proposed work focuses on fabric analysis for complex designs and provides audio and text identity guidelines by correlating the images under study. Colour based feature extraction is implemented to obtain global and local features which serves as an input for elaborative analysis of a fabric texture. The obtained image is partitioned to pixelated grains to achieve the above mentioned target. The features are formed as a cluster with the help of SVM classifier and HMAX(Hierarchical model and X) model. In order to improve the efficiency in complex texture fabric identification HMAX solution is combined with the traditional methods. HMAX provides good result on object recognition for methods involved in computer vision. The detail of each pixel is divided to undergo block operations to obtain solid and uniform colour estimation. The estimation identifies the color and further analysis is carried out for complex textures. The block obtained by that outcome is incorporated with the images under study and their respective description is provided for the user. Image enhancement can be achieved through histogram equalization. The methodology adopted proves reliable and effective solution. The pattern can be defined using parameters likes variance, histogram and gray level features. The existing methodology is incompetent due to lack of analysis in complex fabric patterns. The difficulty arises due to scale transforms, occlusions and light intensity. The challenge lies between preference and invariance. Therefore its important to develop a texture recognition system for complex patterns to analyse a fabric for visually challenged.