Authors : Labonnah F. Rahman, Md. Mamun , H. Husain , Mohd. Marufuzzaman and Yushaizad Yusuf
Abstract: Texture refers to the surface properties that can be easily described by its primitives (ton) and their spatial relationship. Texture analysis is a process to find the shape, segment or identify the region of interest on the object. In this research, researchers approach to implement unsupervised texture segmentation in hardware. First, the Discrete Wavelet Transform (DWT) is used to extract the input image and sample it into different frequency bands. After this process, the input image becomes smaller and compressed. This input image is then fed into fuzzy K-mean clustering algorithm. Fuzzy K-mean is a well-known and precise supervised clustering algorithm that divides the image into different segmentations and many computations are needed for the correct segmentation assignment. Hence, the total execution time for segmentation is improved using compressed image as input. From the simulation result, images with 2-5 types of textures were successfully detected around 0.025-0.033 sec. The proposed hardware approach for wavelet based texture segmentation is able to reduce the execution time to enhance the performance for 128x128 pixels which can be considered as fast segmentation solution.
Labonnah F. Rahman, Md. Mamun , H. Husain , Mohd. Marufuzzaman and Yushaizad Yusuf , 2011. A Hardware Approach for Real-Time Fuzzy Wavelet Based Textures Segmentation. International Journal of Signal System Control and Engineering Application, 4: 46-54.