Defect Detection in Patterned Fabrics Using Local Binary Pattern and Its VLSI Implementation

Abstract

For texture classification in computer vision Local Binary Patterns (LBP) is used as one of the efficient method. In this paper, LBP texture operator is used to label the pixels of an image of any fabric or texture using the property of thresholding the neighborhood of each and every pixel and considers the compared result as a binary number or a digit which will be used for detecting the defects in patterned image of a fabric. This method is simple and gray scale invariant. The application of Local Binary Pattern described is defect detection on Patterned Fabrics in which the defect on patterned fabrics is detected by assigning the 0 pixel value to the defected window in an image. Further we will implement the same in VLSI, which was not specifically invented for image processing applications but its intrinsic parallel computation capability offers high performance image processing solutions, which are often tenfold and even hundred fold faster than traditional processors.

Keyword: FPGA, defect, fabrics, logarithmic likelihood ratio, computer vision, local binary pattern, texture