Research Of Fabric Defect Detection Based On OMAP3530Embedded System

Fabric defect detection and classification are important proceduresforquality controlintextilemanufacturingindustry. The automatic detectionand classification for fabric defects have been a research focus incomputer vision field for several decades. Currently, manual fabric defectdetection and classification are conducted in China with thesedisadvantages such as high labor strength, low efficiency, higher falsedetection rate and so on. Therefore, the research on the system ofautomatic fabric defect detection and classification has important socialand economic significance.In this thesis, the filter group containing threeannular Gaussian band-passfiltersis designed based on the feature of non-defect fabric image. Twodefect detection algorithms based separately on the annular Gaussianband-pass filters in frequency domain and their spatial-domainconvolution masks are designed to detect the fabric defects. Many fabricsample images that contain different types, sizes, and shapes ofdefects.are processed by the two algorithms, and the experiment resultsshow that the fabric defects are successfully segmented from the texturebackground.Also, making use of Linux as the operating system, DVSDK as thesoftware tools, the software design of the fabric defect detectionalgorithm based on two-dimensional spatial-domain convolution masks isaccomplished on the development platform of OMAP3530embeddedsystem. The experimental result shows that this software can finish thefabric image data inputting, data Processing by DSP and the displaying ofdefect image correctly.