The Energy Gradient Method Based on Two-Dimensional Discrete Wavelet to Extract the Feature of Pilling

摘要

Traditional fabric pilling detection susceptible influenced by the expert's subjective factors, so it has high false detection rate, slow speed and low efficiency. In this paper, we proposed an automatic fabric pilling detection method using image processing technology which based on computer vision. We also proposed three kinds of algorithm base on energy to separate the texture and pilling from fabric image and compare these three algorithms. Finally, a number of pilling characteristic values are extracted as a basis of pilling rating.