A Novel Histogram Thresholding Method for Surface Defect Detection

One of the most important applications of machinevision in various industries is automated inspection. Performanceof automated inspection depends directly on the algorithm usedfor threshold selection. Common methods of automaticthresholding are based on image histogram. In previous methods,the threshold selection has been realized by dividing thehistogram into two classes. Also, possibility of misdiagnosis ishigh especially for the textures without defect. This paperproposes a new statistical algorithm for automatic theresholdingwhich can be optimally applied in the presence of different typesof surface defects. The optimum threshold is obtained in theproposed algorithm so that a maximum between-class andminimum within-class variances are provided. Proposed methodsdemonstrate a better performance compared to classichistogram-based algorithm particularly for the textures withoutany considerable defects.

Keywords:

Histogram , Thresholding , Defect Detection