Classification of Solder Joint Based on Statistical Feature

摘要

In this paper, a minimum risk Bayes classifier based on a new statistical feature is proposed. Using a 3-color (red, green, and blue) hemispherical light-emitting diode (LED) array illumination and a charge coupled device (CCD) color digital camera, the images of solder joint are obtained. Then the color features including the average gray level with the percentage of highlights and the new statistical feature are extracted. The minimum risk Bayes classifier is introduced based on these features. And the inspection results of the Adaboost algorithm approve the performance of the Bayes classifier. The experiments show that the proposed algorithm based on statistical feature increased the inspection accuracy.