1998 年 16 巻 2 号 p. 199-206
Binarizing images are used usually for high speed processing in a vision system for an automatic recognition, inspection, or positioning of the micro-joints. In processing used a binarizing image, an automatic setting method of the threshold value is very important for a reliability improvement. The conventional binarizing methods are not corresponding to the change of the lighting. Especially, the threshold value is not select suitably for an image in which the brightness of the processing object exceeds a maximum value of quantization level. Therefore, in the conventional binarizing method, the size of the region divided by the binarizing processing has done a change by lighting environment at picturing. The brightness of the processing object exceeds a maximum value of quantization level in the pictured image of the printed circuit board, because the electric components mounted on a printed circuit board have various reflection rate for the light. Then, in this research, an automatic detection method which corresponds to a change of lighting environment that the brightness of the processing object exceeds a maximum value of quantization level is discussed. We propose a interval discriminate analysis thresholding method in which the threshold value is calculated based on the statistical analysis under the condition that the brightness of class 2 is transformed to the brightness at the boundary. In this proposal method, binarizing can be done at the same level relatively corresponding to a change of lighting environment. Furthermore, every valley part can be detect simultaneously for the gray histogram presenting many peak distribution.