Modified Optimal Threshold Segmentation Method in Small Size Precision Measuring

To improve the accuracy and the efficiency for the small size measurement, a modified optimal threshold segmentation method was proposed. The traditional optimal threshold method can be used to do the image segmentation by calculating the missed classed probability. On the strong background noise condition, the missed classed probability distribution is unknown. And the optimal threshold obtained tends to be a local optimal solution. All these will cause measuring errors. To solve the problem, an over-segmentation coefficient was introduced into the optimal threshold method to restrain the local optimal solution. And the over-segmentation phenomenon caused by the introduction of over-segmentation coefficient will be compensated through the calibration of the small size measurement system. By the simulation and experiment analysis, the method could implement the small size precision measurement, which is simply to process, with less calculation and strong real time characteristics.