基于非接触式温度测量中的高温熔体识别方法

Recognition method in contactless temperature measurement of high temperature melts

提出一种基于灰度阈值分割的高温熔体识别方法。采用双峰法和最大方差自动取阈值法分别确定预分割阈值,再根据高温熔体的灰度值近似相等选择合适的分割阈值,使所分割的目标的灰度值的等级数较少,并使高灰度值的R色总数最多。根据这种方法,利用MATLAB软件,对拍摄的某铜冶炼厂高温熔体图像进行仿真实验,结果表明对于目标和背景对比强烈的图像,采用双峰法分割效果较好;而最大方差自动取阈值法更适用于图像中目标占很大比例的情况;这种识别方法能自动调整不同拍摄条件下的图像分割阈值,从而准确地识别目标。

A high temperature melts recognition method was proposed based on the gray-levels threshold segmentation. By using the double-peak method and the maximum-square-error method, the elementary threshold of segmentation was determined. Then, the suitable threshold was selected according to the approximate equivalency of gray value of high temperature melts. By using this recognition method, less levels of gray value of segmented target and the most sum of higher gray value of R color were identified. The image simulation of high temperature melts from a copper smelting plant was carried out with software MATLAB. The results show that the double-peak method has better effect when the contrast of the object to background is sharp and the maximum-square-error method is preferable when the area proportion of object is dominant. The threshold can be adjusted automatically in different conditions of shoot and the object can be recognized accurately.