基于BP神经网络的图像颜色温度检测方法研究

Study on Temperature Measurement Using Image Color Based on BP Neural Network

高温物体的颜色与物体的温度存在复杂的非线性关系.基于BP神经网络在非线性函数逼近方面有优良特性,提出了用BP神经网络拟合高温物体颜色模型HSV中参数H、S与温度的映射关系的测温法.实验证实该测温法是可行的,且较传统的测温方法简单,误差小.

The color of the high-temperature object has the complex nonlinear relationship with the temperature. Based on that BP neural network has more advantaged on approxirhating the nonlinear function , a method of temperature measurement is proposed , BP neural network is used to approximate the nonlinear relationship between Hue, Saturation of the HSV model and the temperature of the high-temperature object. The experimental result shows that the method is feasible, simple, and has fewer error compared with the traditional method.