基于BP神经网络的彩色光弹性等色线自动定级法

把人工神经网络与光学彩色信息图像处理技术相结合，根据计算机的特点，选取红绿蓝（RGB）颜色模型，通过BP神经网络建立图像中像素点的RGB值与条纹级数F之间的非线性关系，对彩色光弹性条纹进行了插值，从而实现了等色线条纹的自动定级，用对径受压圆盘问题的理论精确解来检验，表明本方法与传统的分数级条纹补偿法相比，不仅可以实现条纹定级技术的自动化，而且精度较高，对提高三维结构光弹性模型切片的等色线测量精度有实用价值。

Automatic ordering of color isochromatics based on BP neural network

A systematic study has been carried out on how to use the red, green and blue (RGB) model for determining fringe order based on artificial neural networks. For this purpose the isochromatics is acquired by means of a color video camera and colors decomposed in the three primary colors. Fringe order is a complex function of RGB. The use of BP neural network to estimate this function is described. Tests of a circular disk under diametral compression are demonstrated. The results agree well with the theoretical exact solutions. The techniques can improve the precision of the experimental stress analysis in 3|dimentional photoelastic thin slices.