基于逻辑回归模型的木片和树皮的图像识别

摘要

为了提高木片生物能的利用率,需要除去木片中的树皮杂质. 识别木片和树皮的图像对于实现二者分离具有重要作用.提出一种基于逻辑回归模型来对木片和树皮进行图像识别的算法. 算法流程包括计算各种可以描述木片和树皮纹理的参数,利用SAS统计分析出最佳参数,将最佳参数计算所得的样本数据用于SAS LOGISTIC过程得到预测方程,再将预测方程用于逻辑回归模型来实现木片和树皮的图像识别. 通过实验可知,最佳参数为灰度方差,该图像识别算法的正确率能达到97.4%.%To improve the utilisation rate of wood chips bioenergy,the impurity of barks in wood chips must be removed.It is important to recognise the images of wood chips and barks for separating them.We propose an algorithm,it recognises the images of wood chips and barks based on logistic regression model.The process of the algorithm includes calculating various parameters capable of describing textures of wood chips and barks,analysing statistical results for best parameter by using SAS,getting prediction equation by applying the sample data calculat-ed with best parameter to SAS LOGISTIC process,and then realising the recognition of images of wood chips and barks by using prediction equation on logistic regression model.It is known through the experiment that the best parameter is the intensity variance and the accurate rate of the image recognition algorithm reaches up to 97.4%.