论文部分内容阅读
针对现今煤岩图像识别方法的缺乏与不足,为了挖掘新的煤岩图像识别方法以及更好地处理高维煤岩图像数据,提出了基于最大池化稀疏编码的煤岩识别方法.本方法在提取煤岩图像特征时加入了池化操作,在分类识别时采用了集成分类器,即多个弱分类器组成一个强分类器.实验结果表明:最大池化稀疏编码的特征提取方式能简单有效表达煤岩图像的纹理特征,大大增强煤岩图像的可区分性,获得较高的识别率,并且具有良好的识别稳定性.研究结果可为煤岩界面的自动识别提供新的思路和方法.
In order to find out the new coal rock image recognition method and to better deal with high-dimensional coal rock image data, a coal-rock identification method based on the maximum pooling sparse coding is proposed in view of the lack of the present coal-rock image recognition methods. In the process of extracting coal and rock image features, a pooling operation is added, and an integrated classifier is used in classification and recognition, that is, a plurality of weak classifiers form a strong classifier. Experimental results show that the feature extraction method of the maximum pooling sparse coding can be simple and effective The texture features of coal and rock images are expressed to greatly enhance the distinguishability of coal and rock images and achieve high recognition rate and good recognition stability.The research results can provide new ideas and methods for the automatic recognition of coal and rock interfaces.