ABSTRACT

The separation of coal and gangue is an important part in the coal-preparation process, which can be realized by image recognition technique. The unique optical texture of coal is the most obvious visual features to distinguish coal and gangue. The current research represented the attempt to implement the multifractal analysis in the optical texture of coal and gangue. Firstly, fractal theory is applied to verify the multiscale characteristics of coal and gangue images. Then a new method of using the multifractal spectrum to describe and identify the optical texture characteristics of coal and gangue is constructed. Finally, we analyzed the coal and gangue images according to the physical meaning of the parameters in the multifractal spectrum. The results demonstrate that the optical texture of coal and gangue presents the characteristics of complexity and self-similarity, and the multifractal spectrum is an appropriate tool for characterizing the optical texture structure of the coal and gangue. The representative parameters in multifractal spectrum can be used as efficient features for the recognition of coal and gangue. It is thus clear that the multifractal model is a novel and feasible method to describe and recognize the optical texture characteristics of coal and gangue images.

摘要

煤矸石分离是选煤过程中的重要环节，可通过图像识别技术实现。煤独特的光学纹理是区分煤与矸石最明显的视觉特征。目前的研究代表了在煤和矸石的光学结构中实施多重分形分析的尝试。首先，应用分形理论验证煤和矸石图像的多尺度特征。进而构建了一种利用多重分形光谱描述和识别煤矸石光学织构特征的新方法。最后，根据多重分形谱中参数的物理意义，对煤矸石图像进行了分析。结果表明，煤矸石的光学织构呈现出复杂性和自相似性的特点，多重分形光谱是表征煤矸石光学织构结构的合适工具。多重分形谱中的代表参数可作为煤矸石识别的有效特征。由此可见，多重分形模型是描述和识别煤矸石图像光学纹理特征的一种新颖可行的方法。