Flotation froth image classification using convolutional neural networks

In recent years, the use of machine vision systems for monitoring and control of the flotation plants has significantly increased. The classification of froth images is a critical step in development of an on-line machine vision based control system. Deep learning is a recent advance in machine learning that uses programmable neural networks to extract high-level features from image data. In this research study a convolutional neural network (CNN) is developed to classify the froth images collected from an industrial coal flotation column operated under various process conditions (air flow rate, frother dosage, slurry solids%, froth depth and collector dosage). In the first step, the froth images captured at different air flow rates are classified by the CNN algorithm and its classification accuracy is compared with a conventional artificial neural network (ANN). The results show that the froth classification system based on CNN significantly outperforms the ANN classifier in terms of classification accuracy and computation time. In the second step, the whole images taken under different operating conditions are classified using the CNN algorithm. The experimental results indicate that the CNN model is able to classify the froth images with an overall accuracy of 93.1%. The promising results of this study demonstrate the significant potential of deep learning neural networks in froth image analysis, which is of great importance for development of machine vision systems.