On-line burning state recognition for sintering process using SSIM index of flame images

Recognition of burning state based on flame images has been an important issue in sintering process of rotary kiln. Existing methods usually adopt techniques of image segmentation, pattern recognition and machine learning, which have high demands on the quality of samples and computational power. It is challenging to the on-line recognition of the burning state, which is essential in realtime control systems. This paper proposes a new approach to the burning state recognition by comparing the structural similarity (SSIM) index of flame images. The burning state is identified according to the maximum SSIM index between the real time flame image and images in two standard libraries which consist reference images with normal-burning state and under-burning state respectively. This method has low computational complexity and is suitable for online control in the rotary kiln system. Simulation results show that the proposed method achieves high recognition accuracy with low computation.