According to the characteristics of the flotation froth
image texture features, a method for the extraction of significant patterns
based on multi-layer SVMs(MLSVMs) is introduced. Firstly, the numerical
flotation froth image is analyzed to extract texture features, such as
energy, entropy and inertia, based on grey-level co-occurrence matrix(GLCM)
to provide qualitative information on the changes in the visual appearance
of the froth. MLSVMs classifier， which is trained with the sampling data
from above texture features, identifies out the three types of flotation
production states. The particle swarm optimization(PSO) algorithm with
improved inertia weights is adopted to optimize kernal function parameters
of MLSVMs. The test results show that the proposed classifier has an
excellent performance on training speed and correct recognization ratio, and
meets the requirement for real-time monitor for the flotation process.