The flotation froth surface appearance includes remarkable information, which can be employed as a helpful index for the qualitative evaluation of the process efficiency. Image analysis is a practical technology for the sake of achieving process related information that can be employed in expert controllers in order to amend flotation performance. In this paper, the intelligent modelling of relationship between froth characteristics and the metallurgical performance in a pilot column flotation of iron ore was established. Column flotation tests were carried out at a wide range of operating conditions and the froth features along with the metallurgical performances were specified for each run. The artificial intelligence models suggested for the performance parameters prediction include (1) multi-layer back propagation neural network (BPNN), (2) hybrid BPNN with principal component analysis (PCA). The hybrid network was on the basis of the PCA employment in order to decrease the number of variables to be given as input for BPNN. The relationships between the froth features and metallurgical performance factors were successfully modelled via the use of the two methods. The simulation results revealed that the prediction precision of BPNN model on the basis of all the data was relatively higher than the estimation precision of BPNN based on PCA algorithm. The Hybrid BPNN model that was trained by the pre-processed database of measurements achieved from the PCA can be considered a robust method when training time is of paramount importance in objectives of process control.