Abstract: In order to improve the quality of image segmentation in engineering applications，an improved quantum-behaved particle swarm optimization (QPSO) algorithm is proposed on the basis of mutated QPSO，which is then combined with the maximum between-cluster variance method to present a multi-threshold image segmentation algo-rithm． The algorithm is characterized by a memory vector constructed from memory information in the search proce-dure of particles using Bayesian theorem． The memory vector is used to predict the future behaviors of particles and to assign the mutation probability of each particle automatically． In this way，the global search ability is enhanced and the convergence ability is preserved for the algorithm． Experimental results on Berkeley datasets show that the proposed algorithm is superior to two existing PSO-based methods because it helps obtain more stable and clearer image segmentation results．

Key words: quantum-behaved particle swarm optimization, memory information exploration, multi-threshold, image segmentation