Image Segmentation Based on Dynamic Particle Swarm Optimization for Crystal Growth

In order to realize the intelligent production of sapphire crystal, it is important to obtain the growth status from the furnace by charge coupled device (CCD). However, a significant challenge is that traditional approaches are often not valid to separate the images of the melting interface well due to the low contrast and uneven brightness from the heater. In this paper, an improved Otsu algorithm based on dynamic particle swarm optimization (DPSO) is proposed to find the exact threshold band as contour of the crystal. In this method, the Otsu method is constructed firstly, then DPSO is used to find the optimal threshold band. Experimental results show that the proposed algorithm can separate the texture of crystal growth images well and has high robustness.