Two Dimension Threshold Image Segmentation Based on Improved Artificial Fish-Swarm Algorithm

To deal with the deficiency of the existing image threshold segmentation algorithm, combining the advantages of Artificial Fish-swarm Algorithm (AFSA), based on the improvement of the AFSA, the thesis proposes a new improved image threshold segmentation algorithm which is the combination of the AFSA and two-dimensional Fisher evaluation function. According to the principle of Fisher pattern recognition, based on two-dimensional histogram, by defining a two-dimensional Fisher criterion functions which is also used as the food concentration function in AFSA, it puts forward a two-dimensional improved image threshold segmentation algorithm. The Experimental results show that the algorithm not only gets the desired segmentation results, but also reduces the time-consuming greatly and thus achieve the fast segmentation goal.