Approach the Interval Type-2 Fuzzy System and PSO Technique in Landcover Classification

In fuzzy classification systems, the estimation of the optimal number of clusters and building base-rules are very important and greatly affects the accuracy of the fuzzy system. Base-rules are often built on the experience of experts, but this is not always good and the results are often unstable. Particle swarm optimization (PSO) techniques have many advantages in finding optimal solutions and have been used successfully in many practical problems. This paper proposes a method using the PSO technique to build base-rules for the interval type-2 fuzzy system (IT2FS). Experiments performed on satellite image data for the landcover classification problem have shown that the proposed method works more stably and effectively than the non-PSO technique. © 2020, Springer Nature Switzerland AG.