The optimal multi-degree reduction of Ball Bézier curves using an improved squirrel search algorithm

As a new nature-inspired swarm intelligence optimizer, squirrel search algorithm (SSA) has shown potential to solve several real-world problems, but for some complex problems, it still suffers from degraded performance. In this paper, a hybrid squirrel search algorithm (NOSSA) combined with optimal neighborhood update and quasi-opposition learning strategies is proposed to overcome the drawback of population update guided only by leading individuals in SSA. NOSSA adopts a stochastic optimal neighborhood update strategy to improve convergence speed and accuracy, and incorporates a Quasi-opposition learning strategy to enhance exploration. To verify its efficiency, NOSSA has been tested on 23 classic benchmark functions. Experimental results show that NOSSA has better performance on search-efficiency, convergence rate and solution accuracy compared with the representative stochastic optimizers. Furthermore, intelligent algorithms are introduced into the optimal multi-degree reduction of Ball Bézier curves and two new methods are proposed for the multi-degree reduction of center curve and radius function of Ball Bézier curve respectively. Experimental results demonstrate the effectiveness of the methods and show that NOSSA performs best among the representative stochastic optimizers in the degree reduction. The methods achieve the automatic and intelligent degree reduction of Ball Bézier curves.