The traditional Slime Mould Algorithm (SMA) often suffers from slow convergence and a tendency to fall into local optima, significantly limiting its applicability in complex optimization problems. To overcome these limitations, a Strategy-based Fractional-order SMA (SFSMA) that integrates multi-strategy integration is proposed in this paper. Population diversity was enhanced through the synergistic combination of differential evolution and fractional order calculus, while a dynamic threshold mechanism monitored the search state in real time. These strategies collectively facilitated escape from local optima and improved convergence speed. To evaluate the performance of the proposed method, SFSMA was first compared with mainstream swarm intelligence algorithms and other fractional-order swarm intelligence variants using 12 classical benchmark functions. Experimental results confirm that SFSMA achieved significant improvements in both convergence speed and optimization accuracy. Furthermore, an SFSMA–Otsu segmentation model was developed by integrating the proposed algorithm with the two-dimensional Otsu algorithm. The model was evaluated on multiple types of images, including human, landscape, and medical images, using four quantitative metrics: peak signal-to-noise ratio, mean squared error, structural similarity index measure, and feature similarity index measure. Quantitative results demonstrate that the SFSMA–Otsu achieved substantially higher segmentation accuracy compared to existing methods. In addition, the convergence speed of SFSMA improved by approximately 82.08% compared to the traditional SMA. In conclusion, the proposed SFSMA effectively addresses the shortcomings of traditional SMA and provides an efficient and reliable solution for complex optimization and image segmentation tasks, exhibiting both theoretical value and practical potential.