Multi-threshold remote sensing image segmentation with improved ant colony optimizer with salp foraging

Abstract Remote sensing images can provide direct and accurate feedback on urban surface morphology and geographic conditions. They can be used as an auxiliary means to collect data for current geospatial information systems, which are also widely used in city public safety. Therefore, it is necessary to research remote-sensing images. Therefore, we adopt the multi-threshold image segmentation method in this paper to segment the remote-sensing images for research. We first introduce salp foraging behavior into the continuous ant colony optimization algorithm (ACOR) and construct a novel ACOR version based on salp foraging (SSACO). The original algorithm's convergence and ability to avoid hitting local optima are enhanced by salp foraging behavior. In order to illustrate this key benefit, SSACO is first put up against 14 fundamental algorithms using 30 benchmark test functions in IEEE CEC2017. Then, SSACO is compared against 14 other algorithms. The experimental results are examined from various angles, and the findings convincingly demonstrate the main selling point of SSACO. We performed segmentation comparison studies based on 12 remote sensing images between SSACO segmentation techniques and several peer segmentation approaches to demonstrate the benefits of SSACO in remote sensing image segmentation. Peak signal-to-noise ratio, structural similarity index, and feature similarity index evaluation of the segmentation results demonstrated the benefits of the SSACO-based segmentation approach. SSACO is an excellent optimizer since it seeks to serve as a guide and a point of reference for using remote sensing image algorithms in urban public safety.