The LSAC model has good adaptability and strong information processing capabilities. Introducing it to solve the problem of human target segmentation in thermal infrared images will bring the following advantages.

 

(1) The LSAC model provides a flexible and efficient speech embedding framework. This framework is easy to integrate various image information and various information processing tools, and work together to achieve the purpose of image segmentation, thereby greatly enhancing the potential of obtaining good segmentation results on thermal infrared images.

(2) The numerical calculation of the model works in the continuous domain, and the segmentation results with sub-pixel accuracy can theoretically be obtained.

Therefore, the segmentation accuracy is higher than many traditional methods.

(3) The model can output continuous smooth and closed target contours, which can be directly used in subsequent human target recognition, tracking and other links, avoiding a large number of post-processing operations, simplifying the processing process and indirectly improving the output accuracy of the system.