Natural image matting based on surrogate model (IMBSM) is an computationally efficient image matting algorithm that use surrogate models to approximate the pixel pair evaluation function and the optimal solution of the surrogate model obtained efficiently as the approximate optimal solution of the pixel pair optimization problem.

If you use this source code please cite:
Liang Yihui, Gou Hongshan, Feng Fujian*, Liu Guisong, Huang Han. "Natural image matting based on surrogate model." Applied Soft Computing 143 (2023): 110407.

cd path     %go to the path of the source files
IMBSM_Demo %run the demo