Primary particles produced in the early fire period evolve into bigger aggregates due to Brown coagulation, which take on fractal properties. In this article, smoke particle images were obtained by a field emission scanning electron microscope (SEM), and the fractal characteristics of the particles were analyzed based on the power law of fractal theory. By using digital image processing technique, more fractal parameters such as the fractal dimension, fractal prefactor, primary particles' size and radius of gyration were obtained. The results can provide a fundamental basis for fire smoke detection research.