Indoor visible light localization has high requirements in terms of accuracy, in order to solve this problem,a Sparrow Search Algorithm (SSA) is proposed to optimize the indoor visible light fingerprint localization algorithm of Deep Belief Network (DBN). Firstly, the signal strength characteristic value and position coordinates are used to establish an offline fingerprint database. Secondly, the good global exploration and local development capabilities of the sparrow search algorithm are used to optimize the initial weight threshold of the deep confidence network, establish a network training model, and predict the position of the positioning target, so as to avoid the problem of DBN falling into local optimization and slow convergence. Finally, using the established offline fingerprint database data,the positioning error is calculated and analyzed. In the space experiment, the results show that the average positioning error of the algorithm in the paper is 3. 51 cm, and the probability of the positioning error within 6cm is 89. 9%, which is about 22. 5% lower than that of the DBN positioning algorithm.