Fly Local Sensitive Hashing (FLSH) is a biomimetic data-independent hashing method inspired by the mechanism of odor processing system in drosophila. In this paper,we propose a novel Randomized Sampling-based Fly Local Sensitive Hashing (rs-FLSH) to model the randomness occurred during the establishment of synapses between neurons.Significant performance improvement can be achieved by applying a novel randomized sampling scheme in rs-FLSH,in which the sample rate is modeled by a Gaussian random variable rather than a fixed value in FLSH. Experimental results on benchmark dataset show that our proposed method outperforms FLSH,and achieves considerable performance comparing to several data-dependent hashing methods.