In view of the limitations of Flower Pollination Algorithm (FPA) such as insufficient global convergence ability, low searching accuracy and premature maturity, a flower pollination algorithm using a niche chaotic search strategy was proposed. In order to increase the breadth of algorithm search, niche technology was applied to keep the diversity of the population and the global optimization ability of the algorithm was improved. The chaotic sequence generated by the logic self-mapping function was introduced to locally optimize the elites of the algorithm. The performance of the algorithm was tested by classic test functions. The results show that compared with Flower Pollination Algorithm, Differential Evolution(DE) and Bat Algorithm(BA), NCFPA has better global search ability and search accuracy.