Collaborative Filtering (CF) has become one of the most successful approaches for providing personalized product recommendations to users. Neighborhood-based CF is one of the main forms among all CFs, which is widely used in commercial domain. However, neighborhood-based CF suffers from new user cold-start problem in sparse rating data. In this paper, we propose an improved neighborhood-based CF recommendation algorithm based on Bhattacharyya Coefficient to address the new user cold-start problem. The proposed algorithm combines the item neighborhood information with the user neighborhood information to improve the recommendation precision. Finally, the proposed algorithm is tested on a real dataset and the results show the proposed algorithm has the better recommendation performance.