Abstract: With the development of modern wide-field optical transient (OT) surveys, the traditional methods of cross identification are facing a great challenge in real-time search of OT sources from data of large volumes. To overcome the difficulties encountered by the traditional methods, we propose a novel algorithm to speed up the search. This algorithm divides the sky coverage of the object catalog into a grid of regions with a constant size in both Right-Ascension (RA) and Declination (Dec) directions. It uses a fast approach to index the grid regions, and maps them into a two-dimensional array in the computer memory. Our tests show that it takes the algorithm 0.3 seconds to perform cross identification in a catalog of 220 000 records if run on a computer of an Intel Core i7 2600k CPU. This is about 34 times faster than the HTM algorithm. This indicates that our algorithm can meet the requirements on fast real-time OT-source search in a modern wide-field OT survey project, such as the Chinese Ground Wide-Angle Camera (GWAC) project. We also explain in detail the parameters of the functions of the algorithm and specify the optimal values of the parameters. We finally discuss the scope of applications and certain features of the algorithm.