Conventional symbol recognition approaches overlook the critical issues of multiple moments and multicore programming. In this study, we develop a symbol recognition system - SymMoment - running multiple moments on a multicore processor. SymMoment integrates multiple moments to speed up the process of symbol recognition.SymMoment optimizes a moment’s order through multiple threads in parallel, thereby tremendously boosting system performance. SymMoment adopts an open-ended design and; therefore, other moments are readily incorporated as plugins. We make use of 6710 symbol samples to evaluate SymMoment. Compared against the popular moments including the Zernike and Tchebichef moments, SymMoment improves the recognition accuracy of the existing moment-based schemes by up to 89.1% with an average of 52.5%. SymMoment’s multicore computing speeds up the performance by up to 21.1% with an average of 8.5%. Furthermore, SymMoment reduces the optimization time of the Zernike and Tchebichef moments by averages of 10.1% and 6.6%.