© 2015, Springer Science+Business Media New York. The detection of gravitational wave usually requires to match the measurement data with a large number of templates, which is computationally very expensive. Compressed sensing methods allow one to match the data with a small number of templates and interpolate the rest. However, the interpolation process is still computationally expensive. In this article, we designed a novel method that only requires to match the data with a few templates, yet without needing any interpolation process. The algorithm worked well for signals with relatively high SNRs. It also showed promise for low SNRs signals.