A noise power spectrum estimation algorithm is presented, which continuously updates the estimated noise by weighted noisy speech for non-stationary noise environments. To avoid noise overestimates in the weak speech regions following strong speech, a projecting smoothing method is developed for weighting factor calculation. The update for varied noise power levels is much faster compared to most other algorithms, and at the same time, noise power spectrum is not overestimated. Experimental results indicate that the proposed noise estimation algorithm yields lower segmental estimation error and better Perceptual Evaluation of Speech Quality(PESQ) scores when integrated in speech enhancement.