To obtain every single LED’s contribution when they illuminate the target locations and to acquire, record or reappear some specific light patterns, this paper explores how to estimate the parameters of Pulse Width Modulation (PWM). These parameters are amplitude, frequency offset and phase delay, and they are generally used to control the LED points. Firstly, the frequency offset-phase delay space is dispersed into reticulated grid spaces, and a sparse model is established based on the measured data with sparsity characteristics in the grid space. Then, on the basis of the sparse model, very few samples are utilized to reconstruct the unknown parameters fast by means of Orthogonal Matching Pursuit (OMP) algorithm. Finally, an iterative refinement grid technique is introduced to optimize the sparse model so as to suppress the estimation error effectively. Experimental results indicate that the method presented in this paper can reconstruct the unknown parameters fast by using only 27.5% samples that of the traditional Nyquist sampling theorem. Simultaneously, this algorithm is found robustness when Signal to Noise Ratio(SNR) is higher than 20 db after lots of contrast tests in different noise conditions.