Transferring of quantization noise in algorithm implementation is tracked, which is caused by the finite sampling width of far field photodetector. The effect of quantization noise on wavefront aberration correction effect and convergence speed is analyzed. According to the changing of Strehl ratio (SR), a modal reconstruction algorithm based on dynamic region extraction is presented. Numerical calculations are carried out with the wavefront which fits the Kolmogrove atmospheric turbulence power spectrum and is obtained by the simulation of 18- and 33-order Zernike terms. The results show that, under the condition of the camera sampling width of 12 bits and 33-order initial aberration, the SR is larger than 0.9, the root-mean-square value of wavefront recuperative residual is 0.058λ (λ is wavelength), and the algorithm is converged after iteration for 31 times. The proposed algorithm greatly improves the correction performance and convergence speed of the wavefront sensorless adaptive optics system and reduces the effect of the quantization noise.