Double helix point spread function possesses continuously rotating invariant features with defocus, and it can be used for performing nano-resolution image of thick samples and for studying singleparticle tracking by combining the single-molecule localization method. However, the inadequacy of the original double helix point spread function is that the transfer function efficiency is very low, so it is difficult to achieve photon limited applications for the bio-imaging. The double-helix point spread function is optimized by introducing an iterative algorithm so that the constraints of the point spread function can be optimized best in three different domains: the Laguerre-Gauss modal plane, the spatial domain, and the Fourier domain. The simulation results show that the peak intensity of high efficiency double helix point spread function is 30 times higher than the original one. The design and fabrication of the phase plate are also proposed. The experimental results are in good agreement with the theoretical predictions, thus proving the feasibility of this method.