The payload launched to deep space can be increased by using lunar gravity assist (LGA). There are two main methods for designing LGA transfer trajectory. The forward patched method firstly parameterizes the constrained orbit and then optimizes the orbit with the global optimization algorithm. The forward method is intuitive, but has many optimization parameters, which causes time-consuming calculation and poor convergence stability. The backward patched method firstly determines the optimal orbit of the heliocentric segment and then calculates the geocentric LGA. The backward patched method consumes less calculation time, but with poor accuracy, and it is not easy to converge to the global optimal solution. To solve the LGA escape trajectory in a short time, this paper proposes a hybrid optimization method based on the existing methods. The proposed hybrid optimization method makes full use of the advantages of fast convergence speed of backward patched method and high precision of forward patched method. The method takes the trajectory corresponding to the Pareto solution set of the multi-objective optimization of the backward patched method as the reference trajectory, and carries out local optimization near the reference trajectory. Then, the forward method is applied to quickly obtain the optimal transfer trajectory by using the global search algorithm. Numerical simulation was carried out for the detection of near-Earth asteroids 1989 ML and 2003 SM84. The results show that compared with the forward patched method, the hybrid optimization method improved the convergence stability and computational efficiency effectively. The convergence stability of the solution was improved to 3-4 times, and the calculation time was reduced by 50%. The results indicate that the hybrid optimization method excels in convergence stability and computational efficiency.