The microtremor survey method is characterized by the advantage of non-destructiveness to the geographical environment, safety, environmental friendliness and high signal-to-noise ratio, which possesses the broad application prospects in seismology and exploration geophysics. However, the traditional "distributed acquisition-centralized processing" model results in significant delays of shear wave velocity structure imaging, failing to meet the real-time exploration requirements. While the edge computing reduces the latency and network load, it faces challenges such as heterogeneous edge nodes and resource constraints, limiting the instant performance and accuracy of microtremor survey. To address these bottlenecks, this paper proposes a microtremor edge imaging method based on modified genetic optimization algorithm. First, a microtremor edge collaborative imaging system architecture is designed by deploying seismic edge servers to centrally manage sensor nodes, which enables the instant network edge imaging. Additionally in order to minimize the latency and control energy consumption, a multi-node collaborative computing framework is designed by tailoring the heterogeneous node resources and modeling the task allocation challenges. Finally, introduce the modified genetic optimization algorithm(MGOA) integrating crossover, mutation strategies and heuristic rules is introduced to efficiently solve the global optimal task allocation. The EdgeCloudSim simulations demonstrate that compared to traditional genetic algorithms, the proposed method reduces the total instant imaging time by 23.36% while maintaining 100% task coverage, which increases the higher computational efficiency by 29.8% than the CPLEX solver in medium-scale scenarios. The robustness tests show stability scores of 44.12 and robustness scores of 52.12, both surpassing the comparison algorithms. The field tests validate the method′s adaptability across different initial geological models with the low imaging errors and optimized edge device memory utilization at just 53%, making it suitable for the resource-constrained environments. In conclusion, it achieves the coordinated optimization of delay and energy consumption, providing an efficient solution for instant imaging in microtremor survey.