Research on lunar soil has great value, and soil hardness is one of the essential parameters to be measured. Currently, the existing method involves contact measurement using special instruments. To reduce the self-weight and increase the scientific payloads of lunar rovers, a reanalysis of the data obtained from existing sensors provides a great solution for achieving different tasks. Therefore, a vision-based and noncontact method based on rutting images is proposed to estimate soil hardness. The overall approach includes two parts: rutting depth measurement and soil hardness measurement. To simulate a complex lunar environment, a variety of simulated lunar soil preparation schemes to achieve different soil states are designed in this paper. More importantly, this paper summarizes the empirical formulas of the simulated lunar soil in different states and the judgment conditions for defining the soil states based on the rutting depth. Finally, the accuracy, feasibility, and applicability of this method are verified. Experiments show the capability of the method we proposed to estimate soil hardness in simulated environments, as well as in the real lunar environment.