Based on a large national sample of the China Family Panel Studies (CFPS) database, the study will to provide some thoughts and suggestions for the transformation and upgrading of China's agriculture, the enhancement of total factor productivity by exploring the agricultural production efficiency from the micro-farmers' perspective. By constructing the models of Charnes, Cooper & Rhodes (CCR), Banker, Charnes & Cooper (BCC), Slacks-Based Model (SBM) and technology optimization Malmquist-Luenberger index, it finally obtained the comprehensive technical efficiency value, pure technical efficiency value, farm household efficiency value and green total factor productivity level value at the micro-farm household level based on the comparative analysis. It was found that: the comparison of the measures based on different models found that although there are differences in the calculated efficiency values, the pure technical efficiency values calculated by BCC are the main factors affecting the micro agricultural production efficiency values at the farmer level, the SBM model should optimize the CCR, BCC models, and more suitable for Chinese government policy formulation and optimization; the technology optimization Malmquist-Luenberger index method is the micro agricultural production efficiency measurement method of choice, with the characteristics of diverse model selection, rich application scenarios and convenient processing of negative outputs; environmental factors in the current evaluation of agricultural green total factor productivity, mainly play a negative inhibitory role, reducing the negative externalities of environmental variables output, become one of the key issues facing the current micro farm layer green total factor production enhancement; the combination of subjective and objective measures of environmental non-desired output is an important way to measure environmental factors of agricultural green total factor productivity, it can be used in practical applications based on a combination of research objectives and data availability.