The emission spectrum detection and diagnosis is one of the most common methods of application to the plasma. It provides wealth of information of the chemical and physical process of the plasma. The analysis of discharge plasma dynamic behavior plays an important role in the study of gas discharge mechanism and application. An air dielectric discharge spectrum measuring device was designed and the emission spectrum data was measured under the experimental condition. The plasma particles evolution was analyzed from the emission spectrum. The numerical calculation model was established and the density equation, energy transfer equation and the Boltzmann equation was coupled to analyze the change of the particle density to explain the emission spectrum characteristics. The results are that the particle density is growing with the increasing of reduced electric field. The particle density is one or two orders of magnitude difference for the same particle at the same moment for the reduced electric field of 40, 60 or 80 Td. A lot of N2(A3), N2(A3) and N2(C3) particles are generated by the electric field excitation. However, it transforms quickly due to the higher energy level. The transformation returns to the balance after the discharge of 10-6 s. The emission spectrometer measured in the experiments is mostly generated by the transition of excited nitrogen. The peak concentration of O2(A1), O2(B1) and O2(A3Σ+u) is not low compared to the excited nitrogen molecules. These particles energy is relatively low and the transition spectral is longer. The spectrometer does not capture the oxygen emission spectrum. And the peak concentration of O particles is small, so the transition emission spectrum is weak. The calculation results of the stabled model can well explain the emission spectrum data.