Rosa rugosa Thunb., Rosa sp. and Rosa chinensis Jacq. all belong to the genus Rosa L. The three kinds of flowers are similar in appearance and easily confused. The genus Rosa L. has important ornamental value and medicinal value, so it is of great significance to quickly identify the genus Rosa L. Laser induced breakdown spectroscopy (LIBS) is used for in?situ detection of the main elements in the three flowers in this work. In addition, CN can also be recognized in the spectrum of the genus Rosa L, sothe CN in the spectrum is simulated by using LIFBASE, and the vibration temperature and rotation temperature of CN are calculated. Then the parameters obtained can be regarded as experimental parameters. By comparing and analyzing the laser induced breakdown spectra of the three different kinds of flowers, the characteristic spectral lines with significant difference in intensity are selected as variables to predict flower genus, and combined with the general regression neural network (GRNN), the prediction accuracy could reach 93.3%.