Photoacoustic tomography is a technology that reconstructs the distribution of light energy in tissue through detecting photoacoustic signals. In recent years, the field of research has been greatly developed and widely used in anatomy, functional science and molecular imaging. However, one of the great challenges is that the efficiency of light to sound conversion is very low due to photoacoustic effect, resulting in a low signal-to-noise ratio (SNR) of photoacoustic signal, and the quality of reconstructed photoacoustic image is not high. Conventional approach to enhance the SNR of photoacoustic signal is the data averaging method, but severely limits the imaging speed. Without sacrificing signal fidelity and imaging speed, firstly, uses empirical mode decomposition (EMD) to realize adaptive decomposition of photoacoustic signals. Then uses conditional mutual information as criterion to determine intrinsic mode function (IMF) which needs noise reduction, and then de-noises the selected intrinsic mode functions to obtain the de-noising photoacoustic signal. Finally, the de-noised photoacoustic image is obtained by using the reconstruction algorithm. The simulation and experimental results show that the proposed method, which combines empirical mode decomposition with conditional mutual information, can achieve better improvement of signal-to-noise ratio of photoacoustic signals and the contrast of reconstructed images than the traditional methods. The effectiveness of the de-noising algorithm is proved. At the same time, this method provides a possibility of development a real-time low-cost PA imaging system with low power laser source and low power amplification signal SNR.