Aiming at the problem of how radar implements intelligent anti-deception jamming, a radar intelligent anti-deception jamming method based on the time-frequency image chromaticity difference and the deep random forest is proposed. Firstly, perform the time-frequency transformation on the radar received signal, and after gray-scale processing, convert the RGB three-dimensional data information into one-dimensional information, and use the global threshold to segment the original time-frequency imag. Secondly, extract the color component features of the segmented image, then use deep random forest (DRF) for interference identification. Finally, design a two-dimensional time-frequency filter based on the identification result and the location information of the real signal to filter out interference and noise. The algorithm does not need to estimate the signal parameters, and the computational complexity is low. The simulation results show that the interference recognition rate can reach more than 90% when the jamming signal ratio (JSR) is greater than 3 dB and less than -2 dB, and when the jamming is equivalent to the signal power, the recognition rate can be further improved by adding features such as box dimension and information entropy.