Current research of void defects testing and identification for carbon fiber reinforced plastic (CFRP) mainly focuses on sheet-shaped CFRP, not thick-section CFRP. To fill this need, this paper proposes a void defects identification method for thick-section CFRP based on ultrasonic pulse echo backscattered signal processing. The analysis results of backscattered signal characteristics show that the backscattered signal is composed of resonance structure noise, the exponential decay signal, the reflection and scattering signal caused by voids, and stochastic noise. First, in order to obtain the reflection and scattering signal caused by voids, the stochastic noise is removed by lifting wavelet transform, which has good denoising ability. Then, the low pass filter and adaptive filter are designed to remove the influence caused by, respectively, resonance structure noise and the decay signal. The processing results of the experimental signal show that the above processing method can effectively remove corresponding signal components. Based on these signal processing methods, the C-scan imaging of the ultrasonic backscattered signal amplitude is designed. The experimental results show that regions containing voids in the CFRP workpiece can be determined effectively when the voids recognition method is used to test a thick-section CFRP workpiece.