Total focusing method(TFM)based ultrasound array is considered as the “gold standard” for the detection of internal defects, such as voids, impurities and cracks in solid. And the TFM method is also potential for high-precision imaging of internal void defects in concrete. However, the fact that concrete structure contains aggregates, gravel and steel bars, results in the significant structural noise when ultrasound propagates in concrete, and brings about the reduction of internal void defects image quality. A numerical concrete model based on random aggregate model is established, and ultrasound propagation in concrete is simulate in order to research the relation between the aggregates distribution and structural noise. A process of convolution filtering for the received signal is carried out to suppress the structural noise in advance. Then the full matrix capture ultrasound data are processed for imaging according to the total focusing algorithm, and the model detection result is displayed as color map. Finally, the quantitative relation between the values in color map and the internal defect location is discussed. The results show that the imaging signal to noise ratio can be effectively improved, and the image regions with higher value indicate more accurate internal defect locations.