Abstract: In order to detect the flow state and concentration distribution of gas/liquid two-phase flow in the process industry, ultrasonic tomography technique was used to simulate gas/liquid two-phase flow. Considering the influence of the number of sensors on the quality of image reconstruction, a contrastive study of sensors number was completed at first. Results show that the more the number of sensors, the better the quality of image reconstruction is. Then, the linear back projection (LBP) algorithm was used to reconstruct different flow patterns and the adaptive threshold filtering (ATF) algorithm was introduced to improve reconstruction images. Finally, the correlation coefficients between the binarized images and the standard images after threshold processing were calculated. Results show that LBP algorithm can effectively reconstruct the image of various flow patterns, and the ATF algorithm can significantly improve the imaging quality.