Total Focusing Method (TFM) for ultrasonic imaging, a novel ultrasonic total focus method-intelligent partitioning based on target elements (TFM-IPTE) is proposed. The TFM-IPTE first pre-locates suspected defect areas by analyzing the transmitted and received signals of the main elements. Subsequently, square, horizontal, and vertical division strategies are applied to the full matrix capture (FMC) data, and efficient focused imaging is achieved by combining these divided submatrices with a phase shift algorithm. FMC data were collected through ultrasonic total focusing detection on two carbon fiber composite test blocks containing subsurface layered defects with different depths. Imaging calculations were conducted on the full matrix data and the divided submatrix data respectively, and the imaging effects of the full matrix data were compared with those of the submatrix data obtained by three different division methods. Experimental results show that compared with the traditional total focusing imaging using full matrix data, the TFM-IPTE significantly reduces the computation load and thereby improves imaging efficiency. Among the three division methods, vertical division provides superior performance when detecting shallow, middle, and deep defects. While maintaining imaging resolution and side lobe suppression effect, the relative error of horizontal position prediction for defects are reduced by 7.995 2% and 7.633 4%, 2.603 0% and 2.447 9%, 0.595 2% and 0.496 5% compared with square division and horizontal division respectively. The TFM-IPTE effectively balances imaging efficiency and quality through the data division strategy, providing an efficient solution for automated ultrasonic non-destructive testing (NDT) in fields such as aerospace and nuclear power.