Abstract: During the process of complex terrain surveying, the sparse distribution of beams in multibeam echosounders (MBES) will result in decreased measurement accuracy and loss of detailed information. Traditional beam interpolation methods can improve measurement accuracy, but they require manual identification for encryption areas, which is challenging for continuous surveying operations in large sea areas. In this article, a wavelet transform adaptive method to increase beam density for multi-beam sonar is proposed. The method utilizes a single-chip heterogeneous processor scheme consisting of FPGA+ARM. It performs wavelet analysis on the previous frame’s terrain data at the ARM end to obtain the pre interpolated area for the next frame. Subsequently, local beam encryption is conducted on this pre interpolated area using FPGA. Experimental data from tanks and natural water bodies demonstrate that this approach enables adaptive high-precision measurements of complex underwater terrains. The method holds promising applications in underwater detection and resource exploration.