Abstract: An acoustic tomography temperature measurement based on grid non-uniform discretization was proposed, and discretization factors were defined to quantify the sparse and dense distribution characteristics of grid cells in space. After which, Green's biharmonic interpolation algorithm was introduced, while comparisons and analyses were conducted on the reconstruction performance of various discretization methods, including uniform discretization, non-uniform discretization based on different discretization factor settings, and odd/even-numbered grid divisions. Results show that by non-uniform discretization with fewer measurement points, the accuracy of reconstruction results can be improved, and the peak value of temperature field can be reconstructed more accurately. In addition, when reconstructing the single-peak symmetric temperature field, odd-numbered grid discretization is prior to even-numbered grid discretization.