Abstract: The compound imaging method uses the direct superposition of multiple emission imaging results for imaging, which is simple to calculate but has poor imaging quality. The adaptive weighting method based on coherence factor calculates the weighting factor according to the global characteristics of the imaging vectors at all emission angles, which leads to the degradation of the background imaging quality. The paper proposes a magnitude squared coherence factor (MSF) weighting method. This method determines the weighting coefficient based on the coherence between spectrum of the imaging vectors of adjacent emission angles, and performs weighted output on the compound imaging results. The results of the body-mode experiments show that the MSF method results in smoother background and improves the noise contrast and speckle signal-to-noise ratio compared to the conventional weighting methods of coherence coefficient, general coherence factor, and circular statistics vector, in which the speckle signal-to-noise ratio is improved by 91.6%, 45.9%, and 52.3%, respectively. The contrast ratio was improved by 69.8% relative to the unweighted method. The human experimental results further demonstrate the effectiveness of the proposed method. The human experimental results further demonstrate the effectiveness of the proposed method, in which the contrast radio was improved by 15.8%, 9.3%, and 26.4%, the contrast-to-noise ratio by 36.3%, 24.8%, and 32.4%, and the speckle signal-to-noise ratio by 80%, 66.3%, and 44.3%, respectively.