Cement mixing techniques are widely used to improve the mechanical properties of weak soils in geotechnical engineering. However, due to the influence of various factors such as material properties, mixing conditions, and curing conditions, cement-mixed soil exhibits pronounced spatial variability which is greater than that of natural soil deposits, introducing additional uncertainty into the measurement and evaluation of its unconfined compressive strength. The purpose of this study is to propose a novel framework that integrates image analysis with Bayesian approach to evaluate the unconfined compressive strength of cement-mixed soil. A portable scanner is used to capture high-quality digital images of cement-mixed soil specimens. Mixing Index (MI) is defined to effectively evaluate mixing quality of specimens. An equation describing the relationship between water cement ratio (W/C) and unconfined compressive strength (qu) is introduced to estimate the strength of uniform specimens. To estimate the strength of non-uniform specimens, the equation is developed by integrating MI with the strength of uniform specimens. The coefficients of equations are obtained using Bayesian approach and Markov Chain Monte Carlo (MCMC) method, which effectively estimating the strength of both uniform and non-uniform specimens, with coefficients of determination (R2) of 0.9858 and 0.8745, respectively. For each specimen, a distribution of estimated strength can be obtained rather than a single fixed estimate, providing a more comprehensive understanding of the variability in strength. Bayesian approach robustly quantifies uncertainties, while image analysis serves as a convenient and non-destructive method for strength evaluation, providing accurate method for optimizing the mechanical properties of cement-mixed soil.