An essential requirement for efficient heavy oil production is liberation of heavy oil from host rocks, which is determined by the wettability of the rocks, and the interfacial tension and viscosity of heavy oil. Recent progress on the design of an online visualization flow cell allows capture of dynamic heavy oil liberation processes from the surfaces of sand grains in real time under a water flooding environment. However, the accurate assessment of heavy oil liberation remains a challenge, due to uncertainties in defining oil-free areas of heavy oil contaminated rock surfaces. In this study, three new image-processing algorithms of modified empirical, Gridding, and Edge-covering methods were applied to image transformation for heavy oil liberation analysis. These methods were found to be more accurate and robust in determining the threshold value distinguishing liberated from unliberated sand surfaces. The use of wavelet transform theory in the Gridding and Edge-covering methods led to faster calculations with a typical error of less than 2 % in the quantitative analysis on the threshold value determination and the degree of heavy oil liberation. Among these three methods, the Gridding method with a sound theoretical foundation was shown to be the most reliable. The results showed that the threshold value determined was highly dependent on the types of ores and the image capture settings such as lighting conditions, exposure time, and microscope magnification.