To reduce the influence of factitious factors on triangle orientation discrimination (TOD) threshold measurement and so enhance the measurement efficiency, a machine vision system to perform TOD instead of human observers was studied, and an objective discrimination model (ODM) for TOD was developed. Based on the main effect and course of signal processing of human visual system (HVS), ODM uses oriented filters to extract essential information from image-signal and discriminate the orientation of a triangle by comparing the visual character vectors with standard ones. With the triangle images generated by the virtual staring imaging system, experiments were held to compare the performance of ODM with human observers' one. Three groups of experimental results of different triangles size were given and the results show that ODM can predict the observers' performance well.