Polarization navigation is a promising orientation-determination method inspired by insects' foraging behavior that offers the advantages of autonomous and high precision. In this paper, using the solar meridian as an azimuth reference is proposed. The model of the distribution pattern of the polarized skylight projected onto an imaging sensor is analyzed. The sufficient features of the solar meridian are proven. According to these features, an angle algorithm for an imaging polarization navigation sensor based on a machine-vision algorithm is proposed. In consideration of noise in images, the relation between the measured angle and the noise in images is modeled. This model cannot only optimize the threshold tolerance R in the algorithm but also describe the effects of several primary factors that can affect the measuring precision. In the simulation test, the measurement accuracy was better than 0.34°. When the algorithm was tested on the polarization-detection system, the measurement accuracy was better than 0.37°.