The purpose of this study is to verify the feasibility of feature tracking method on the images taken by a digital camera near the coastal area. This technology is applied to the surface circulation at Yin-Yang Bay, a small bay at northeastern Taiwan, with an obvious front of cloudy yellow fresh water and clear brine. The primary images taken at Yin-Yang Bay have to do geometric transformation, image smoothing and edge detection before work on feature tracking. Two feature tracking methods are applied to these processed images for estimating surface current velocity. They are the subjective method of estimating distance among the sequence images by naked eyes and the objective method of Maximum Cross Correlation (MCC), respectively. The results show that the velocities of surface currents at Yin-Yang Bay derived from both methods are similar and are comparable to the results of in-situ measurements with a difference of 2%. This indicates that the technology developed in this study is suitable for coastal circulation research. However, it is noted that not every front velocity can reflect the behavior of tidal current. Nevertheless, the front velocity during neap tide is apparently slower than that during spring tide.