Significance and attributes of subpixel estimation on area‐based matching

Area‐based matching and subpixel displacement estimation using similarity measures are common methods used in various fields. Subpixel estimation using parabola fitting over three points with their similarity measures is also a common method to increase matching resolution. However, such estimation can contain systematic error depending on image characteristics, the similarity function, and the fitting function used for subpixel estimation. In this paper, characteristics of subpixel estimation error were clarified using a simple analytical model. Although this model is quite simple, it allows us to make a theoretical analysis for general images without assuming any specific image pattern. Through this analysis, it has been shown that there are correct combinations of the similarity functions and the fitting functions. Also, the so‐called “pixel‐locking” effect, by which estimated positions tend to be biased toward integer values, has been explained. Finally, analyses were verified with an experiment using real images.