This paper deals with direct comparison of traditional pixel and chosen subpixel corner detection approaches. For this reason the Harris corner detector and two subpixel algorithms used to improve the detection accuracy were considered. First one is based on the orthogonal vectors theory and iteration process and the other one is refining the found pixel coordinates by fitting quadratic curve to the cornerness map in both x and y directions separately. As the procedure to verify the performance of these approaches the real displacement determination between two positions of the same camera using information coming from image stereo pair was chosen. To fulfill this task requires employing the several computer vision and image processing areas which are described in presented paper. The appropriate sets of image pairs were created and used for further testing and comparison. There were also the statistical analysis and evaluation performed and the obtained results were listed in corresponding tables.