OTDS (optical triangulation displacement sensors) have been widely used in industrial measurement applications because they can perform a non-contact measurement with sub-micron resolution. Moreover, they are reasonably fast and have a simple structure, good repeatability, and a long operation range. OTDS convert the change of a spot position on the detector into object displacement on the assumption that the spot is a perfect point. However, actually, there exists a light intensity distribution on the detector. Various factors such as inclinations of the object surface, speckle effects, optical power fluctuations of the light source, and noises in the detector change properties of the light intensity distribution. One easy way to make the light intensity distribution robust against the error sources is to use an incoherent LED as the light source and a linear CCD (charge coupled device) as the detector. In such a system, a peak position of the light intensity distribution is exactly proportional to the displacement of the object even if there are various error sources. Therefore, it is possible to minimize errors of OTDS. However, adopting the CCD inevitably causes problems; the resolution of the sensor system is limited by the pixel size of the CCD and there must be large amounts of calculation because output data from the CCD are discrete. To overcome these problems, a signal processing system for error-minimized OTDS is proposed.