The effects of camera focus and sensor exposure on accuracy using computer vision

Abstract

As industrial camera become smarter, due to the increase in supporting algorithms, so the quality of data produced offers higher precision and accuracy of industrial inspection operations. This paper aims to focus on the accurate effect of camera focus on a calibration process under the influence of different sensor exposure and degrees of focus. A computer vision methodology is used to determine the effects of different sensor exposures and at different degrees of camera focus. Endocentric and telecentric lenses are used to acquire images, and a comparative analysis was achieved using the least square method. A sample of 2176 images was used to generate the population for analysis. The analysis showed that the human eye cannot visually determine a hundred percent focus of an image. This means that it is difficult to determine when an image is 100% focused from when it is 95% focused. The result from the experiment showed that as the camera focus tends towards 100% the calibration error decreases, with a minimum calibration error of 0.06 pixels, leading to the conclusion that the developed method achieved a calibration error accuracy with the most effective camera sensor exposure and the camera focus.