Systematic depth error modeling in range measurements on PMD CamCube 3.0

Range cameras are devices to measure the distance between the sensor and the object surface for each pixel within an image. Comparing to laser scanner they have the advantage of measuring range for several points at the same time without scanning systems. The measurements obtained by a range camera may have systematic errors that must be modeled to correct the measurements. Some factors as the integration time used, the distance to be measured, and the illumination of the scene, may affect the range measurement. In this paper we examine the influence of the integration time and the distance between the camera and the targets on the measured distance, proposing a model for the cyclic error of a Camcube PMD 3.0 camera using the Discrete Fourier Transform. The model allowed to reduce the root mean square error (RMSE) from 15.01 cm to 5.05 cm in observations made with the integration time of 4000 (s. It was also verified that the error amplitude is proportional to the integration time.

Depth Error; Range Camera; Integration Time; Fourier Analysis