1 September 2016 in International Journal of Thermal Sciences
This articles investigates the temperature errors due to chromatic aberration in multiwavelength thermography methods. The Chromatic aberration leads to a shift in the perspective projection of a point in the 3D space on the image formed at different wavelengths. This shift causes an error in the temperature field calculated by polychromatic methods, from the fusion for each pixel of radiance temperature images at different wavelengths. The temperature error can reach 40% on a sample with high spatial non-uniformities, due to wide variations of emissivity. This paper suggests an approach to correcting the chromatic aberration that is based not on equipment but on software, coupled with a calibration, using Digital Image Correlation (DIC). This experimental technique is a 2D optical method used in mechanical engineering, for inferring a deformation of a plane structure's surface from a displacement field calculated by correlation between pixels of two successive images. This paper applies the technique to achieve the field displacement induced between two images at two wavelengths by chromatic aberration. After applying this correcting displacement field, the temperature error decreases from 40% to 1% for the pixels located at the boundary of two areas with different emissivities.
22 October 2026 in SSRN Electronic Journal
Photometric Stereo (PS) recovers the surface normal and the albedo of an object from several images acquired from a single viewpoint under varying illumination. In its usual form, the unknown intensity of each light source is absorbed into the estimated albedo, which is therefore recovered only up to an unknown scale factor and cannot be interpreted as a physical reflectance. This paper turns PS into a quantitative, spatially resolved reflectance measurement method. The intensity of each source is calibrated against a Spectralon® white diffuse reflectance standard of certified directional-hemispherical reflectance, independently at each pixel and each wavelength, so that the calibration coefficient also accounts for the photo-response non-uniformity of the sensor and for the spectral response of the whole imaging chain. PS is then performed with the Energy-preserving Oren–Nayar (EON) diffuse reflectance model which, unlike the Lambertian model and earlier Oren–Nayar variants, conserves energy and therefore delivers a physically meaningful albedo together with the surface normals and a diffuse roughness parameter. The method is validated on samples of increasing complexity: Spectralon® standards of nominal reflectance 99 %, 50 % and 20 %, flat samples coated with white and gray diffuse paints, spherical samples coated with the same paints, and a 3D printed replica of the ”Bear” object of the DiLiGenT dataset. The estimated albedo is compared, at 450 nm, 630 nm and 780 nm, with measurements from previous studies and with reference measurements performed with a spectrometer equipped with an integrating sphere.