Need for the normative joint research project
High-level needs
The complexity of the human visual system allows its adaptation to extremely dark and bright lighting conditions. Due to its very large dynamic range for lightness perception (11 orders of magnitude of luminance), human can safely and comfortably navigate the world (e.g., driving, working under daylight conditions, welding, etc) and perform tasks involving vision in lighting environments with very high luminance contrast. However, some lighting environments, e.g. office or workshop workplace installations and daylight scenes, can be disturbing for some tasks, and it is convenient to adapt them for more adequate lighting. There are several other social needs that present high contrast scenes containing light sources with very different luminance levels compared to a much lower luminance of the background. As expressed in the 2019 revision of the EU’s Green Public Procurement Criteria for Road Lighting and Traffic Signals [7], obtrusive light is an important issue for wildlife (high insect mortality, disruption of the migration of birds) and human quality of life (sleep pattern disruption), as well as for astronomical observations. Other applications such as machine vision in the manufacturing process, vision for autonomous driving under daylight or night-time conditions and material appearance measurement would also benefit from the capability of reliably measuring wide ranges of luminance contrast. All those situations require measuring instruments specifically designed to perform well on such extreme luminance and contrast scenes (luminance of glare sources at a few Mcd/m² beside a background of several mcd/m²) and to deliver traceable results.
Specific needs
The measurement of high contrast sources is in general challenging, as the most adequate acquisition conditions for measuring the luminance of a light source vary with the brightness of each source. Whereas the brightest sources in the scene require, for example, a short integration time to avoid saturation of the corresponding pixels, the dimmest ones need a long integration time to increase the signal well-above the readout noise of the pixel matrix imaging sensor. The widely used solution is to capture several images at different integration times so that the full luminance range of the scene is correctly acquired by all each pixel in at least one of these images. It implies that the measurement no longer relies on a single acquisition but on several acquisitions. An algorithm must be implemented to generate the final spatial luminance distribution of the scene from the measurements performed with different integration times. This algorithm, which enhances the measuring system by enlarging the dynamic range with respect to that of a single image, is usually known as an HDR algorithm. However, from a metrological point of view, it is important to consider the implications of combining different images acquired with different integration times. There is currently no guideline regarding the calibration of HDR imaging measurement systems for traceable luminance measurements. Therefore, the results of different users, different algorithms implemented in the software, different lenses and different ILMDs lead to different glare ratings for the same luminance distribution. The average luminance for small glare sources can vary up to 3 times [8]. The consequence is that it is impractical to perform field measurements in order to check whether a new lighting equipment fulfil the requirements regarding glare [9]. Only when residents complain about glare or obtrusive light from street lighting into their workplaces or rooms do manufacturers become active and equip the luminaire with simple shields. In the worst case, these shields influence the light distribution and affects the illumination on the road, which frequently leads to reduced visibility and thus to a decrease in traffic safety. It is also difficult for manufacturers to develop glare-reduced luminaires because they cannot clearly measure and compare the performance of their products in the field. Accordingly, reliable and accurate measurements are needed for the development and optical design of luminaires, to judge the conformity of individual lighting installations and to enable potential legislative activities. The absence of metrologically-supported standards for glare and obtrusive light evaluation using HDR imaging luminance measurement systems might lead to major shortcomings in safety and comfort for many visual activities.
Due to the different needs stated earlier, CIE requested EURAMET to prioritise the research effort regarding HDR luminance imaging by identifying a “lack of traceable SI calibration, poor long-term stability, and inadequate relative spectral responsivity” and “the calibration and characterisation of HDR-cameras used for luminance distribution measurements and glare evaluation”. These issues had also been stated by the CIE as topics of research priority.
Many stakeholders’ sectors need reliable recommendations for traceable HDR imaging luminance measurements. Amongst them are standardisation bodies such as the CIE and CEN/TC-169, NMIs and DIs, HDR luminance imaging devices manufacturers, scientific community working on glare, obtrusive light, and any stakeholders working on applications requiring HDR imaging luminance evaluation (sparkle measurements, gloss measurements), as well as all end-users of HDR imaging systems dealing with quantitative glare and obtrusive light evaluation, and communities impacted by obtrusive light, glare and high‑contrast luminance scenes.
Objectives related to the needs
HDR systems will be subject to the same sources of uncertainty and error as ILMDs, but due to the HDR‑algorithm, the impact on the result is different. CIE 237:2020 “Non-Linearity of Optical Detector Systems” [10] describes CCD-based sensors with low dynamic range, and CIE 244:2021 “Characterisation of Imaging Luminance Measurement Devices” [6] describes the characteristic of ILMDs but without specific consideration of an HDR mode, and, therefore, a more detailed consideration must be given for updating it or to cover such aspects in another CIE publication (objective 4). Identified sources of uncertainty include measurement range dependent non-linearity, spectral dependent non-linearity, stray-light (diffraction/flare, ghosts, scattering), spectral mismatch, local tone mapping. These contributions need to be modelled and quantified (objective 2 and 4) and their effects should be traceable through the algorithm. Most of the algorithms currently used do not provide this possibility, as the uncertainty evaluation of the measurements is not included. The commercial devices that propose an HDR algorithm are usually proprietary. A unified and open access algorithm (objective 3) including the implementation of uncertainty analysis (objective 4) therefore needs to be developed. A reference standard source needs to be developed (objective 1), allowing the characterisation of different HDR imaging systems.
Due to the variety of expertise needed (instrumentation, HDR algorithm implementation, uncertainty evaluation) and the different applications (glare and obtrusive light), this project would be very difficult to realise without the contribution from experts from different fields and is therefore very well suited to be conducted as a European project.