Progress beyond the state of the art

This project builds on and will progress beyond different projects:

In the EMPIR normative joint research project 19NRM02 RevStdLED, the traceability of luminance imaging measurement devices is addressed. The general model of evaluation and uncertainty budget for luminance measurements will be picked up in this project as a foundation. However, RevStdLED aims to reduce its complexity for selected individual applications relevant to test laboratories, and namely does not consider RGB matrix cameras nor HDR imaging systems for which the uncertainty contributions will have different significances.

In the EMPIR joint research project 18SIB03 BxDiff, HDR imaging systems are used for studying the relation of reflectance measurements at different scales, on texturized, structured and translucent materials. However, BxDiff does not address the standardisation of an HDR algorithm nor the development of a reference source with luminance contrast.

In the EMPIR normative joint research project 20NRM01 MetTLM, the research regarding imaging devices focusses on spatial resolved temporal light modulation (TLM) measurement modes of luminaires and extended scenes, and vivid examples demonstrating its feasibility using ILMDs and RGB matrix cameras. In this context, high contrast luminance scenes or sources will be included in the targeted measurements. This project will gain from the experience gained in MetTLM and will collaborate on the HDR measurement of TLM sources.

High contrast reference standard source of at least 6 orders of magnitude (objective 1)

Current state of the art

Single test charts have been usually used to assess the performance of imaging measurement systems for detecting object contrasts. However, these charts do provide only a limited number of data points. Existing luminance standard sources can offer various luminance levels but generally without the potential of defined multiple luminous surface elements operating at different calibrated luminance level simultaneously. Light sources with high luminance contrast do exist, but they are usually intended to test the HDR performance of photography camera, machine vision systems and automotive cameras. For instance, a luminance source has been recently designed and tested to recreate dynamic contrast scenes in a test lab, similar to those typically observed by automotive cameras [11]. It is claimed that this source provides a complete luminance contrast of 6 orders of magnitude, with a stated preliminary luminance range of 0.06 cd/m² – 60 kcd/m². However, this system seems to be dedicated to automotive cameras, and does not allow the characterisation of HDR imaging measurement systems with very wide field of view, as, for instance, those required when assessing glare with a fisheye lens. In addition, this previous and very recent source is not designed to offer traceable calibration of luminance measurement systems, where the luminance of each consisting light source must be accurately determined in each operating condition.

Progress beyond the state of the art

In this project, a high contrast reference standard source of at least 6 orders of magnitude of luminance will be developed, which will include several luminance levels, covering the range 0.1 cd/m² to about 100 kcd/m² or more and an additional light trap of <0.01cd/m². The luminance of the consisting sources will be determined with an expanded uncertainty no larger than 1 % for the brightest source and no larger than 2 % for the dimmest one. This standard source will cover the needs for testing and characterising measurement systems with not only narrow, but also wide measurement fields up to 180º via a modular design concept. It will be designed and developed to meet the requirements of luminance dynamic range for applications and measurement needs described in the relevant standards and documents, like CIE 232:2019 [1], CIE 150:2017 [5].

Validation of HDR luminance measurements and demonstration of the inter-comparability of HDR luminance measurements (objective 2)

Current state of the art

Currently, due to the use of HDR algorithms that do not provide traceability and lack uncertainty evaluations, the validation or equivalence of measurement data cannot be judged. This is particularly true for the field of glare assessment, where many self-made procedures are in use, with a lack of proper adjustment and calibration of the devices, leading to results that sometimes cause difficultly in interpretation and in worst case are unreliable. This creates a significant lack of trustable and repeatable field measurements of glare and obtrusive light which can lead to safety risks and visual discomfort.

Progress beyond the state of the art

To overcome the limitation of untraceable measurements, a characterisation procedure will be developed using the reference standard source and additional sources (e.g. conventional luminance standards and small sources such as bare LEDs with a higher luminance (several Mcd/m²) that are stable during the measurement sequence). The characterisation procedure will be based on the specific need of the HDR imaging process and the investigated technologies. This will allow the characterisation of the HDR imaging measurement systems and consequently, the ability of a traceable validation of the glare and obtrusive light in complex outdoor scenes as well as from daylight. In addition, the metrological comparability of results obtained with different HDR imaging technologies will be demonstrated. At least three types of HDR imaging measurement systems will be tested (ILMDs, commercial DSLR cameras and industrial/scientific cameras based on RGB matrix sensors). The comparability of the measurements will be investigated though laboratory and field tests using the characterised systems. For this purpose, the developed luminance standards will be used as well-defined references.

Harmonised HDR algorithm for traceable HDR luminance measurements (objective 3)

Current state of the art

The recent development of algorithms used for the generation of HDR images has been strongly motivated by applications in computer graphics and digital photography. Many consumer-grade or professional cameras (e.g. photography cameras and smartphones) offer integrated features for HDR image acquisition and processing. Many HDR algorithms aim at facilitating image acquisition (e.g. automatic calibration, image alignment, white balance, etc), at offering high quality images without the need for an expensive imaging device (e.g., smartphone cameras) and at obtaining the best images from a visual and artistic perspective. Vision systems with HDR functionality are also widely used in automotive, aerospace and industrial applications (e.g. production, inspection, etc). In these cases, the HDR algorithms aim to enhance specific information (e.g. geometrical information about size and shape, object or scene outlines, etc) to the intended purpose. However, the captured and processed images generally do not yield to quantitative luminance measurements. HDR imaging for luminance measurement of daylight and outdoor night scenes for the evaluation of glare and obtrusive light assessment cannot benefit from such processing features, as the measurement accuracy, reliability and traceability of the process matter the most. Furthermore, the HDR algorithms associated with ILMDs, which provide luminance images, are generally either proprietary algorithms or lack documentation and are not compatible with SI traceability. Most of these algorithms perform destructive process in images which breaks the virtual chain between the source images and the final HDR product. Therefore, an HDR algorithm that allows for the propagation of measurement uncertainties and provides traceable results does not yet exist.

Progress beyond the state of the art

This project will go beyond the state of the art by developing a dedicated HDR algorithm which will include all the functionalities that are missing from existing algorithms and will serve the metrological needs of HDR imaging measurements, including the propagation of uncertainties. The aim is to design an algorithm with a metrological orientation dedicated to quantitative measurements rather that image beautification or vision applications. The HDR algorithm will process the source images (LDR ones) and generate a final HDR image that will be accompanied by the measurement uncertainty. It will be implemented in source code that will be distributed under an open-source license in an open access repository for the wider possible adoption by the stakeholders. This source code will be accompanied with a user guide and exemplary datasets that will be generated for this purpose during the project.

Uncertainty estimation of HDR luminance measurements, and propagation to glare and obtrusive light assessment (first part of objective 4)

Current state of the art

To ensure comparability and reliability of HDR luminance measurements, and assessment of glare and obtrusive light, the uncertainty associated to measurements must be provided. In addition, this is crucial for compliance with standards and requirements. At present, there are no recommendations nor guidance for the determination of uncertainty budgets of HDR luminance measurements and for the propagation of uncertainty in the assessment of glare and obtrusive light.

Progress beyond the state of the art

One of the main targets of this project is to develop a model of the uncertainty propagation of luminance measurements and of glare and obtrusive light evaluation using HDR imaging systems. The model for uncertainty propagation of HDR luminance measurements will be developed using data from the characterisation of the investigated HDR imaging technologies and will be directly implemented with the algorithm developed for HDR processing. It will be demonstrated with the newly developed reference standard source, and field measurements. The part of the model regarding the uncertainty propagation in glare and obtrusive light assessment will be validated using dedicated measurements in well documented lighting installations. Given that there are no such models yet, the targeted uncertainty cannot be estimated, but a key outcome of this project will be the ability to perform an uncertainty estimation for such measurements and assessments for the first time.

Report on the relevance of existing quality indices and test methods regarding HDR imaging luminance systems (second part of objective 4)

Current state of the art

The characterisation of ILMDs is described in CIE 244:2021 “Characterisation of Imaging Luminance Measurement Devices” [6]. The quality indices, which are the most relevant for HDR luminance measurements are the i) Effect of surrounding field (f₂₃), ii) Stray light influence for negative contrast (f₂₄), iii) Edge function (f₂₅), iv) Influence of smear (f₂₆), v) Size-of-Source Effect (f₂₉), vi) Linearity using a fix measurement range (f_3,1), and vii) Linearity using range change (f_3,2)]. In addition, the “Contrast detection probability” index (CDP) [11], exclusively designed to assess the performance of automotive cameras to detect the object contrast in its field of view, was recently proposed.

This CIE publication provides test methods for determining these indices, but does not provide any indication on the impact those effects have on HDR luminance measurements.

Progress beyond the state of the art

In this project, the above-mentioned effects will be considered for the different integration times because different parts of the image will be selected by the HDR processing. The impact on the uncertainty evaluation will therefore be dependent on the HDR algorithm. This project will evaluate the relevance of the existing indices regarding HDR imaging luminance systems and determine how the related effects impact the uncertainty of the measurements for HDR luminance measurements. This project will also evaluate if there is a need for other quality indices specific to HDR, and aspects of potential new quality indices will be reported.