52
J. Eur. Opt. Society-Rapid Publ. 21, 36( 2025)
limited to about 15 % of the sRGB space and is highly dependent on the viewing angle. Similarly, Marek Mezera et al.( 2023) [ 3 ] discussed how laser-induced periodic surface structures( LIPSS) can generate diffractive or plasmonic structural colors, but these methods struggle to reproduce a wide range of hues with high saturation.
Other studies further highlight these limitations. J. Bonse and S. Gräf( 2021) [ 4 ] outlined several unresolved challenges in the fabrication of LIPSS, emphasizing the constraints in achieving a broad color range and practical scalability. S. Maragkaki et al.( 2020) [ 5 ] investigated the influence of defects in LIPSS-generated structural colors, and found that such defects can cause color inconsistencies and further limit the achievable color gamut. In addition, S. Zhang et al.( 2023) [ 6 ] investigated laser-written multilayer structural colors for full-color marking, but found that such techniques suffer from viewing angle limitations, making color perception inconsistent depending on the observer’ s perspective.
As a result, these techniques cannot replicate the full range of colors accessible through standard printing techniques such as inkjet printing, where a wide color gamut is achieved using saturated cyan, yellow, magenta, and black primary inks [ 7 ].
Since the reference image often contains colors outside the achievable gamut, a gamut mapping step is necessary to adapt the printer reproduction to the constraints of the distorted gamuts. The gamut mapping process in this study is primarily based on the method proposed by Chosson and Hersch [ 8 ]. In this approach, primary colors – which include black, white and chromatic primaries – are first selected to define the target color gamut for the printing process. To account for differences in viewing conditions and ensure color consistency, a chromatic adaptation transform [ 9 ] is used. To preserve perceptual differences between light and dark areas, a lightness soft-compression is applied in [ 10 ], gently adjusting values to fit within the achievable range of our output medium. Gray axis alignment ensures that neutral tones remain consistent, preventing unwanted color casts in achromatic regions. Finally, chroma softcompression is performed to maintain color saturation while keeping colors within the target gamut boundaries.
This structured approach enables effective gamut mapping, ensuring that source colors are accurately reproduced while maintaining visual consistency.
Due to the limited gamut of the laser-induced colors, traditional image quality metrics such as resolution, blur, and other Image Quality Assessment( IQA) criteria may prove inadequate for optimizing the laser printing process for color reproduction [ 11 ]. Several other metrics are commonly used in color reproduction studies that cover a range of characteristics. These include color [ 12 – 15 ], hue and chroma as defined by the CIE [ 16, 17 ], saturation [ 13, 15 ], color rendition [ 12, 18 ], processed color gamut [ 12 ], color reproduction [ 19 ], color shift [ 19, 20 ], gamut size [ 21 ], correctness of hue [ 22 ], and the proportionality of printed gamuts to their original counterpart [ 14 ]. In addition, perceptually driven metrics such as the Spatial Hue Angle MEtric( SHAME) [ 23 ] account for both spatial and color distortions, making them particularly relevant when evaluating color differences in constrained gamuts. However, the quality of the reproduction is markedly different when using the laser-induced gamut, which is considerably smaller and not centered around a grey axis in comparison to the typical printer gamuts referenced in the literature.
The aim of this study is to identify appropriate objective metrics for assessing the image quality of color images reproduced by laser processing with a limited color gamut. A one-way Analysis of Variance( ANOVA) [ 24, 25 ] method was conducted in conjunction with an Akaike Information Criterion( AIC) model selection test [ 26, 27 ] to determine which objective metrics are most appropriate for estimating the subjective IQA. These metrics will guide the optimization of laser parameters to enhance image quality or to compare color gamuts produced by various materials or laser parameter sets.
Given the large number of possible laser parameter combinations and the inherent difficulty in predicting their effect on color, an empirical approach is required. This approach involves laser printing millions of metasurfaces using a wide range of laser parameters and then measuring them in different observation modes. Once the colors have been measured, they are used to simulate images with different laser-induced gamuts corresponding to different types of initial random plasmonic metasurfaces and different observation modes. The challenge of ensuring optimal observation modes for each observer on different real samples for classification purposes led us to conduct a survey on digital images. The simulated images were evaluated on a monitor by human observers, who reported normal or corrected-to-normal vision in an uncontrolled viewing environment on their personal computer, using comparison rather than absolute color evaluation to mitigate the effect of viewing condition variability. The observers were asked to classify the images, which had been simulated with different gamuts being given the reference of the image obtained with a sRGB gamut. They were asked to do this based on how faithful to the original the reproductions appeared to them.
In Section 2, we introduce the foundations of image quality analysis, define image quality, review existing metrics, and present our approach for extracting a final metric from established methods. Section 3 details the subjective image quality assessment( IQA) conducted, including the survey setup, gamuts and images used. In Section 4, we present the results of our analysis and propose a new objective metric aligned with the subjective ratings. Finally, the conclusion discusses the strengths and limitations of our approach, along with perspectives for future improvements and applications.
2 Material and methods
Image quality can be defined as the assessment by human observers of the visual elements that make an image pleasingtothem [ 28 ]. The main factors that affect image quality are noise, blur, and color reproduction.
Since images are one of the most common methods of conveying information, whether in digital or physical documents, it is of the utmost importance to create the