Development of an image-based gloss measurement instrument

Orange peel has been defined as ‘the appearance of irregularity of a surface resembling the skin of an orange’.13 It is a generally undesired visible surface nonuniformity which typically results from an improper paint application technique, and can be observed as a wavy pattern of bright and dark areas. Within industry, orange peel is generally quantified with an orange peel meter, by the use of which the optical profile of the sample surface is scanned with a laser or LED light source.14 From the profile structure, a set of numbers is calculated using mathematical band-pass filter functions, which form a structure spectrum. This structure spectrum then allows for an analysis of the orange peel. As an alternative, a technique called ‘phase stepped deflectometry’ might be used.15 In this method, a display is used to project a fringe pattern onto the test surface, and a camera captures the reflected image of the pattern. Knowing the geometric relationship between the display, surface, and camera, the optical profile can be spatially modelled from the captured reflections. Finally, equivalent to the former method, band-pass filter functions are used to simulate the human eye resolution at various distances.

Three samples out of a dedicated set of 10 orange peel visual reference standards developed by ACT Test Panels LLC were selected for use to test the suitability of the i-GM to capture and quantify the phenomenon. The acquired image of these three samples, denoted as Orange Peel Standard (OPS) #1, #5, and #10, respectively, is presented in Fig. 3. The samples exhibit a high, medium, and low degree of orange peel, respectively. As stated before, for these measurements, the collector lens of the i-GM is positioned to image the sample plane onto the detector array. A clear difference between the three surface topographies can be observed.

The images can be analyzed by the use of digital image processing techniques. Figure 4 presents the result of a discrete fast Fourier transform of a horizontal cross section for each sample. A clear distinction in the Fourier magnitude spectrum of the three samples is obtained. As expected, the fundamental spatial frequency predominates the image of the test surface with no visual orange peel effects (OPS #10). For the surface with moderate orange peel (OPS #5), larger frequencies come into play, while the range of encountered spatial frequencies as well as the importance of larger frequencies further increases for the surface that exhibits the highest orange peel (OPS #1) (Fig. 5).

The visual contrast between an observed highlight and the sample background is known to be of primary importance in visual gloss appraisal, especially for high-gloss surfaces.16,17 The concept of contrast gloss has been reported for a longer time. Hunter defined contrast gloss as the type of gloss that produces a characteristic appearance of contrast between specularly reflecting areas and other areas of the surface.6 Jones proposed an instrument for quantifying contrast gloss, based on a comparison of the reflectance from the sample in the specular reflection direction (45° viewing direction for a 45° light incidence angle) with the reflectance at normal viewing direction.18 However, to the best of our knowledge, no industrial measurement device is currently in use, neither is there any applicable (normalization) standard in which a measurement method for contrast gloss is defined.

More recently, the concept of contrast gloss received increased attention through a study using image synthesis, performed by Ferwerda et al.16 Lately, a visual gloss prediction formula was proposed and validated by Leloup et al., based on the measured luminance contrast between the reflected virtual image of the specular light source (\(L_{\text{im}}\)) and the surrounding surface area (\(L_{\text{b}}\))19;

To exemplify the concept of contrast gloss, a rendered picture of a white vs. a black sphere with the same physical gloss parameters (in terms of the specular reflectance and spread of the specular lobe, according to Ward’s light reflection model),20 enclosed in a checkerboard box illuminated by an overhead area light source, is presented in Fig. 6.16 Clearly, the black sphere produces a higher visual gloss impression, although the same physical characteristics are attributed to both spheres. A classical gloss meter will report the same gloss value, but the higher contrast clearly induces a larger visual gloss perception.

This is further illustrated in the results of Table 2. The specular gloss values of a white, gray, and black high-gloss sample of the NCS series (i.e., sample NCS W G90, NCS G G90, and NCS B G90, respectively) all number approximately 90 GU. It has been shown, however, that their gloss perception is clearly different.17,19

By inclusion of a second light source (called the a-specular light source) in the i-GM (see Fig. 1), positioned in the direction of the normal on the test surface, it becomes possible to quantify the background luminance of the test surface in conjunction with the specular reflection originating from the specular light source. Therefore, a second Xicato XSM 80 LED module was introduced to act as the a-specular light source in the instrument and baffled to prevent stray light from entering the CMOS detector directly. Similar to the specular source, this a-specular light source was driven with a constant current of 10 mA.

As a measurement example, the acquired images of test samples NCS W G90 and NCS B G90 (white and black test samples with nominal specular gloss of 90 GU, respectively), captured with the updated i-GM with both sources turned on, are presented in Fig. 7. An obvious difference in background luminance between both test surfaces can be observed. In practice, the luminance contributions originating from specular and a-specular reflection can be most easily separated by taking two images in sequence, respectively, with only the specular and only the background source turned on. Based on our proposed visual contrast gloss equation,19 the visual contrast gloss impression of the black sample is thereby quantified to be 1.5 × higher than that of the white sample.

For several decades, conventional specular gloss meters have been used through multiple branches of industry for the assessment of the surface quality of products. Nevertheless, instrumental gloss readings are not easily related to visually perceived gloss. Studies have demonstrated that visual gloss sensitivity is higher at the extremes of the scale, i.e., for matte and high-gloss surfaces.21,22 The fact that observers might rely on variations of other attributes of gloss such as, for instance, the distinctness of the reflected image, or the contrast between the specular highlight and surface background, has been put forward as a possible explanation. Further studies followed the multiple-dimension approach to perceptual gloss, in which the visual system treats these dimensions and features as a whole, a so-called gestalt.5,12,23

Results from specular gloss meters are solely based upon the amount of specular reflection. In that sense, they do not offer information about additional cues to surface gloss. In this paper, a prototype i-GM has been proposed that does offer the possibility to quantify multiple attributes of surface gloss simultaneously. The instrument makes use of an image-based CMOS detector. The i-GM has been designed such that specular gloss readings correspond with results from a conventional gloss meter obtained in a 60° measurement geometry; on a series of 16 matte to high-gloss test samples, an average absolute deviation of only 1.5 GU has been reported. For the new camera-based instrument to be accepted, this accordance with the current traditional and widely used instrumentation is essential.

Besides specular gloss, it has been demonstrated that the i-GM can be used to quantify orange peel and contrast gloss. For the assessment of orange peel, new industrial instruments have already been proposed. However, these instruments do not provide information on specular gloss or contrast gloss and thus need to be used in combination with other instrumentation if a full characterization of the surface appearance quality is to be considered. In comparison with a standard specular gloss meter, they are moreover expensive. Finally, these instruments consider and assess orange peel indirectly from a characterization of the surface profile or texture. With the proposed i-GM, orange peel is assessed directly from the acquired reflection image using well-known image post-processing procedures. The use of a low-pass, band-pass, or high-pass filter in the frequency domain seemed thereby appropriate to distinguish between surfaces with high, medium, or low orange peel, respectively. Yet, the selection of the cutoff frequencies from the amplitude spectrum was made rather arbitrarily. Clearly, the development of a procedure for appropriate selection of the cutoff frequencies is a point of attention. Therefore, more in-depth investigations on different materials (paper, wood, plastics, etc.) are needed, in order to find out which type of band-pass filter is the most appropriate for orange peel segmentation. Ultimately, it remains to be explored whether a so-called orange peel index could be inferred for quantifying orange peel by the use of just one metric, in analogy with specular gloss.

While alternatives have been proposed for orange peel, to the best of our knowledge, no industrial instrument exists which quantifies contrast gloss. By simple inclusion of a second, a-specular light source into the i-GM, it was demonstrated that contrast gloss can be assessed rather easily by performing two image acquisitions in sequence, with the specular and a-specular light source alternately turned on. As an example, white and black samples with equal specular gloss were assessed. While the concept of contrast gloss and the search for a suitable metric has been investigated before, the i-GM has proven to be able to determine a numerical value, at least on achromatic samples. Further psychophysical research should be performed on a dedicated set of chromatic stimuli to find a correlation between visual contrast gloss and the specular and off-specular characteristics, taking into account colorimetric information. The i-GM, being equipped with an RGB sensor, will have the potential to record all the necessary input parameters.

Finally, the presented i-GM offers some additional opportunities which were not discussed in this paper. Surface defects and nonuniformities such as scratches, dents, or gloss homogeneity can easily be detected, and even the nonuniform texture of effect coatings, i.e., sparkle and graininess,24 can be analyzed. These aspects need further exploration to discover the full potential of the i-GM.

An important aspect of gloss that has not been addressed in this manuscript is the so-called distinctness-of-image (DOI), which is perceptually related to the distinctness and sharpness of the reflected images.6 The reason for this is that in order to assess DOI, the camera lens should be focused on the projection lens instead of on the sample surface. For this, the distance between camera and collector lens thus needs to be changed, which is impossible with the current prototype. In order to overcome this issue, an objective lens with variable focus might be envisaged. This alternative will be investigated in the near future.