MondialMarmi: A collection of images of polished natural stones for colour and texture analysis

Description

MondialMarmi is a collection of colour images of polished natural stones for image processing purposes. It was originally conceived for benchmarking colour and texture descriptors.

The current version contains 25 classes, and there are four images for each class. Each image correspond to one tile. The images were acquired under controlled illumination conditions (see Image acquisition) for a description of the acquisition setup). Each tile was imaged under the following in-plane rotation angles: 0°, 10°, 20°, 30°, 40°, 50°, 60°, 70°, 80° and 90°.

Classes

Acquamarina	Azul Capixaba	Azul Platino	Baltic Brown	Bianco Cristal
Bianco Sardo	Blue Pearl	Dakota Mahogany	Giallo Antico	Giallo Napoletano
Giallo S. Cecilia	Giallo Veneziano	Kashmir Gold	Nero Africa	Paradiso Bash
Paradiso Classico	Rosa Monçao	Rosa Porriño	Rosso Multicolor	Sky Brown
Verde Bahía	Verde Marino	Verde Ming	Verde Oliva	Violetta

Usage and organisation

1. Download the dataset from images.zip
2. Unzip the folder

Once uncompressed the dataset has the following structure:

./<rotationAngle>/<className_rotationAngle_sample>.bmp

• <rotationAngle> indicates the in-plane rotation angle at which the tile was imaged. This can be 00, 10, 20, 30, 40, 50, 60, 70, 80 or 90.
• <className> represents the class name — that is, the commercial denomination of the granite. This can be any of the 25 class names listed above.
• <sample> is a two-digit code that uniquely identifies the tile. There are four tiles for each class and angle).

Image acquisition

The acquisition process was carried out through a system composed of a base with a slot to accommodate the tile, a dome illuminator and a digital camera (see Fig. 1). A rotatable support at the top of the dome made it possible to acquire images under different rotation angles. Each tile was placed in the slot and 10 images were taken at different rotation angles from 0° to 90° by steps of 10°.

Fig. 1 The acquisition system (IT patent no. 1413266)

The illumination conditions were kept constant throughout the whole acquisition process. The images were captured at the natural resolution of the camera's sensor (2560px × 1920px), and were finally cropped to a central part of size 1500px × 1500px, corresponding to an area of ≈21cm × 21cm.

For colour calibration an X-Rite Digital SG colour checker (140 colour targets) was also put in place and imaged. The RGB values of the 140 colour targets were acquired using the same settings adopted for the tiles; the corresponding device-independent CIE XYZ coordinates (under illuminant D65) were measured through a Minolta CR-200 Chroma Meter (CalibrationData.csv, CalibrationData.pdf).

Technical data

Illuminator

Spectrum Illumination Monster Dome Light 18.25" with white HB LEDs. Technical data available here.

The illumination settings were:

• Voltage 15V
• Average illumination level at the surface of the tiles ≈600lx

Camera

Edmund Optics EO-5012C LE.

The camera settings were:

• Image size: 2560px x 1920px
• image format: BMP
• Encoding: linear (no gamma correction)
• Frame rate: 2.40fps
• Exposure time: maximum allowed by the frame rate
• Overall gain: 1x
• Colour gain (determined through white balancing): R = 14x, G = 1x and B = 5x

Contributors

Francesco Bianconi and Antonio Fernández

How to cite

If you wish to use this dataset for your research activity please consider referencing the following papers (where appropriate):

1. Bello-Cerezo, R., Bianconi, F., Fernández, A., González, E., Di Maria, F. Experimental comparison of color spaces for material classification (2016) Journal of Electronic Imaging, 25 (6), art. no. 061406
2. Bianconi, F., González, E., Fernández, A., Saetta, S.A. Automatic classification of granite tiles through colour and texture features (2012) Expert Systems with Applications, 39 (12), pp. 11212-11218.

Acknowledgements

We are grateful to Mondial Marmi s.r.l. (Perugia, ITALY) for providing the natural stone tiles used to make up this dataset.

The work was partially funded by the European Commission under project LIFE12 ENV/IT/000411 “Enhanced material recovery and environmental sustainability for small scale waste management systems” - EMaRES”.

For their support during the image acquisition process we would like to thank:

• Raquel Bello Cerezo
• Stefano Baglioni

For technical support and logistics:

• Stefano Antonio Saetta

License

Distributed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

Disclaimer

The information and content available on this repository are provided with no warranty whatsoever. Any use for scientific or any other purpose is conducted at your own risk and under your own responsibility. The authors are not liable for any damages - including any consequential damages - of any kind that may result from the use of the materials or information available on this repository or of any of the products or services hereon described.