Automatic mineral identification using color tracking. (May 2017)

• Record Type: Journal Article
• Authors: Aligholi, Saeed; Lashkaripour, Gholam Reza; Khajavi, Reza; Razmara, Morteza
• Is Part Of: Pattern recognition. Volume 65(2017:May)
• Page Start: 164
• Page End: 174
• Publication Date: 2017-05
• Subjects: Automated mineral identification -- CIELab color space -- Modified Hausdorff distance -- CIEDE2000 color distance; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2016.12.012
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: Optical properties of crystals are one of the most powerful features for mineral identification. In this study, a novel automated color-based mineral identification (MI) procedure is developed, which can efficiently incorporate color variations under plane-polarized (PPL) and cross-polarized (XPL) illumination modes. Any unknown mineral is registered by two 19-point sets (under PPL and XPL) in the CIELab color space and the proposed scheme tracks mineral color variations in the CIELab color space. If the corresponding PPL and XPL modified Hausdorff distances (with those of the known mineral) are within acceptable ranges, it can be recognized as a specific mineral. The method is easy to implement compared to methods that examine different color-based features. In addition, not much color information is lost as is the case with feature-based MI schemes. The proposed method, therefore, provides a significant improvement in discriminatory power and compared to conventional methods, the procedure enables reliable and consistent identification over a larger range of minerals. Highlights: Color variances in minerals under plane polarized light (PPL) and cross polarized light (XPL) are patterned as color paths. Color tracking in CIELab space is employed for mineral identification. Modified hausdorff distance (MHD) equipped with the CIEDE2000 color distance is employed for compare the minerals color paths under PPL and XPL illuminations. This method is effectively superior to multi-feature MI approaches due to its simplicity and ease of application. In addition, not much color information is lost as is the case with feature-based MI schemes.