Towards Enriching the Estimation of Stone Degradation through Computer Vision Analysis

Black crusts on stone surfaces, mainly composed of gypsum crystals, not only aesthetically affect the stonework but also cause further decay. Appropriate conservation methods should nondestructively detect the pollutant type and extent. In this paper, we propose a computer aided diagnostic method for automatically detecting deterioration patterns on black crusts. The proposed non-destructive method can be applied in situ. The applied detection process quantifies the alteration due to weathering and assesses the efficiency of several chemical cleaning methods. The detection process is based on a combination of Gaussian and morphological filtering. The results of both methods are fused via a Conditional Thickening operation in order to obtain reliable results concerning the spatial density, topology and extent of the deterioration patterns, while preserving their extent and shape characteristics. The algorithm is tested with a series of images depicting the corrosion state on marble surfaces. Decay patterns, classified by their chemical composition as black and white particles, are significantly decreased after the chemical cleaning, as indicated by the applied algorithmic approach. The corrosion alteration after cleaning is quantified in terms of robust statistical metrics expressing the extent, spatial density and thickness of the crusts.