Analyzing the color of forensic textile using smartphone-based machine vision

Color is an important characteristic of textile, and its analysis is of great significance for the forensic characterization of textile. The colorimetry method based on visual observation provides a subjective assessment; the instrument-based color analysis method is objective but requires expensive equipment and professional technicians. In this study, a smartphone-based machine vision method for color analysis was established. A smartphone with a camera was used for image acquisition, and the free software ImageJ was used for image processing. The captured RGB image was first converted to a Lab Stack, and then the target area was selected for L*a*b* value analysis. The influence of acquisition equipment, light source, illumination/photography angle and distance, and sample on color analysis was investigated. Fifteen red textile pieces were analyzed using optimized machine vision methods, and the results were compared with those obtained using the microspectrophotometry by both hierarchical cluster analysis and K-means clustering method. The results of the two methods were consistent, thereby confirming the reliability of the machine vision method. The smartphone-based machine vision color analysis method is cheap, simple, accurate, and objective; thus, it has great potential to be widely used in forensic science.