Estimation of age in forensic medicine using multivariate approach to image analysis

A new method for victims' age estimation, based on the image processing and analysis of remains bones structure, is proposed. Digital images of lumbar vertebras cuts were used as a major information source. The age related properties were extracted from the images using classic texture analysis approach based on statistical analysis of grey-level co-occurrence matrix, fractal analysis, wavelet transformation and Angle Measure Technique. Projection on latent structures regression was chosen for calibration and prediction. The method has been applied to 70 male and 63 female individuals aged from 21 to 93 and results were compared with traditional approach. Some important questions and problems have been raised.