Measurement technique based on hyperspectral imaging for occulted blood stain detection

The identification of washed blood plays a crucial role in forensic investigations, contributing to the identification and reconstruction of crime scenes. The standard methods for detecting washed blood are based on chemical tests or the use of Luminol. These methods can damage and alter the crime scene. In this paper, we present a novel approach to recognize bloodstains in tissues before and after washing up using hyperspectral imaging. Our technique involves the acquisition of hyperspectral images of various tissues at different exposure times, followed by data pre-processing and classification using a neural network. The hyper-parameters of the developed model are optimised by means of Bayesian optimisation technique. Through extensive experiments, we demonstrate the effectiveness of our approach in accurately classifying bloodstains in tissues before and after washing up, thus providing valuable insights to forensic practitioners. Our results pave the way for better forensic practices and improve the ability to identify blood-washed tissues.