Book Title: Computational Criminology: AI Applications in Forensic Science and Criminal Justice
Editors: Dr. Xavier Louis, Dr. Surbhi Girdhar, Ms. Aswathi Chandran Nair, Mr. Ravi Kumar, and Ms. Nandini Katare
Chapter: 15
DOI: https://doi.org/10.59646/704/15
Author: Satish Rai
Abstract
Bloodstain pattern analysis (BPA) has long been a foundational discipline in forensic science, enabling investigators to reconstruct the dynamics of violent crimes from the spatial and morphological characteristics of blood deposits. Traditionally reliant on expert visual interpretation, BPA has been subject to criticism regarding subjectivity, reproducibility, and cognitive bias. The emergence of computer vision, photogrammetry, and three-dimensional reconstruction technologies offers a transformative opportunity to place BPA on a more rigorous, objective, and quantitative footing. This chapter examines the theoretical foundations of BPA, catalogues the principal stain typologies, and critically evaluates the application of deep learning architectures including convolutional neural networks (CNNs) and vision transformers to the automated detection, classification, and measurement of bloodstains. Particular attention is accorded to photogrammetric and structured-light 3D scanning methods that enable accurate area-of-origin determination. The chapter further addresses validation challenges, courtroom admissibility considerations, and the ethical dimensions of algorithmic forensic evidence.
When blood is shed at the scene of a violent crime, it does not simply disperse randomly across surfaces. Rather, the behaviour of blood in motion is governed by the interacting principles of fluid dynamics, surface tension, viscosity, atmospheric resistance, momentum, and gravity. These physical forces determine how blood droplets travel through space, impact surfaces, fragment into secondary spatters, or accumulate into pools and transfer patterns. The resulting stains are not merely biological residues; they constitute a form of physical evidence whose geometry, distribution, and morphology preserve a silent but highly informative record of the events that produced them.