Elvira Sukma Wahyuni, Kern Cesarean Ahnaf, Firdaus Firdaus, Nurul Ashikin Abdul-Kadir, Nor Aini Zakaria, Idha Arfianti Wiraagni, Diwangkoro Aji Kadarmo

Wound Depth Measurement System in Forensic Cases using Image Processing and Machine Learning

Introduction

Wound depth measurement system in forensic cases using image processing and machine learning. Improve forensic wound depth assessment with an 85% accurate SVM system. This image processing method classifies wound stages (2, 3, 4) from color features, enhancing investigations.

Abstract

Accurate evaluation of wound depth is crucial in forensic investigations, as it significantly affects case assessments and outcomes. This study introduces a method for classifying wound depth using a Support Vector Machine (SVM) model and compares its performance with Decision Tree and Logistic Regression models. The classification was based on color features extracted from HSV and LAB color spaces. The da-taset consisted of 76 images categorized into three stages: stage 2 (36 images), stage 3 (12 images), and stage 4 (28 images). Model performance was evaluated using confusion matrices, precision, recall, and F1-score. The SVM model achieved an overall accuracy of 85%, demonstrating higher precision and re-call across all stages compared to the Decision Tree and Logistic Regression models, which achieved 50% and 70%, respectively. The results indicate that the SVM model performed particularly well in distinguish-ing stage 2 wounds, although differentiating between stages 3 and 4 remained challenging. Overall, the proposed system shows potential to enhance the accuracy and efficiency of forensic wound evaluation by providing a rapid and objective classification tool. However, as the system was tested on a limited dataset under controlled conditions, further research should expand the dataset, incorporate additional features, and explore other machine learning algorithms to improve robustness and applicability in real forensic contexts.

Review

This study introduces a relevant and timely approach to a critical area in forensic science: the objective classification of wound depth. By proposing a system that leverages image processing for feature extraction (color features from HSV and LAB color spaces) and machine learning for classification, the authors aim to enhance the accuracy and efficiency of forensic wound evaluation. The core contribution lies in demonstrating the potential of a Support Vector Machine (SVM) model to classify wound depth, offering a promising avenue for a rapid and standardized tool in forensic investigations where precise assessment significantly impacts case outcomes. The methodology involved comparing the performance of SVM against Decision Tree and Logistic Regression models on a dataset of 76 images, categorized into three wound stages (stage 2, 3, and 4). Model evaluation, utilizing confusion matrices, precision, recall, and F1-score, revealed the SVM model as the most effective, achieving an overall accuracy of 85%—a notable improvement over the Decision Tree's 50% and Logistic Regression's 70%. Specifically, the SVM model demonstrated particular strength in distinguishing stage 2 wounds. However, a crucial limitation acknowledged by the authors is the persistent difficulty in differentiating between stages 3 and 4, which is compounded by the limited and somewhat imbalanced dataset, particularly for stage 3 (12 images). While this paper presents a valuable initial exploration and a promising proof-of-concept, its practical applicability in real forensic contexts is currently constrained. The system's testing under controlled conditions and reliance on a relatively small dataset significantly impact its generalizability and robustness. For future development, it is imperative to substantially expand the dataset to encompass a wider variety of wound types, skin tones, lighting conditions, and image qualities typically encountered in actual forensic cases. Furthermore, incorporating additional image features, such as texture, shape, or geometric properties, and exploring more advanced machine learning or deep learning architectures could markedly improve the model's ability to differentiate challenging stages and enhance its overall reliability for critical forensic applications.

Full Text

Comments

You need to be logged in to post a comment.