Collection of evidence from crime scenes generally aims to detect and extract biological stains to support downstream processing of forensic samples and to generate rapid intelligence to help the investigation service identify the person to whom it belongs. the biological stain. Current approaches traditionally rely on chemical and immunological techniques, which lack sensitivity and specificity. Moreover, existing evidence collection techniques are mostly manual, opposing the current era of automation that rules various areas of daily life.
The oldest forms of biological forensic analysis rely on fingerprints or bodily fluids, such as saliva, blood and semen, detected at a crime scene.15. Alternative methods exploit DNA profiling16 and detect gunshot residue on an object or even detected footprints. Nevertheless, most of these techniques are applied manually, increasing the workload of the crime scene evidence collection team, and are largely governed by the technical capabilities of the operator. Spurred by this concern, the “Future of Forensic and Crime Scene Science Conference” defined the critical drivers of system and technology design17 such as (1) miniaturization to increase portability and ease of use, (2) faster analysis, (3) simple “black box” interpretation, (4) easy integration of case information, and 5. low cost. Driven by these developing drivers, the iForenLIBS18 was developed for collecting gunshot residue (GSR) from a crime scene, which exploits laser-induced breakdown spectroscopy (LIBS) technology to analyze GSR particles by simultaneously detecting the three typical ammunition elements (lead, barium and antimony). An alternative method19 analyzed GSR using a chemical imaging system operating in the short-wave infrared (SWIR) region. Although this hardware configuration does not adhere to the portability design driver mentioned above, it is still an attractive solution for automatic collection of criminal evidence.
It should be noted that according to the literature, currently, these are the only devices developed at a technological readiness level (TRL)20 of at least 6, where the device undergoes field testing while considering (at least partially) the proposed design drivers mentioned above17. However, several attractive approaches exist that are still at a lower TRL of up to 3 or 4. Indeed, the great success of modern computer vision techniques, including deep learning strategies, has already been extended to collecting evidence at crime scenes or in support of criminal investigation teams. in general. For example, in a shoe print retrieval system21, the shoe print pattern is characterized based on shape features and an assigned relational graph. Recently, 3D crime scene reconstruction22 Technology has been developed, producing more insightful, complete and objective documentation for crime scenes. This system includes a laser scanner, a situational structured light scanner for fine measurements and a detailed structured light scanner to extract the maximum possible detail.
Currently, the collection of fingerprints and body fluid stains is manual, and the surface of the investigation scene should be considered to determine the appropriate collection methods. Depending on the type of surface, i.e. absorbent, non-absorbent smooth surface and non-absorbent rough surface, the suitable type of powder for fingerprint extraction is chosen. In addition, the collection methods depend on the type of fingerprint, i.e. latent, patent and plastic.23. For example, to collect latent fingerprints from body skin oil and sweat, fingerprint visualization requires an additional process24. Therefore, considering the current technological era and to facilitate criminal investigation services, developing a fully autonomous, i.e. one-click, device to capture fingerprints is essential. Nevertheless, to our knowledge, such an automatic fingerprint extraction system that meets the design drivers required17 and exceeds a TRL level of 4, i.e. beyond the proof of concept stage, has not yet been proposed.
To be complete, in contrast to the visualization and extraction of fingerprints and body coloring fluids, most current work focuses on the automatic classification and comparison of collected evidence against a database. rather than collecting them. Therefore, significant improvements have been made at the software level. For example, fingerprint classification, i.e. the grouping of fingerprints in a consistent way so that different impressions of the same finger are grouped together in the same group, relies on deep learning.25,26,27,28,29graph theory using directional data30crest flow and crest lines31.32and rule-based33 diets. Nevertheless, these techniques are purely software and constitute the process of following the process of visualization and extraction of this article.
Stimulated by this technology gap, this paper develops a “one-click device for rapid visualization and extraction of latent evidence”, aiming to develop and evaluate a fully autonomous, easy-to-use fingerprint and body fluid stains, fast and very efficient. extraction (image capture) system for crime scene evidence collection at crime scene. The proposed device improves the science and technology content of crime scene investigation, simplifies the evidence collection steps, and quickly completes the display and extraction of latent evidence, such as fingerprints and body fluid stains from the scene, based on a one-step process. This is important because our system overlooks the impact of cumbersome operation steps and the technical level of personnel on the quality of forensics, ensuring standardization, efficiency and collection of high-quality crime scene evidence, overcoming the shortcomings of existing methods. According to user reports from several frontline investigation departments in Shanghai, the developed device has acquired multi-angle fingerprint photos of non-aircraft objects at the scene of many criminal cases. After image fusion, clear and complete fingerprint images were obtained. Thus, the proposed device provides a robust technical guarantee to public security teams to solve criminal cases. Overall, the contributions of this work can be summarized as follows:
Exploit the optical properties of biological evidence and its media, use the mechanism and technology of various light source bands to display biological evidence, and integrate multiple light sources and image processing techniques.
Develop a “one-click device for rapid visualization and extraction of latent evidence at the scene” using the insights presented above. The developed system is portable with high battery life, meeting suggested design drivers for criminal evidence collection systems.
To solve the current problem of quickly finding and discovering potential biological evidence at a crime scene while avoiding complicated operating procedures, eliminating the effect of the technical skills of different personnel on the quality of evidence collected, and meeting the requirement for rapid disposal of crime scene evidence.
Through the proposed one-click device, our system acquires multiple photographs with multi-spectral and multi-angle lights, while geometry registration, extraction, optimization and evidence fusion are completed in the background before the display of the resulting image.
Although the developed device has many capabilities, it is an inexpensive device as its manufacturing cost does not exceed $1200.
The rest of this article is as follows. In the section “Review of the literature”, presents the work in progress in the collection of evidence from crime scenes. In the section “Study on the integration of multi-mode light source and light guiding technology”, the section introduces the integration of multi-mode light source and light guiding technology, while the section “Design of the proposed device” presents the design of the developed one-click device for rapid visualization and extraction of latent evidence at a crime scene. Finally, the “Conclusion” section concludes this work.