3D Forensics: Documentation, Analysis, and Visualization of Evidence

Understanding 3D Forensics

• Definition: The use of three-dimensional technologies to assist with documentation, analysis, and visualization of evidence. These three aspects are crucial for forensic work.
  • Documentation: Capturing the scene and evidence.
  • Analysis: Processing and interpreting the collected data.
  • Visualization: Presenting the findings clearly, often in 3D, which is highly beneficial for juries.
• Documentation Mantra: In 3D documentation, the goal is to keep the object or evidence geometrically and visually "true to form" – as close to its original 3D state as possible.
  • A 2D photo translates a 3D world to a 2D representation, which isn't true to form unless the original surface is flat (e.g., bloodstains on a flat wall).
  • For 3D, this means accurately representing complex scenes like bedrooms, outdoor crash scenes, or fire scenes.
• Lord Kelvin's Quote: William Thompson, First Baron Kelvin, emphasized the importance of quantification: "I often say that when you can measure what you're speaking about and express it in numbers, you know something about it. But when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind."
  • In 3D scanning, everything is recorded numerically (e.g., with $XYZ$ coordinates for points), providing quantifiable data about the space and evidence.

3D Documentation Technologies

• Chronological Order of Development:
  • Photogrammetry: Oldest, dating back to the 1850s with the first cameras. A Frenchman used hot-air balloons to take pictures for surveying.
  • Total Station: Laser-based instrument from the late 1980s / early 1990s, popularized by police for mapping crime and crash scenes.
  • Laser Scanner: The 'big brother' to the total station, emerging in the late 1990s and gaining popularity with police agencies around 2010.
  • Structured Light Scanners: (Discussed in detail below).
  • iPhone LiDAR: Released in May 2020 on Apple devices.
• Most Popular for Forensics Today (The 'Holy Trinity'):
  • Photogrammetry: Widely used, particularly with drones for crash scenes, due to ease of data acquisition.
  • Total Station (now often GNSS): Evolved into GPS-based systems; often used in conjunction with photogrammetry.
  • Laser Scanner: Essential equipment for many police agencies globally for mapping various scene types.

Structured Light Scanners

• Functionality: Consists of a projector (emitting a pattern) and a camera (recording the image).
  • White Light Scanners: Project visible 'fringe patterns' (straight lines) onto an object.
    • The deformation (curving, bending) of these stripes allows 3D information to be recovered rapidly.
    • Can project $30$ to $60$ patterns per second.
    • Captures both 3D geometry and color.
    • Limitations: Light-based, so struggles outdoors in sunlight.
    • Applications: Useful for smaller evidence like footwear impressions, shoe prints, cartridge cases, capturing fine details. Creates a mesh structure.
  • Infrared Structured Light Scanners: Use lasers to emit an invisible infrared pattern (e.g., a grid of dots).
    • Features two infrared cameras and a color camera.
    • Larger, brighter corner dots in the grid help resolve 3D data on darker surfaces, which are problematic for laser reflection/absorption (e.g., shiny black cars).
    • Data acquisition is very fast (e.g., a small bedroom scene in $60$ seconds), capturing high-density data suitable for bloodstain pattern analysis or bullet trajectories.
• Face ID as a Common Example: iPhones use an infrared camera and a projector to blast $30,000+$ points at a face and background.
  • This is a structured light scanner for geometric facial recognition, not just color or imagery.

Photogrammetry

• Etymology: Greek words 'photos' (light), 'grammar' (to write/draw), 'metron' (to measure). Meaning 'light drawing' to measure.
• Definition: The art and science of taking reliable measurements from photographs.
• Types of Photogrammetry:
  • Aerial: Popular with drones for mapping large areas easily and cheaply.
  • Terrestrial: From ground level, for large landscapes or scenes; sometimes uses poles for higher camera positions.
  • Close-range: For objects the size of a car, tires, or smaller evidence (e.g., footwear impressions).
  • Micro-photogrammetry: For very small objects, even with scanning electron microscopes.
• Workflow:
  1. Taking Photographs: Requires many overlapping photos (more overlap generally means better results).
  2. Feature Matching: Software detects high-contrast features (corners, light/dark, texture). Matches these features across multiple photos to determine camera locations and relationships.
  3. Sparse Point Cloud Generation: Creates a sparse distribution of points representing camera positions. These points are used to calculate where cameras were located.
  4. Dense Point Cloud Generation: Once camera positions are known, the software goes back to pixel level to create a dense 3D model (millions of points). Breaks apart when viewed too closely.
  5. Meshing: 'Shrink-wrapping' or draping a sheet over the dense point cloud, creating a network/lattice. Reduces data weight and is preferred by computer programs, especially game engines.
  6. Texturing: Pixels from original images are 'pasted' onto the mesh patches to create a realistic model.
• Applications:
  • Postmortem Fingerprint Identification: Scanning fingers to capture ridge details for comparison.
  • Video to 3D Models: Extracting frames from video (e.g., clandestine grave scenes) to create detailed 3D models. Allows documentation of intricate details difficult to capture manually.
  • Large-Scale Mapping: Drones can map vast expanses (cities, fields) with billions of points, not just millions.
• Software Examples: RealityCapture, 3DF Zephyr, Metashape, Coldmap (some free options available).

Terrestrial Laser Scanners

• Manufacturers: Trimble (Paris office), FARO (Germany HQ), Leica (Switzerland), ZNF (Germany), Rigel (Austria). FARO, Leica, and Trimble are most popular in forensics.
• Measurement Principle: Heart of the scanner contains a mirror (often $45^ ext{o}$ angle) and a laser module.
  • A laser beam is emitted, reflected $90^ ext{o}$ by the mirror, and travels to a surface.
  • It measures the distance to the surface.
  • The mirror spins, and the whole unit turns horizontally $360^ ext{o}$. Both vertical and horizontal angles are measured.
  • This creates polar coordinates (distance, vertical angle, horizontal angle) which the scanner converts to Cartesian coordinates ($XYZ$).
  • Newer scanners use a $905$ nanometer laser (visible with infrared cameras).
  • Measures very small areas (e.g., $2$ mm diameter) rapidly.
• Underlying Technology: LiDAR (Light Detection and Ranging, a portmanteau of light and radar).
  • Emits a signal (light), which bounces off a surface, and part of the signal returns to the sensor.
  • Time of Flight (ToF) Method:
    1. A discrete light pulse (on/off) is emitted, starting a timer.
    2. The pulse hits a surface and is partially reflected back to the scanner.
    3. The timer stops upon return.
    4. Calculation: The distance is calculated using the formula: D = rac{C imes T}{2}, where $D$ is distance, $C$ is the speed of light (approx. $3 imes 10^8 ext{ m/s}$), and $T$ is the total time for the round trip (divided by $2$ for one-way distance).
• Applications: Fantastic for both indoor and outdoor scenes. Can document complex multi-floor crime scenes (e.g., a crime house in an hour or two).

iPhone LiDAR

• Hardware: Available on iPad Pro and iPhone 12 Pro / Pro Max and newer models (up to 16 Pro, 17 announced) since May 2020.
  • A small sensor (emitter and receiver) on the device.
• Functionality: Determines distances of objects in front of the device.
  • Emits an interlaced pattern of dots (e.g., $12 imes 12$ series on older phones, shifting rapidly to create a larger overall pattern).
  • The receiver determines the distance from the device to the wall for each point.
• Depth Maps: Represents distance as color (e.g., blue for farther, red for closer). Useful in 3D imaging, graphics.
• Applications Beyond 3D Forensics: Augmented reality, portrait mode (uses LiDAR to determine subject distance/blur areas).
• Recon3D App: Uses iPhone LiDAR for mapping.
  • Output: High-density point clouds (e.g., $1-2$ mm density).
  • Examples: Body on the ground (approx. $60$ seconds to scan), bloodstain scenes (for 3D analysis of impacts), vehicles (for crash scenes, approx. $2$ minutes to scan), exteriors, interiors, and roadways (smaller areas).

Analysis with 3D Data

• Reconstruction vs. Recreation:
  • Reconstruction: Substantive, used to prove something in court. Based on physical evidence (e.g., vehicle scans, tire marks, black box data in a crash). Carries significant weight at trial.
  • Recreation: Demonstrative tool, used to supplement an expert's opinion, not necessarily to prove something definitively. Less evidence may be available. Can show what didn't happen. Has less weight at trial.
• Specific Analysis Types:
  • Bite Mark Analysis: Using 3D to scan dentition and compare bite marks (an area receiving criticism due to reliability concerns).
  • Footwear Impression Analysis: Scanning an impression and a suspect's shoe, then overlaying them to create a 'heat map'. Provides visual comparison and a statistical histogram of point distribution, offering a different analytical paradigm.
  • Suspect Height Analysis: Combining laser scan data of a scene with video of a suspect to estimate height.
  • Bullet Trajectory Analysis: Simplifying documentation of complex trajectories (through walls, vehicles, ricochets) that are tedious to do manually. 3D capture makes analysis easier.
  • Bloodstain Pattern Analysis: Enabling analysis of complex scenarios previously difficult by hand (e.g., spatter from a hammer impact). Allows analysis regardless of surface curvature (flat wall or curved surface).
    • Cast-off Pattern Research: Using 3D motion capture (e.g., with a blood-soaked rod) to study blood droplet ejection and create new analysis methods. Red lines show backtracked bloodstain trajectories, blue lines show rod tip path, revealing how trajectories follow the swing path. This leads to concepts like 'path volume envelope' to define origin areas.
    • Cessation Patterns: Patterns created when a blood-carrying object abruptly stops.
    • Expirated Patterns: Patterns from coughed/spit blood (e.g., using human blood, slowly demonstrated to show pattern formation for analysis).

Visualization of 3D Data

• Planned Drawings: Traditional 2D drawings are still valued in court for simplicity and printability. 3D data can easily be converted to 2D plans.
• 3D Virtual Tours: Laser scanners capture panoramic images that can be combined with 3D data to create virtual tours.
  • Allows integration of crime scene photographs, video, and measurements.
  • Can be exported (e.g., on USB) for judges and juries.
  • Allows viewing in panoramic and infrared spectrum options.
• 3D Printing: Converts 3D models into physical objects. Can scale objects (large to small, small to large).
  • Examples: A car tire, a footwear impression, a murder victim's vertebra.
  • Emerging Concern: 3D printing of self-fabricated firearms is a growing topic in forensics.
• Virtual Models / 3D Players: Presentation of evidence in 3D (e.g., a recovered bullet) via an interactive player.
  • Can be distributed on USB, allowing inspection without bringing physical evidence to court.
  • Can include embedded photographs, information, and measurement tools.
• Game Engines: Specialized area becoming more popular for reconstructing events.
  • Allows re-creation of scenes (e.g., officer-involved shootings) with freedom of camera movement (different perspectives: top-down, following officers).
  • Officers' positions can be mapped from video onto the 3D environment.
• Virtual Reality (VR):
  • Training: Used for forensic training (e.g., swabbing, brushing, fingerprinting, evidence filing).
  • Scene Immersion: Experiencing a crime scene with analysis results in 3D (e.g., a 'CAVE' room with projectors on all surfaces).
    • Allows judges, jurors, and investigators to virtually step back into a scene and interact with the analysis results (e.g., seeing bloodstain trajectories in 3D).

Further Learning Resources

• Recon3D YouTube Channel: Focuses on the iPhone app and its functionalities.
• 3D Forensics YouTube Channel: Features interviews with forensic professionals, videos on photogrammetry, laser scanning, and other related topics.
• Contact Information: Website and email provided for reaching out (students can inquire about app trials).