Impact Angle Estimation Using an Active Bloodstain Shape Model
Key idea
Instead of estimating the impact angle using only the stain's width and length, this method analyzes the entire stain shape and learns the relationship between stain geometry and impact angle from experimental data.
Why this model was developed
In Bloodstain Pattern Analysis (BPA), the impact angle of a spatter stain is often estimated using a simple geometric method. Analysts measure the length and width of a stain, assume the stain is an ellipse, and compute the angle using:
While this method is widely used, it is based on several simplifying assumptions:
- Airborne droplets are assumed to be spherical.
- Droplets are assumed to form perfect elliptical stains upon impact.
- Physical effects such as surface texture, droplet dynamics, and fluid properties are ignored.
In reality, bloodstains often deviate from ideal ellipses. They may show irregular edges, tails, and small variations from their presumed elliptical shape.
Because of this, the traditional ellipse-based formula can produce systematic errors in the estimated impact angle.
To address these limitations, this research introduces a data-driven model that learns the relationship between bloodstain shape and impact angle directly from experimental data, without assuming stains are perfect ellipses.
Overview of the Method
The proposed method estimates the impact angle using two main components:
- Active Bloodstain Shape Model (ABSM): A statistical model describing how bloodstain shapes vary.
- Regression Model: A mathematical function that relates stain shape to impact angle.
Together, these components allow the system to estimate the impact angle from the full shape of the stain, rather than only its width and length.
The sections below describe how the model was created, how it is used, and what users can expect from it.
Step 1 – Collecting Experimental Bloodstain Data
The model is trained using experimentally generated bloodstains.
In controlled experiments:
- Blood droplets were dropped onto an angled surface.
- The true impact angle of each droplet is known.
- The resulting stains are photographed and digitized.
Each stain therefore has two key pieces of information:
- The actual impact angle
- The observed stain shape
This dataset forms the foundation for building the statistical model.
Step 2 – Representing Bloodstain Shapes
Instead of describing stains with only two measurements (width and length), the method captures the entire stain outline.
Each stain is represented as a set of landmark points along the stain boundary. Images from stains are processed in the following way to place the required landmarks: images are converted to grayscale, the stain is segmented from the background, tail information is removed, and landmarks are placed around the stain's outline.
These landmarks describe the contour of the main stain body, including asymmetries and deviations caused by impact and other fluid dynamics.
By representing stains this way, the method preserves much more information about stain geometry than the traditional approach.
Info
Note that the landmarks are only placed around the main body of the stains. We ignore tails, because they contain no useful information for impact angle estimation.
Step 3 – Building the Active Shape Model
Once all stains are represented using landmarks, the next step is to analyze how stain shapes vary across the dataset. To do so, all landmark sets are aligned, removing differences in scale, position, and rotation. This ensures that any remaining variation between shapes is caused by differences in stain geometry rather than by how the images were captured.
To extract how stain shapes vary in our dataset, we use an Active Shape Model (ASM). The ASM identifies the principal modes of variation in the dataset using Principal Component Analysis (PCA). In simple terms, the model learns the main ways bloodstain shapes change as the impact angle varies.
The result is a statistical model of bloodstain shapes consisting of:
- An average stain shape
- A set of shape variation parameters
Any stain can now be described as the average shape + a weighted combination of shape variations. These weights become the shape parameters that characterize a specific stain.
The image below shows several variations of the model by varying the first principal component.
Step 4 – Linking Shape to Impact Angle
After describing each stain using its shape parameters, the next step is to determine how these parameters relate to the impact angle. This relationship is learned using non-linear polynomial regression. This allows the model to follow the actual experimental data, rather than forcing the data to fit a predefined equation.
The regression model takes the shape parameters as input and predicts the impact angle, as illustrated in the graph below.
Step 5 – Estimating the Impact Angle of a New Stain
When analyzing a stain in practice, the method works as follows:
- Extract the stain contour from the image.
- Fit the Active Shape Model to the stain.
- Extract the shape parameters describing that stain.
- Apply the regression model to estimate the impact angle.
The final output is an impact angle estimate based on the complete stain shape.
Note on Stain Tails
Our initial model was a closed contour. This closed contour, however, does not accurately represent stains with tails, as shown on the left in the figure below. To resolve this, we adapted our model to be an open contour. This way, the model can better adapt to stains that have tails, as shown on the right below.
Advantages of the Method
Compared to the traditional ellipse method, this approach offers several important advantages.
Uses the full stain geometry
Instead of relying only on width and length, the method considers the entire stain shape, capturing more information.
Data-driven
The relationship between stain shape and impact angle is learned from real experiments, rather than imposed by theoretical assumptions.
Handles non-elliptical stains
Because the model is based on statistical shape variation, it can still work with stains that deviate from perfect ellipses.
Better alignment with experimental observations
Experimental results show that the traditional width-to-length ratio does not always follow the expected inverse sine relationship. The proposed model better matches the true data.
Fast
Analyzing stains with the proposed method is fast, because it eliminates the need to manually fit ellipses or measure a stain's width and length.
Summary
The Active Bloodstain Shape Model estimates impact angles by analyzing the full geometry of bloodstains rather than relying solely on width-to-length ratios. By learning the relationship between stain shape and impact angle from experimental data, the method provides a robust and efficient alternative to traditional ellipse-based calculations.
Scientific Reference
The Active Bloodstain Shape Model described on this page is based on the following peer-reviewed publication:
Joris, P., et al. (2014).
Calculation of bloodstain impact angles using an active shape model.
Journal of Forensic Sciences.
This publication presents the statistical shape modelling approach used to estimate bloodstain impact angles. The study also includes validation experiments using controlled bloodstain patterns and compares the proposed method with traditional ellipse-based impact angle estimation techniques.
Readers interested in the full methodology, experimental setup, and detailed evaluation results are encouraged to consult the original publication here .
Found a mistake?
Did you find a mistake, or does a page contain outdated information or something that is not working correctly? Please let us know so we can review it and fix it as soon as possible.
Forentrics - Bringing Evidence into Focus