Get the picture: Using photogrammetry as a measure for comparing surfaces

When I first started my PhD over a year ago, I was told to expect the unexpected when it came to research. Lo and behold, for several weekends of the last few months I have been taking 100’s of pictures of the floor outside, all part of my research. Given that my background is in human biomechanics this seems at first particularly odd, but makes more sense given that my PhD is looking at human locomotion outside of the lab! Admittedly, it appears a loose connection right now between the two, but it can all be answered by a technique known as photogrammetry.

Photogrammetry is the technique of making 3D models from a series of overlapping photographs. Mostly used in mapping, architecture and archaeology. I am aiming to use this technique to compare different outdoor surfaces.

As you are all probably aware, the surfaces we walk on everyday differs greatly. Whether this be on paving stones, gravel, grass, tarmac, pebbles or over steps, the surfaces we walk on ultimately change how we interact with the environment. Some surfaces can be more complex than others, causing us to walk with reduced stability. This means we have a greater chance of falling over. Falls are particularly problematic for the elderly, cited as the most common cause of death from injury for people aged 75+. Therefore understanding how we navigate across different surfaces is essential in reducing fall risk. To do this, we need to not only understand how people move across different surfaces but also how these surfaces themselves differ. What features of the surface cause the greatest change to behaviour? Is this related to compliancy? slope? unevenness? Currently there is no standard method for comparing surfaces and that is what I hope to address in this current study. One technique I have used to compare surfaces is through physical measurements. Enter photogrammetry.


Figure 1: Photographs of one of the 17 surfaces used in the study.

For my study, photogrammetry models were created for 17 different surfaces. These surfaces were either flat, uneven, compliant, sloped, across multiple levels or had a combination of these features. Surfaces were chosen to give a broad range of surface types, but include surfaces that are often reported as more likely to cause a fall. An example of one of the surfaces used in the study is shown in Figure 1. This surface consists of loose stones, therefore the surface is both uneven to walk on and compliant.

For each surface, an area of 1m2 (length x width = 2m x 0.5m) was measured and drawn out. Consistent area size allowed an equal comparison between surfaces in the analysis. Photographs were then taken from multiple angles, approximately 40 photographs required to build our models. Once taken, photographs could then be uploaded to the Agisoft PhotoScan software to create the model. Figure 2 shows a model of the surface shown in Figure 1.


Figure 2: A 3D model of the surface shown in Figure 1.

Once created, models could then be exported into Meshlabs software, where physical measurements could be calculated. In this study, we compared surfaces geometrically by calculating a relief index. This index is defined as the ratio between the surface area of the model and the planar area. For surfaces like that shown in Figure 2, the relief index is higher compared to a smoother surface.

Relief index is one of the physical measures we are using to compare surfaces in this study. Along with other physical measurements, we are also comparing surfaces through behavioural and perceptual measures. Behavioural measures were assessed in a separate study where participants walked over each surface whilst their gaze and gait behaviour was recorded. This information gives us an idea of people’s stability and risk of falling whilst walking. We also assessed perceptual measures in a third study, asking participants to rate surface roughness and their perceived level of stability before and after walking across each of the surfaces. Part of this research is ongoing but from the results we should have a good idea, from our multiple measures, of the how surfaces differ and how we can compare them.  Understanding the surfaces we walk on is essential to reducing the prevalence of falls in the elderly. Check back for future posts on the full analysis of results.


Blog post by Nick Thomas

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s