Although online shoe retails are flourishing, the increase in product return rates is limiting profit margins. While shoe stores offer customers various high-tech options for determining foot length for the appropriate fit, buyers using the Internet are stuck with trying for the right measurements at home. The CMOS chip technology typical in smartphone cameras can be useful here, as it can be applied as a tool for optical foot measurements. Experts at the Stralsund-based Steinbeis Transfer Center for Image Processing and Medical Technology have developed a method to accurately, quickly and easily determine foot length and width using conventional smartphones.
Biometric foot measurements are important when it comes to selecting the right footwear, especially for kids. When trying shoes on for size, children often can’t give reliable feedback regarding the fit of the shoe. That’s because the bones of growing children’s feet are soft and malleable, reducing their sensitivity to pain or discomfort: but shoes that don’t fit right can cause foot damage early on – damage that could have been prevented. Well-fitting footwear offers the foot plenty of room to expand under pressure, without nudging the inside of the shoe. Also, if the shoe fits in terms of width, it gives the foot the right support while not constricting free movement. This is why many shoe manufacturers have developed sizing options for a variety of widths: one shoe length for several widths.
So it’s very important to determine the right shoe size when buying a new pair. This is particularly challenging in the case of online shoe purchases. Retail sites now generally offer various options for determining the best fit possible: anything from instructions for measuring with paper, pencil and a ruler, to separate sizing guide printouts – some even recommend holding the foot up to the screen and tracing the outline with the mouse/cursor. But the new smartphone app developed by professor Hans-Heino Ehricke and his team at the Steinbeis Transfer Center for Image Processing and Medical Technologies offers a more elegant and more precise option. It works as follows: first, a sheet of paper is laid on the floor with one edge to a wall. The foot to be measured is then placed on the sheet of paper so that the heel touches the wall. This defines a measuring point for the heel of the foot. Next, only the forefoot is captured with the smartphone camera. The image analysis software integrated in the app then determines the precise length of the paper’s edge, in addition to the tip of the foot and the outermost medial and lateral points. These measurements are then used to calculate the foot’s length and width. The paper serves as a calibration aid, so that distances between pixels in the image can be converted to metric measurements.
The biggest challenge for the image analysis experts was dealing with various constraints such as image lighting, structure and shape of the background, color and patterns of socks, or quality and focal distance of camera optics. Among other things, the project team developed an algorithm for removing glare caused by the camera’s flash hitting the ground. The fruit of their efforts is a method that can be used to accurately measure feet with a precision of 1 to 3 mm. As CMOS chip technology develops, even better results will be available in the future.
The app also offers the option of integrating various shoe sizing scales. This way, users can have the measurement data displayed not only based on the metric system, but in accordance with any desired shoe sizing system. Converting a shoe size to another sizing system can be difficult if the exact foot length is not known. This is most often caused by rounding errors. And it’s why common conversion charts (as those found on most shoe retailer websites) aren’t generally very helpful. However, if the precise foot length is known, the foot can be classified accurately within a sizing system, helping customers find just the right fit.
In 2011, nearly 12 million smartphones and 2 million tablet PCs were sold in Germany. With statistics like that, it’s feasible that foot measurements determined by smartphone can become a standard method for sizing – a method that will be employed by conventional shoe retailers as well. And that’s why the apps developers applied to patent the method last year.
In addition to foot measurements, the proper measuring and labeling of footwear is another important aspect of shoe retail. Three-dimensional recognition of shoe geometry using computer tomography is already being implemented by shoe manufacturer’s as a modern day quality assurance measure. But this isn’t a simple solution: the automated recognition of image data presents its challenges; especially when entire shoe collections have to be measured. The Steinbeis team in Stralsund worked with several partners from the footwear industry and shoe retail to develop a viable technique and software solution. These solutions may help to standardize and automate image analysis to a great extent.
The interplay between the two image-based methods for foot and footwear measurements can improve the way a shoe fits. Simultaneously, modern technology can maximize customer satisfaction in online sales and improve sales by reducing the number of products returned.
Professor Dr. Hans-Heino Ehricke
Steinbeis Transfer Center for Image Processing and Medical Technology (Stralsund)
su0384@stw.de