#qoncept

Or what I could at least discern from their technology page from their site.

Since the first time Ice Scope was used to measure figure skating jumps at Japanese Nationals, many FS fans are curious about how does Ice Scope (will be referred as "the program" onwards within this paragraph) measure the jumps, or whether one can trust the accuracy of the measurements gotten from this program. While I cannot answer for the second question (ie. the accuracy of the measurement), I can answer the first part of the question, namely the technology the program is based upon. As written in the subtitle (😛), I explain the technology using the information available on the site page linked above.

(I mean I am sure there should be some white papers lying around the vast wide web but 1. they are most definitely written in Japanese and 2. my Japanese skills isn't that far advanced yet, so I cannot write here more information than that 😫)

Technology

So, from what I was able to discern from the page:

Of all 3 technologies Qoncept has to offer, two of them (4D tracker and Visual SLAM) are the base for the Ice Scope measurement. 4D tracker is for tracking things in real time and they use two ways to track the things — real-time tracking and predictive tracking (ie. the program predicted the trajectory of the thing they track from a model — IMO model here most likely refers to a type of mathematical formula, to say it broadly). They can even track small things (not specified how small — in the example image of volleyball tracking they were able to track a volleyball with a recorded diameter (in the image) of 15 to 20 pixels). After the tracking the data is mapped out onto a 3D space (eg. the site page shows a 3D reconstruction of volley ball trajectory tracking).

In the case of Ice Scope, this software is used to track the skaters' blade (eg. the blade tip, the parts that show the trajectory of the jump).

Example of the 4D tracking technology used to track the trajectory of a volleyball

The planes in the 3D space itself is derived from the Visual SLAM which can detect and map surfaces (eg. the orientation of the surface, size, etc.) from 2D visual recording of the data, in real time no less (seriously, this is the remarkable part of the software). The video resolution they use for the measurement is 1920x1080 square pixels with 60 fps rate. It is written on the page that they accept customized *order*, though I am not sure if this means they can adjust this software for video specs lower than the one they stated on the site page. 

In the case of Ice Scope, this part of software is used to map the ice rink surface.

Look how Visual SLAM is able to extract plane information from a 2D image data by 1. Acquiring the feature points (cyan dots), 2. Mapping the relation between the feature points (green line).

Measurement Flow

Ice Scope is a measurement system, and every measurement system is composed by various stages of process, from data recording (ie. what is usually named unit sensing) to measurement result display. In Ice Scope, the measurement flow is likely to be:

2D Image recording → data acquisition by the computer → data processing (Visual SLAM) | data processing (4D tracking)* → calculation (eg. the jump exit speed, height, distance) → data display

Example of Ice Scope data display, in this case measuring the throw pairs jump (throw 3F)

*this process can be executed simultaneously (especially if the hardware has enough capacity), but in the display processing block, the order of processing would be likely be Visual SLAM first, and then the 4D tracking.

That's all for the explanation about the technology the Ice Scope is based upon. Feel free to send me asks if you have any questions and I might write a follow-up post about how to possibly assess accuracy of Ice Scope.