We can't tell you or show anything about cases where we cooperated, but we can provide some typical example. The needs are different for any requests, so it is often needed to develop ad hoc software for special cases. We also suggest to see Five samples page to visually evaluate some of the possible results.

Extracting a fingerprint from the background

We have a fingerprint of a suspect on a banknote, but it is impossible to compare it to the database because it is mixed with the watermark of the banknote. We can separate the two information and thus extract only the fingerprint for further recognition.

Measuring the height of the suspects from a picture

We have pictures of the crime scene, but as always the faces are not clear enough for a sustainable recognition. We can improve them, correcting the blur or low resolution, always within physical limits. But to add more information we can compare it with the length of other object in the scenes, even from a single picture, and measure their height, providing also a reliable error margin based on the geometry of the scene.

Extracting a license plate from some low quality frames

This is one of the most common tasks and, unfortunately, many times the quality is really too bad to do anything. But, if some minimum conditions are met, sometime is possible to obtain unexpected results. If you can see the model of the car even the possibility to have three numbers of the license plate is very useful.

Measuring the size of a vehicle

A car has been involved in a crime. The quality is so bad that you can't even understand the model of the car. But thanks to other checks you have two suspects, one on an utility car and one on a van. We can compare street sizes and obtain approximatively the size of a car, helping to discriminate the model.

Analysis of a scene in a complex system

Very often analog systems record the signal coming from different cameras on the same tape. In this case it is very difficult to analyze a scene because you always need to search for the next frame of the camera of interest. We can separate the different scenes and then analyze more efficiently what's happening, then proceeding to analyze and enhance the video.

Stabilizing operative activity video

Operative departments of law enforcement usually film in difficult conditions, often leading to shaking video and other problems. We can compensate the motion and enhancing the overall features of the video to obtain a better visualization of the events.

Facial comparison

We have a frame where there is the face of the responsible of a robbery. We have a suspect but we are not sure about his identity, also because usually surveillance devices optics strongly distort the image. What we can do is calibrate the camera that filmed the crime, and with our software video mixer compare the crime scene picture with the real time acquisition of the suspect, compensating or simulating camera distortion and automatically saving most similar images. Then we can easily compare facial features on saved images.