G06V 20/90

Definition

(Für diese Definition ist die deutsche Übersetzung noch nicht abgeschlossen)

Identifying an image sensor based on characteristic sensor noise patterns, sensor imperfections, artifacts, or optical defects. Defective pixels which, individually, are normally not perceptible to the human eye may be detected, and the repeatability of their occurrence at the same spatial position may be used for sensor/camera identification.

Notes – technical background

These notes provide more information about the technical subject matter that is classified in this place:

The process of digital camera identification may involve three steps:

1. Photo response non-uniformity (PRNU) noise extraction. The PRNU-pattern from the image under investigation is extracted using a denoising filter;

2. Extraction of sensor pattern noise (SPN), also known as the camera fingerprint, is obtained by taking a series of flat-field images with the camera under investigation. From each image the PRNU-pattern is extracted and then these patterns are combined to estimate the SPN;

3. Comparison. The SPN-pattern of the camera and the PRNU-pattern of the image are compared by calculating for example a correlation metric.

Examples

Determining whether the two images are taken by the same camera, by implementing the basic three-step process described above.

Querverweise

Informative Querverweise

Glossar