W3C home > Mailing lists > Public > public-credentials@w3.org > September 2021

RE: Rotationally Invariant Digital Fingerprint (RIDF) for Photos of Flat Objects? [RE: Rotationally Invariant Digital Fingerprint (RIDF) for Images?]

From: Michael Herman (Trusted Digital Web) <mwherman@parallelspace.net>
Date: Mon, 6 Sep 2021 15:13:50 +0000
To: Leonard Rosenthol <lrosenth@adobe.com>, "public-credentials (public-credentials@w3.org)" <public-credentials@w3.org>
Message-ID: <MWHPR1301MB209470A2204070AACFF6B308C3D29@MWHPR1301MB2094.namprd13.prod.outlook.com>
Thank you Leonard,

Also note because these are flat objects (e.g. crackers, slices of toast, etc.), not all of the solutions need to deal with the "2D geometry" of the image ...other dimensions such as the spectral histogram also look interesting/promising in terms of generating a unique digital fingerprint (for these classes of natural objects).

Other promising search terms include:

  *   Recognizing slightly rotated scaled photos
  *   Digital fingerprinting photos (differentiated from digital watermarking of a photo)
  *   Shazam algorithm
  *   SIFT - https://en.wikipedia.org/wiki/Scale-invariant_feature_transform

Thank you again.

Have a great day,
Michael

From: Leonard Rosenthol <lrosenth@adobe.com>
Sent: September 6, 2021 7:54 AM
To: Michael Herman (Trusted Digital Web) <mwherman@parallelspace.net>; public-credentials (public-credentials@w3.org) <public-credentials@w3.org>
Subject: Re: Rotationally Invariant Digital Fingerprint (RIDF) for Photos of Flat Objects? [RE: Rotationally Invariant Digital Fingerprint (RIDF) for Images?]

In general, the space of  "image fingerprinting", "perceptual hashing", etc. is a very active and evolving space with the recent moves from classic/heuristic approaches to more AI/ML-based ones.  If you want the latest in the field, search the literature.  If you just want to something to play with - search for terms like "pHash".

Leonard

From: Michael Herman (Trusted Digital Web) <mwherman@parallelspace.net<mailto:mwherman@parallelspace.net>>
Date: Sunday, September 5, 2021 at 6:09 PM
To: public-credentials (public-credentials@w3.org<mailto:public-credentials@w3.org>) <public-credentials@w3.org<mailto:public-credentials@w3.org>>
Subject: Rotationally Invariant Digital Fingerprint (RIDF) for Photos of Flat Objects? [RE: Rotationally Invariant Digital Fingerprint (RIDF) for Images?]
Retitled to: Rotationally Invariant Digital Fingerprint (RIDF) for Photos of Flat Objects?

From: Michael Herman (Trusted Digital Web)
Sent: September 5, 2021 2:23 PM
To: public-credentials (public-credentials@w3.org<mailto:public-credentials@w3.org>) <public-credentials@w3.org<mailto:public-credentials@w3.org>>
Subject: Rotationally Invariant Digital Fingerprint (RIDF) for Images?

Does anyone have any experience with this or know of a solution for computing an RIDF (rye-diff)?

Problem Statement

1. Alice takes a digital photo Pa of a relatively flat physical object T (e.g. a cracker or a slice of toasted bread)
2. Locally, Alice processes the photo Pa to compute a value RIDF(Pa)
3. Alice sends the physical object T to Bob
4. Bob receives the object T from Alice
5. Bob uses a similar camera to take a digital photo Pb of T ... at a slightly different distance and camera angle than Alice used
6. Bob, local to Bob, using the same software as Alice, processes Pb to compute a value RIDF(Pb)

Question

What kind of algorithm can be used so that RIDF(Pb) is equal, within some sort of standard deviation, to RIDF(Pa)?  Ideally, they're exactly equal to some level of precision.

Some sample images are attached. Assume each cracker is it's own image file. 3 images of each cracker.

Michael Herman


Get Outlook for Android<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Faka.ms%2FAAb9ysg&data=04%7C01%7Clrosenth%40adobe.com%7C7c57c2dc27994ce6c6b308d970b9ccd7%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C637664765580314778%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=qZRLylZbkPzmtCiTvFENZR7c6YE2FrpdCW5sAWKdKRs%3D&reserved=0>
Received on Monday, 6 September 2021 15:14:05 UTC

This archive was generated by hypermail 2.4.0 : Monday, 6 September 2021 15:14:07 UTC