Skip to content

[Standard]: 6.2.5.2 - incorrect description of RVU algorithm #415

@LarsKooijman

Description

@LarsKooijman

Member Body/National Committee

AS ISO/IEC 29794.4:2025

Clause/Subclause

6.2.5.2.

Paragraph/Figure/Table

No response

Type of Comment

General

Comments

In the norm it is stated that:

compute $q^{RVU}_{local}$ as the ratio of the width of ridges against valleys in $S'$.

This implies that the local variable $q_{local}^{RVU}$ is a scalar, whereas in the NFIQ2 code it is a vector of ratios. In the NIST code, as well as Olsen's code, this local vector is appended to a 'global' vector, which is subsequently averaged to get $q^{\mu}_{RVU}$.

The implication that $q_{local}^{RVU}$ is a scalar is reinforced by the fact that other local quality metrics are scalar values (e.g., $q_{local}^{OFL}$, $q_{local}^{OCL}$), which are averaged to obtain a global value.

You will get a different outcome compared to the norm if you compute the average of the local ratios to obtain a scalar $q_{local}^{RVU}$ and then compute the average of all the local $q_{local}^{RVU}$ to get $q_{RVU}^{\mu}$. As such, the norm should clarify that in this case, $q_{local}^{RVU}$ is a vector.

Proposed Change

Rewrite k) as follows:

Compute $q^{RVU}_{local, i}$ for each ridge–valley pair $i \in {1, \dots, N}$ as the ratio of ridge-width to valley-width.

Metadata

Metadata

Assignees

Labels

standardText updates to ISO/IEC 29794-4:2024

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions