Methods and arrangements for assessing camera system performance
Performance metrics for an imaging system are assessed in the field, without the need for carefully-controlled conditions, such as lighting. In some embodiments, a hand-held test target is used to measure MTF, and to identify any undesired noise-reduction operation applied by an image signal processor. A metric-based heuristic enables prediction of whether an imaging system will be suitable to detect and decode a digital watermark signal of a particular resolution. A variety of other features and arrangements are also detailed.
1 . A method employing an optical test card to assess performance of a camera system, the optical test card comprising a substrate with plural regions including a first region printed with a uniform grey tone and a second region printed with a pattern, the pattern comprising a noise pattern or a 2D machine readable code, the method comprising:
receiving imagery depicting said uniform grey tone in the first region and the pattern in the second region, the received imagery corresponding to image data previously captured by the camera system from the optical test card;
determining a first quality metric for the received imagery, in which the first quality metric indicates correlation between (a) depiction of the uniform grey tone, and (b) depiction of the pattern;
determining a second quality metric for the received imagery based on depiction of the pattern; and
combining the first quality metric and the second quality metrics to yield a composite quality metric, the composite quality metric indicating performance of the camera system.
2 . The method of claim 1 in which said combining comprises computing a ratio between the first quality metric and the second quality metric.
3 . The method of claim 1 in which the second quality metric indicates correlation between depiction of the pattern, and ground truth image data for the pattern.
4 . The method of claim 3 in which the ground truth image data comprises original sample values for the pattern, unaffected by printing and imaging.
5 . The method of claim 4 wherein the second quality metric is a normalized cross-correlation metric.
6 . The method of claim 5 in which the normalized cross-correlation metric is a signal presence metric of the form:
N
tf
[
i
,
j
]
=
∑
m
∑
n
(
(
f
[
m
,
n
]
-
μ
f
)
×
(
t
[
m
-
i
,
n
-
j
]
-
μ
t
)
)
∑
m
∑
n
t
2
[
m
-
i
,
n
-
j
]
where f is the received imagery depicting the pattern; t is a patch excerpt from the ground truth image data for the pattern; i,j are dimensions of the patch t; m,n are pixels in the patch t, μ f , μ t are the average values for f and t respectively, and N is the neighborhood over which the score is maximized.
7 . The method of claim 1 wherein the pattern is a continuous tone digital watermark.
8 . The method of claim 1 wherein the pattern is a continuous tone chrominance digital watermark or a continuous tone luminance digital watermark.
9 . The method of claim 1 that includes:
determining a homography that expresses viewpoint distortion of the optical test card as depicted in image data captured by the camera system;
resampling the image data captured by the camera system in accordance with said determined homography, said resampling yielding resampled image data; and
employing the resampled image data as said received imagery.
10 . A method employing an optical test card to assess performance of a camera system, the optical test card comprising a substrate with plural regions including a first region printed with a uniform grey tone and a second region printed with a pattern, the pattern comprising a noise pattern or a 2D machine readable code, the method comprising:
receiving imagery depicting said uniform grey tone in the first region and the pattern in the second region, the imagery corresponding to image data previously captured by the camera system from the optical test card;
filtering the imagery from the camera system with a non-linear filter, in which said filtering comprises, for each of plural pixels in the imagery, assigning a filtered value to said each of plural pixels based on a count of neighboring pixels that have an original value higher than an original value of said each of plural pixels, said filtering yielding filtered imagery;
determining a first quality metric for the filtered imagery based on depiction of the uniform grey tone;
determining a second quality metric for the filtered imagery based on depiction of the pattern; and
combining the first quality metric and the second quality metric to yield a composite quality metric, the composite quality metric indicating performance of the camera system.
11 . An optical test card useful in assessing performance of a camera system, the optical test card comprising:
a substrate with plural regions including:
(i) a first region printed with a uniform grey tone having an area of at least one square inch;
(ii) a second region printed with a noise-like pattern having an area of at least one square inch; and
(iii) a third region printed with a black area and an adjoining white area defining a knife-edge boundary that spans the third region; and
at least one machine-readable marker printed on the substrate that enables determination of viewpoint distortion of imagery depicting the optical test card and encodes metadata about the optical test card.
12 . The optical test card of claim 11 in which the noise-like pattern is comprises a digital watermark pattern.
13 . The optical test card of claim 11 in which said third region includes an area of at least one square inch.