Comparison of ID performance using real and synthetic imagery

Jacobs, Eddie L.; Edwards, Thomas M.; Miller, Brian; Hodgkin, Van A.

doi:10.1117/12.477465

Cited by 2 publications

(12 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Figure 7a reproduces the results from Jacobs et al 2 . This figure show the proportion of correctly identified targets (PID) as a function of the viewing distance.…”

Section: Resultssupporting

confidence: 78%

“…In both cases the thresholds corresponding to the original synthetic data are higher than the reference data, but the corrected synthetic data matches the reference data. Figure 8 shows the relationship between PID (measured by Jacobs et al 2 ) and identifiability (in pixels). PID-score is plotted against identifiability, averaged over the two observers.…”

Section: Resultsmentioning

confidence: 99%

“…The resulting 100 reference images were also simulated twice, using two different parameter settings of the simulation program. In their validation experiment Jacobs et al 2 found that the set of parameters that was initially used to generate the synthetic images was not entirely complete, and additional parameters were needed. They therefore generated another set of corrected synthetic images.…”

Section: Methodsmentioning

confidence: 99%

“…Jacobs et al 2 recently performed an experiment in which they compared identification performance on nominally identical sets of real and synthetic thermal imagery. Real imagery representing notional first generation and advanced scanning sensor systems was obtained.…”

Section: Validation Experimentsmentioning

confidence: 99%

“…An example of these images is shown in Figure 5. For details about the image registration and generation procedures we refer to Jacobs et al 2 In this study we use a subset of their imagery and the corresponding identification data to validate our new identifiability metric.…”

Section: Validation Experimentsmentioning

confidence: 99%

See 4 more Smart Citations

Conspicuity and identifiability: efficient calibration tools for synthetic imagery

Toet

Hogervorst

Bijl

2004

Targets and Backgrounds X: Characterization and Representation

View full text Add to dashboard Cite

We argue that visual conspicuity and identifiability are two efficient task-related measures that can be deployed to calibrate synthetic imagery that is intended to be used for human visual search and detection tasks. The conspicuity of a target is operationally defined as the region around the center of the visual field where the target is capable to attract visual attention. Visual conspicuity predicts human visual search performance in realistic and military relevant complex scenario's. Conspicuity can easily and quickly be measured either in the field (in complex environments) or in the lab. This eliminates the need for costly and time consuming visual search experiments. The agreement between field and lab measurements implies that conspicuity can be used to validate synthetic imagery. Target identifiability is operationally defined as the amount of Gaussian blur that is required to reduce the target signature to its identification threshold. It is an efficient metric that can be used to gain insight into human identification performance without having to resort to elaborate and costly experiments. Identifiability is directly related to PID-performance, and is therefore well suited for comparing synthetic and realistic imagery. We conclude that synthetic imagery can be calibrated for human visual search and detection tasks by setting the conspicuity and identifiability of targets equal to those of their real world counterparts.

show abstract

“…Figure 7a reproduces the results from Jacobs et al 2 . This figure show the proportion of correctly identified targets (PID) as a function of the viewing distance.…”

Section: Resultssupporting

confidence: 78%

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Validation Experimentsmentioning

confidence: 99%

Section: Validation Experimentsmentioning

confidence: 99%

See 3 more Smart Citations

Conspicuity and identifiability: efficient calibration tools for synthetic imagery

Toet

Hogervorst

Bijl

2004

Targets and Backgrounds X: Characterization and Representation

View full text Add to dashboard Cite

show abstract

<title>Identifiability: a fast way to measure identification performance</title>

et al. 2004

Self Cite

View full text Add to dashboard Cite

We present a new measure called target identifiability, as an efficient alternative for measuring identification scores. Identifiability is operationally defined as the amount of blur required to reduce the target signature to its identification threshold. It can quickly be determined using a simple adjustment procedure. To validate the new measure, we measured the identifiability of targets in a set of real and simulated thermal images. The identification scores for these targets were available from a previous study. Our results show that identifiability indeed determines identification performance. Sufficient accuracy can be obtained with only a few (typically 2 or 3) trained observers. The associated measurement procedure is simple and requires only a limited amount of time.

show abstract

Comparison of ID performance using real and synthetic imagery

Cited by 2 publications

References 0 publications

Conspicuity and identifiability: efficient calibration tools for synthetic imagery

Conspicuity and identifiability: efficient calibration tools for synthetic imagery

<title>Identifiability: a fast way to measure identification performance</title>

Contact Info

Product

Resources

About