Comments about the coincident bit counting (CBC) criterion for image registration

Area of Science:

• Computer Vision
• Image Processing
• Signal Processing

Background:

• Nonparametric approaches offer robust image registration by not relying on specific pixel values.
• Previous work introduced stochastic sign change (SSC) and deterministic sign change (DSC) criteria for image registration.
• Chiang and Sullivan recently proposed the coincident bit counting (CBC) criterion.

Purpose of the Study:

• To comment on the accuracy of the comparison between CBC and prior criteria (SSC, DSC) made by Chiang and Sullivan.
• To highlight the potential interest and need for further study of the CBC criterion.
• To address perceived inaccuracies in the performance evaluation of image registration methods.

Main Methods:

• Comparison of the coincident bit counting (CBC) criterion against stochastic sign change (SSC) and deterministic sign change (DSC) criteria.
• Analysis of the nonparametric approach's performance in image registration.
• Evaluation of similarity measures that are independent of specific pixel values.

Main Results:

• The authors contend that the comparison of CBC with SSC and DSC by Chiang and Sullivan was not entirely accurate.
• The CBC method quantifies image registration by comparing coincident bits between pixels across frames for a given displacement.
• The underlying nonparametric approach, which disregards specific pixel values, is noted for its robustness.

Conclusions:

• The coincident bit counting (CBC) criterion warrants further investigation.
• The authors suggest that the comparative analysis presented by Chiang and Sullivan requires re-evaluation.
• Further studies are recommended to accurately assess the performance and utility of the CBC image registration method.