Relative performance of two-dimensional speckle-tracking techniques: normalized correlation, non-normalized correlation and sum-absolute-difference

The authors present computer simulations that compare the performance of three two-dimensional block-matching algorithms as applied to ultrasonic speckle. The three algorithms, normalized correlation, non-normalized correlation and sum absolute difference (SAD), were applied to simulated speckle patterns with varying levels of noise and with different kernel sizes. The results show that SAD and normalized correlation have similar performance characteristics and are able to track speckle motion with a kernel size as small as one resolution cell. Non-normalized correlation performed poorly with small kernel sizes but improved as the kernel size increased.