Method of multi-resolution adaptive correlation processing
First Claim
1. A method of processing at least first and second images of an image flow of an object to determine a relative displacement map of elements of the image flow over a predetermined time interval, the method comprising:
- recording a first array of pixel values associated with the first image of the image flow;
recording a second array of pixel values associated with the second image of the image flow;
defining a first plurality of interrogation regions on the first array of pixel values, each of the first plurality of interrogation regions including a first number of pixels indicative of a first resolution at which to correlate the first and second images;
determining a first correlation plane formed from correlations between each of the first plurality of interrogation regions and at least one associated region in the second array of pixel values, the first correlation plane including a maximum correlation value for each of the first plurality of interrogation regions indicating which of the at least one associated regions each of the first plurality of interrogation regions are most highly correlated with;
determining a direction and magnitude associated with each of the maximum correlation values which represents the relative displacement between each of the first interrogation regions and the respective most highly correlated associated region in the second array of pixels over the predetermined time interval;
for each of the determined directions and magnitudes that are resolved, adding each resolved direction and magnitude to the displacement map at a location associated with the respective first interrogation region;
for each of the determined directions and magnitudes that are unresolved, expanding the associated first interrogation region to form a second plurality of interrogation regions, each of the second plurality of interrogation regions having a second number of pixels, greater than the first number of pixels, indicative of a second resolution at which to correlate the first and second images and determining a second correlation plane formed from correlations between each of the second plurality of interrogation regions and one or more associated regions in the second array of pixel values, the second correlation plane including a maximum correlation value for each of the second plurality of interrogation regions indicating which of the at least one associated regions each of the second plurality of interrogation regions are most highly correlated with, and determining a direction and magnitude associated with each of the maximum correlation values which represents the relative displacement between each of the second interrogation regions and the respective most highly correlated associated region in the second array of pixels over the predetermined time interval; and
for each of the determined directions and magnitudes that are resolved, adding each resolved direction and magnitude to the displacement map at a location associated with the respective second interrogation region.
1 Assignment
0 Petitions
Accused Products
Abstract
A method of multi-resolution adaptive correlation processing of images, such as seeded fluid flow images, to efficiently increase the spatial resolution and dynamic range of detecting particle image displacements in the images. The technique takes full advantage of the multi-resolution characteristic of the discrete correlation function by starting the processing at the smallest scale and if necessary gradually building correlation planes into larger interrogation areas based on the result of inter-level correlation correction and validation. It is shown that the method can be implemented in both direct and FFT based correlation algorithms with greatly reduced computational complexity. Processing the images at the lowest scale (e.g. pixel or particle image size) allows the combination of correlation planes of various shapes both in space and in time for maximizing the correlation plane signal-to-noise ratio or for estimating statistical flow parameters.
21 Citations
9 Claims
-
1. A method of processing at least first and second images of an image flow of an object to determine a relative displacement map of elements of the image flow over a predetermined time interval, the method comprising:
-
recording a first array of pixel values associated with the first image of the image flow; recording a second array of pixel values associated with the second image of the image flow; defining a first plurality of interrogation regions on the first array of pixel values, each of the first plurality of interrogation regions including a first number of pixels indicative of a first resolution at which to correlate the first and second images; determining a first correlation plane formed from correlations between each of the first plurality of interrogation regions and at least one associated region in the second array of pixel values, the first correlation plane including a maximum correlation value for each of the first plurality of interrogation regions indicating which of the at least one associated regions each of the first plurality of interrogation regions are most highly correlated with; determining a direction and magnitude associated with each of the maximum correlation values which represents the relative displacement between each of the first interrogation regions and the respective most highly correlated associated region in the second array of pixels over the predetermined time interval; for each of the determined directions and magnitudes that are resolved, adding each resolved direction and magnitude to the displacement map at a location associated with the respective first interrogation region; for each of the determined directions and magnitudes that are unresolved, expanding the associated first interrogation region to form a second plurality of interrogation regions, each of the second plurality of interrogation regions having a second number of pixels, greater than the first number of pixels, indicative of a second resolution at which to correlate the first and second images and determining a second correlation plane formed from correlations between each of the second plurality of interrogation regions and one or more associated regions in the second array of pixel values, the second correlation plane including a maximum correlation value for each of the second plurality of interrogation regions indicating which of the at least one associated regions each of the second plurality of interrogation regions are most highly correlated with, and determining a direction and magnitude associated with each of the maximum correlation values which represents the relative displacement between each of the second interrogation regions and the respective most highly correlated associated region in the second array of pixels over the predetermined time interval; and for each of the determined directions and magnitudes that are resolved, adding each resolved direction and magnitude to the displacement map at a location associated with the respective second interrogation region. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
Specification