Inspection systems and methods for determining matches between sets of images of rotary parts
Devices and methods are provided herein useful to aligning or linking together data from multiple inspections of a part. In some embodiments, the inspection methods include detecting key points on inspection data associated with the part, such as images. Key points provide the locations of features of interest on the part. Key points for the part are represented by feature descriptors. The inspection methods then link together data from two or more inspections by matching the feature descriptors and calculating a distance between matched key points. The inspection methods identify the proper alignment of inspection data from two or more inspections by identifying the alignment that achieves a minimum distance between matched key points on the inspection data. In this manner, the inspection methods align or link together inspection data from two or more different inspections to enable the comparative analysis of inspection data for a part.
1 . An inspection system comprising:
one or more inspection devices;
a control unit having at least one processor and at least one memory device, the at least one memory device storing instructions that when executed by the at least one processor causes the at least one processor to perform operations, the at least one processor configured to:
receive a first set of ordered images from the one or more inspection devices, each image in the first set of ordered images corresponding to a particular part of a plurality of ordered parts, images in the first set of ordered images being ordered in a fixed arrangement relative to each other;
analyze the first set of ordered images to detect a plurality of first groups of key points, each first group of key points associated with one of the first set of ordered images;
extract a plurality of first groups of feature descriptors, each first group of feature descriptors associated with one of the key points in a first group of key points;
receive a second set of ordered images from the one or more inspection devices, the second set of ordered images including images of at least some of the plurality of ordered parts, wherein it is unknown to which particular part of the plurality of ordered parts each image in the second set of ordered images corresponds, images in the second set of ordered images being ordered in a fixed arrangement relative to each other, wherein the second set of ordered images and the first set of ordered images define a plurality of possible orientations for aligning the second set of ordered images relative to the first set of ordered images;
analyze the second set of ordered images to detect a plurality of second groups of key points, each second group of key points associated with one of the second set of ordered images;
extract a plurality of second groups of feature descriptors, each second group of feature descriptors associated with one of the key points in a second group of key points;
determine Euclidean distances between feature descriptors in the plurality of first groups of feature descriptors and feature descriptors in the plurality of second groups of feature descriptors for the plurality of possible orientations;
for each first group of key points, determine matches between key points in the first group of key points and key points in each second group of key points based on the Euclidean distances to obtain a plurality of matched key points for the plurality of possible orientations;
for each image in the first set of ordered images, determine a sum of the Euclidean distances for the plurality of matched key points for the plurality of possible orientations; and
link images in the first set of ordered images with images in the second set of ordered images based on the sums of the Euclidean distances for the plurality of matched key points for the plurality of possible orientations.
2 . The inspection system of claim 1 , wherein the one or more inspection devices includes at least one of a camera, an eddy current instrument, an infrared detector, a computerized tomography scanner, a flexible borescope, a rigid borescope, a flexible endoscope, a rigid endoscope, and an X-ray fluorescence spectrometer.
3 . The inspection system of claim 1 , wherein the plurality of ordered parts are rotary parts, and wherein at least one of the one or more inspection devices is coupled to a stationary component and configured to capture images of the rotary parts.
4 . The inspection system of claim 3 , wherein the plurality of ordered parts are stationary parts, and wherein the at least one of the one or more inspection devices is coupled to a rotary component and configured to capture images of the stationary parts.
5 . The inspection system of claim 1 , wherein the plurality of ordered parts are airfoils.
6 . The inspection system of claim 1 , wherein linking images includes matching images in the first set of ordered images with particular images in the second set of ordered images to generate matched images that correspond to the same part.
7 . The inspection system of claim 1 , wherein the at least one processor is further configured to determine that at least one image in the first set of ordered images and at least one image in the second set of ordered images are not properly matched based on the sum of the Euclidean distances.
8 . An inspection method comprising:
acquiring a first set of inspection data comprising a first plurality of ordered images with each image in the first plurality of ordered images corresponding to a particular part of a plurality of ordered parts, images in the first plurality of ordered images being ordered in a fixed arrangement relative to each other;
analyzing the first set of inspection data to detect a plurality of first groups of key points, each first group of key points associated with inspection data for one of the plurality of ordered parts;
extracting a plurality of first groups of feature descriptors, each first group of feature descriptors associated with one of the key points in a first group of key points;
acquiring a second set of inspection data comprising a second plurality of ordered images of at least some of the plurality of ordered parts, wherein it is unknown to which particular part of the plurality of ordered parts each image in the second plurality of ordered images corresponds, images in the second plurality of ordered images being ordered in a fixed arrangement relative to each other, wherein the second plurality of ordered images and the first plurality of ordered images define a plurality of possible orientations for aligning the second plurality of ordered images relative to the first plurality of ordered images;
analyzing the second set of inspection data to detect a plurality of second groups of key points, each second group of key points associated with inspection data for one of the plurality of ordered parts;
extracting a plurality of second groups of feature descriptors, each second group of feature descriptors associated with one of the key points in a second group of key points;
determining Euclidean distances between feature descriptors in the plurality of first groups of feature descriptors and feature descriptors in the plurality of second groups of feature descriptors for the plurality of possible orientations;
for each first group of key points, determining matches between key points in the first group of key points and key points in each second group of key points based on the Euclidean distances to obtain a plurality of matched key points for the plurality of possible orientations;
for each first group of key points, determining a sum of the Euclidean distances for the plurality of matched key points for the plurality of possible orientations; and
linking images in the first set of inspection data with images in the second set of inspection data based on the sum of the Euclidean distances for the plurality of matched key points for the plurality of possible orientations.
9 . The inspection method of claim 8 , wherein the first set of inspection data and the second set of inspection data are captured via one or more inspection devices including at least one of a camera, an eddy current instrument, an infrared detector, a computerized tomography scanner, a flexible borescope, a rigid borescope, a flexible endoscope, a rigid endoscope, and an X-ray fluorescence spectrometer.
10 . The inspection method of claim 8 , wherein the inspection data includes an image of at least a portion of at least one of the plurality of ordered parts.
11 . The inspection method of claim 8 , wherein the first group of key points and the second group of key points are identified using at least one of: a scale-invariant feature transform (SIFT) algorithm, a speeded up robust features (SURF) algorithm, a features from accelerated segment test (FAST) algorithm, a binary robust independent elementary features (BRIEF) algorithm, an oriented FAST and rotated BRIEF (ORB) algorithm, and a superpoint algorithm.
12 . The inspection method of claim 8 , wherein the second set of inspection data is acquired at a time after the first set of inspection data is acquired.
13 . The inspection method of claim 8 , wherein the Euclidean distances are calculated in feature space.
14 . The inspection method of claim 8 , wherein the first group of key points and the second group of key points are indicative of features on at least some of the plurality of ordered parts.
15 . The inspection method of claim 8 , wherein the first groups of feature descriptors include numeric identifiers indicative of a location in an assembly relative to a reference unique to a first inspection; and wherein the second groups of feature descriptors include numeric identifiers indicative of a location in the assembly relative to a reference unique to a second inspection.
16 . The inspection method of claim 8 , wherein the plurality of ordered parts are parts of a rotational assembly, and wherein the inspection method further includes determining a clocking position that defines a rotational position of the parts of the rotational assembly to link the images in the first set of inspection data with the images in the second set of inspection data.
17 . The inspection method of claim 8 , wherein matching includes comparing at least one feature descriptor in the first groups of feature descriptors to each of the feature descriptors in the second groups of feature descriptors.
18 . The inspection method of claim 8 , further comprising:
determining a rate of change or an expected life for the particular part based on a portion of the first set of inspection data and a portion of the second set of inspection data.
19 . An inspection method comprising:
acquiring a first set of ordered images including at least a first image of a first part and a first image of a second part;
analyzing the first set of ordered images to detect a first group of key points for the first part and a first group of key points for the second part;
extracting a first group of feature descriptors for the first group of key points for the first part and a first group of feature descriptors for the first group of key points for the second part;
acquiring a second set of ordered images including at least a second image of the first part and a second image of the second part, wherein it is unknown to which of the first part and the second part each image in the second set of ordered images corresponds, images in the second set of ordered images being ordered in a fixed arrangement relative to each other, wherein the second set of ordered images and the first set of ordered images define a plurality of possible orientations for aligning the second set of ordered images relative to the first set of ordered images;
analyzing the second set of ordered images to detect a second group of key points for the first part and a second group of key points for the second part;
extracting a second group of feature descriptors for the first group of key points for the first part and a second group of feature descriptors for the second group of key points for the second part;
determining a set of Euclidean distances for the plurality of possible orientations, the set of Euclidean distances including a Euclidean distance between the feature descriptors of each key point in the first group of key points for the first part and the feature descriptors of each key point in the second group of key points for the first part and the feature descriptors of each key point in the second group of key points for the second part;
determining matches between key points in the first group of key points for the first part and key points in the second group of key points for the first part and key points in the second group of key points for the second part based on the set of Euclidean distances to obtain a plurality of matched key points for the plurality of possible orientations;
determining a sum of the set of Euclidean distances for the plurality of matched key points for the plurality of possible orientations; and
linking the first image of the first part with the second image of the first part and the first image of the second part with the second image of the second part based on the sums of the Euclidean distances for the plurality of matched key points for the plurality of possible orientations.
20 . The inspection method of claim 19 , further comprising:
determining a match between the first image of the second part and the second image of the second part based on the sum of the set of Euclidean distances.
21 . The inspection method of claim 19 , further comprising:
comparing the first image of the first part with the second image of the first part to determine a rate of change or remaining life for the first part.
22 . The inspection method of claim 19 , further comprising:
analyzing the first image of the first part and the second image of the first part using one or more automated defect detection algorithms.