Structural characteristic prediction for a honeycomb body using image abstraction
A method and system for inspecting a honeycomb body. The method includes capturing a first image. Instances of at least one feature in the first image that correlates to a structural characteristic of the honeycomb body are detected. One or more detected instances of the at least one feature identified in the first image are abstracted by creating a graphical representation of each of the one or more detected instances of the at least one feature. A second image is generated by augmenting the first image with the graphical representation in place of or in addition to each of the one or more detected instances of the at least one feature identified in the first image. The second image is analyzed using a machine learning algorithm to classify the honeycomb body with respect to the structural characteristic of the honeycomb body.
1. A method of inspecting a honeycomb body, comprising:
capturing a first image;
detecting one or more instances of at least one feature in the first image, the at least one feature correlating to a structural characteristic of the honeycomb body;
abstracting one or more detected instances of the at least one feature identified in the first image by creating a graphical representation of each of the one or more detected instances of the at least one feature;
generating a second image by augmenting the first image with the graphical representation in place of or in addition to each of the one or more detected instances of the at least one feature identified in the first image; and
analyzing the second image using a machine learning algorithm to classify the honeycomb body with respect to the structural characteristic of the honeycomb body.
2. The method of claim 1 , wherein the generating the second image further comprises reducing a resolution of the first image such that a first resolution of the first image is greater than a second resolution of the second image.
3. The method of claim 1 , wherein the at least one feature relates to at least one of an estimated stress, a web thickness, a broken cell wall, an angled cell wall, or a bent cell wall.
4. The method of claim 1 , wherein the abstracting the first image comprises characterizing each of the one or more detected instances of the at least one feature identified in the first image by at least one of a feature type of each of the one or more detected instances, a magnitude quantifying each of the one or more detected instances, or a location on the honeycomb body of each of the one or more detected instances.
5. The method of claim 4 , wherein the graphical representation comprises a color, a shape, a size, an orientation, a pattern, or a combination thereof, corresponding to at least one of the feature type, the magnitude, or the location on the honeycomb body.
6. The method of claim 1 , wherein the detecting each of the one or more detected instances of the at least one feature in the first image comprises collecting measurement data of the honeycomb body via machine vision software.
7. The method of claim 1 , wherein the first image has a resolution of at least about 10 megapixels and each of the one or more detected instances of the at least one feature has a size of at most about 1000 pixels.
8. The method of claim 1 , wherein each of the one or more detected instances of the at least one feature has a size of at most about 0.001% of a resolution of the first image.
9. The method of claim 1 , wherein the machine learning algorithm is a deep neural network, and the deep neural network is a convolutional neural network.
10. The method of claim 1 , wherein the structural characteristic comprises an isostatic strength of the honeycomb body.
11. The method of claim 1 , wherein the structural characteristic comprises an accuracy with respect to a target shape of the honeycomb body.
12. The method of claim 1 , further comprising determining whether the honeycomb body has passed or failed inspection by comparing a predicted value of the structural characteristic to a target threshold value of the structural characteristic.
13. The method of claim 12 , further comprising displaying a result of the determining whether the honeycomb body has passed or failed inspection.
14. The method of claim 12 , further comprising moving the honeycomb body to a first area if the determining results in a pass and to a second area if the determining results in a fail.
15. A method of manufacturing a honeycomb body comprising inspecting a honeycomb body according to the method of claim 1 after extruding the honeycomb body from an extruder, the method further comprising controlling operation of one or more components of the extruder in response to the inspecting.
16. The method of claim 15 , wherein controlling the operation of one or more components of the extruder comprises moving a plate of a flow control mechanism of the extruder radially with respect to an axis of the extruder, or injecting a liquid addition into the extruder to adjust a rheology of a batch mixture from which the honeycomb body is formed.
17. An imaging system for inspecting a honeycomb body comprising:
a camera configured to capture a first image of an end face of the honeycomb body;
a controller in signal communication with the camera, the controller comprising a machine learning algorithm, the controller configured to:
receive the first image from the camera;
detect one or more instances of at least one feature of the honeycomb body in the first image, wherein the at least one feature correlates to a structural characteristic of the honeycomb body;
abstract the first image by creating a graphical representation of one or more detected instances of the at least one feature;
generate a second image comprising the graphical representation in place of or in addition to the one or more detected instances of the at least one feature; and
analyze the second image using a machine learning algorithm to classify the honeycomb body with respect to the structural characteristic of the honeycomb body.
18. The imaging system of claim 17 , wherein the structural characteristic comprises an isostatic strength or an accuracy with respect to a target shape of the honeycomb body.
19. The imaging system of claim 17 , wherein the first image has a first resolution and the second image has a second resolution that is lower than the first resolution.
20. The imaging system of claim 17 , wherein the controller is configured to create the graphical representation by characterizing the feature by at least one of a feature type of each of the one or more detected instances, a magnitude quantifying each of the one or more detected instances, or a location of each of the one or more detected instances on the honeycomb body.