Image analysis system for testing in manufacturing
A vision analytics and validation (VAV) system for providing an improved inspection of robotic assembly, the VAV system comprising a trained neural network three-way classifier, to classify each component as good, bad, or do not know, and an operator station configured to enable an operator to review an output of the trained neural network, and to determine whether a board including one or more “bad” or a “do not know” classified components passes review and is classified as good, or fails review and is classified as bad. In one embodiment, a retraining trigger to utilize the output of the operator station to train the trained neural network, based on the determination received from the operator station.
1. A vision analytics and validation (VAV) system for providing an improved inspection of a robotic assembly system comprising:
an image combiner to combine a plurality of images from one or more camera-based inspection machines, the image combiner to create a unified image of a board;
a grid and ROI logic to create a grid aligned with the board, and to divide the unified image into regions of interest including a component and associated pads;
a processor implementing a trained neural network three-way classifier, to classify each component as good, bad, or do not know using the regions of interest; and
a clustering system to implement a clustering model wherein a cluster of data defining “good” components is built using statistical features of the “good” components, and wherein an object which is not within the cluster of data is classified as bad.
2. The VAV system of claim 1 , wherein the neural network attaches a probability to the classification for the component, and when the probability is below a decision threshold, the neural network classifies the component as do not know.
3. The VAV system of claim 2 , wherein the probability decision threshold is between 80% and 98%.
4. The VAV system of claim 2 , wherein the probability is adjustable.
5. The VAV system of claim 1 , wherein the plurality of images include one or more of: images with different lighting, images from different angles, and X-ray images at different depths.
6. The VAV system of claim 1 , further comprising optical character recognition (OCR) and element match to identify the component on the board based on optical recognition.
7. The VAV system of claim 1 , further comprising masking logic to mask extraneous information within a viewing window, prior to passing data to the classifier.
8. The VAV system of claim 1 , further comprising:
a synthetic training data generator to create synthetic training data based on the cluster of data, representing components within the cluster of data that do not occur in the training data.
9. The VAV system of claim 1 , further comprising:
an overtagging trigger and alert system to determine whether the VAV system classifies more than a preset percentage of components in the bad and the do not know classifications, which are overridden by an operator.
10. A vision analytics and validation (VAV) system for providing an improved inspection of a robotic assembly system comprising:
an image combiner to combine a plurality of images from one or more camera-based inspection machines, the image combiner to create a unified image of a board;
a grid and ROI logic to create a grid aligned with the board, and to divide the unified image into regions of interest including a component and associated pads;
a clustering system to implement a clustering model wherein a cluster of data defining “good” components is built using statistical features of the “good” components, and wherein an object which is not within the cluster of data is classified as bad;
a processor implementing a trained neural network three-way classifier, to classify each component as good, bad, or do not know using the regions of interest, wherein the neural network attaches a probability to the classification for the component, and when that probability is below a decision threshold, the trained neural network to classify the component as do not know.
11. The VAV system of claim 10 , wherein the probability decision threshold is between 80% and 98%.
12. The VAV system of claim 10 , wherein the probability is adjustable.
13. The VAV system of claim 10 , wherein the plurality of images include one or more of: images with different lighting, images from different angles, and X-ray images at different depths.
14. The VAV system of claim 10 , further comprising optical character recognition (OCR) and element match to identify the component on the board based on optical recognition.
15. The VAV system of claim 10 , further comprising masking logic to mask extraneous information within a viewing window, prior to passing data to the classifier.
16. The VAV system of claim 10 , further comprising:
a synthetic training data generator to create synthetic training data based on the cluster of data, representing the components within the cluster of data that do not occur in the training data.
17. The VAV system of claim 10 , further comprising:
an overtagging trigger and alert system to determine whether the VAV system classifies more than a preset percent of the components in the bad and the do not know classification, which are overridden by an operator.
18. A vision analytics and validation (VAV) system for providing an improved inspection of a robotic assembly system comprising:
an image combiner to combine a plurality of images from one or more camera-based inspection machines, the image combiner to create a unified image of a board including one or more components;
a grid and ROI logic to create a grid aligned with the board, and to divide the unified image into regions of interest including a component and associated pads;
a processor implementing a trained neural network three-way classifier, to classify each component as good, bad, or do not know using the regions of interest, wherein the neural network attaches a probability to the classification for the component, and when that probability is below a decision threshold classifies the component as do not know.
19. The VAV system of claim 18 , further comprising:
a clustering system to implement a clustering model wherein a cluster of data defining “good” components is built using statistical features of the “good” components, and wherein an object which is not within the cluster of data is classified as bad.
20. The VAV system of claim 18 , further comprising:
optical character recognition (OCR) and element match to identify the component on the board based on optical recognition; and
masking logic to mask extraneous information within a viewing window, prior to passing data to the classifier.