Method and system for vision-based defect detection

Patent metadata

Field	Value
Publication number	US-11636583-B2
Application number	US-202117328775-A
Country	US
Kind code	B2
Filing date	May 24, 2021
Priority date	Oct 21, 2019
Publication date	Apr 25, 2023
Grant date	Apr 25, 2023

Abstract

A method and a system for vision-based defect detection are proposed. The method includes the following steps. A test audio signal is outputted to a device-under-test (DUT), and a response signal of the DUT with respect to the test audio signal is received to generate a received audio signal. Signal processing is performed on the received audio signal to generate a spectrogram, and whether the DUT has an unacceptable defect with respect to the predefined auditory standard is determined by analyzing a distribution of signal strength according to the spectrogram.

First claim

What is claimed is:

1. A method for vision-based defect detection of a device-under-test (DUT) comprising:
outputting a test audio signal, wherein the test signal continuously ranges from a high frequency to a low frequency, over the passage of time;
receiving a response signal that is generated with respect to the test audio signal;
converting the received response signal to a received audio signal;
performing signal processing on the received audio signal to generate a spectrogram;
determining whether a defect exists by analyzing a distribution of signal strength, over a preset frequency range, according to the spectrogram;
in response to determining that the DUT has the defect, determining whether the defect is unacceptable with respect to a predefined auditory standard through computer vision according to the spectrogram;
wherein the step of determining whether the defect is unacceptable with respect to the predefined auditory standard through computer vision according to the spectrogram comprises:
obtaining a plurality of sub-spectrograms associated with the spectrogram;
transforming each of the sub-spectrograms into a projection curve;
obtaining a plurality of segments associated with each of the projection curves;
generating a feature quantification result corresponding to each of the segments of each of the projection curves; and
determining whether the DUT has the unacceptable defect according to the feature quantification results and a classifier.

2. The method according to claim 1 , wherein the step of receiving the response signal that is generated with respect to the test audio signal comprises:
receiving the response signal by using a microphone; and
performing analog-to-digital conversion on the response signal to generate the received audio signal.

3. The method according to claim 2 , wherein the step of performing signal processing on the received audio signal to generate the spectrogram comprises:
performing fast Fourier transform (FFT) on the received audio signal to generate the spectrogram.

4. The method according to claim 1 , wherein the step of determining whether the defect exists by analyzing the distribution of signal strength according to the spectrogram comprises:
determining whether the DUT has the defect through computer vision according to the spectrogram.

5. The method according to claim 1 , wherein the step of obtaining the sub-spectrograms associated with the spectrogram comprises:
extracting a region of interest (ROI) from the spectrogram, wherein the ROI corresponds to a preset frequency range; and
dividing the ROI with respect to different levels of frequencies to generate the sub-spectrograms.

6. The method according to claim 1 , wherein the step of transforming each of the sub-spectrograms into the projection curve comprises:
averaging energy values at each time in each of the sub-spectrograms to generate the projection curve.

7. The method according to claim 1 , wherein the step of obtaining the segments associated with each of the projection curves comprises:
dividing each of the projection curves into the segments with respect to different time intervals.

8. The method according to claim 1 , wherein the feature quantification result corresponding to each of the segments of each of the projection curves are associated with a plurality of statistical parameters of data points in the corresponding segment and a weight designated to the corresponding sub-spectrogram.

9. The method according to claim 1 , wherein the step of determining whether the DUT has the unacceptable defect according to the feature quantification results and the classifier comprises:
inputting the feature quantification results corresponding to all of the segments of all of the projection curves into the classifier;
receiving an output result of the classifier; and
determining whether the DUT has the unacceptable defect according to the output result of the classifier.

10. The method according to claim 9 , wherein the classifier is a support vector machines (SVM) classifier that is constructed based on a plurality of defective training objects with acceptable defects with respect to the predefined auditory standard.

11. The method according to claim 9 , wherein the step of determining whether the DUT has the unacceptable defect according to the output result of the classifier comprises:
obtaining a defect confidence level as the output result;
determining whether the defect confidence level is greater than a preset defect confidence threshold;
in response to the defect confidence level being greater than the preset defect confidence threshold, determining that the DUT has the acceptable defect; and
in response to the defect confidence level not being greater than the preset defect confidence threshold, determining that the DUT has the unacceptable defect.

12. The method according to claim 1 , wherein the DUT is an electronic device having a speaker.

13. The method according to claim 1 , wherein the defect is rub and buzz of the DUT.

14. A defect detection system comprising:
a signal outputting device, configured to output a test audio signal, wherein the test signal continuously ranges from a high frequency to a low frequency, over the passage of time;
a microphone, configured to receive a response signal with respect to the test audio signal;
an analog-to-digital converter, configured to convert the response signal to a received audio signal; and
a processing device, configured to perform signal processing on the received audio signal to generate a spectrogram and determine whether a device-under-test (DUT) has an unacceptable defect by analyzing a present frequency range with respect to the predefined auditory standard through computer vision according to the spectrogram;
wherein the processing device further pre-stores a classifier, and the processing device obtains a plurality of sub-spectrograms associated with the spectrogram, transforms each of the sub-spectrograms into a projection curve, obtains a plurality of segments associated with each of the projection curves, generates a feature quantification result corresponding to each of the segments of each of the projection curves, and determines whether the DUT has the unacceptable defect according to the feature quantification results and the classifier.

15. The system according to claim 14 , wherein the processing device determines whether the DUT has a defect through computer vision according to the spectrogram and determines whether the defect is unacceptable with respect to the predefined auditory standard through computer vision according to the spectrogram in response to the determination that the DUT has the defect.

16. The system according to claim 14 , wherein the DUT is an electronic device having a speaker.

17. The system according to claim 14 , wherein the defect is rub and buzz of the DUT.

Assignees

• Wistron Corp

Inventors

• Yang Jie-Ci
• Kao Meng-Chao
• LIU Wen-Kuong
• Lee Tsang-Hsing
• Pan Li-Ping

Classifications

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• H04R29/008

  Visual indication of individual signal levels (visual indication of stereophonic sound image H04S7/40) · CPC title

• G06T2207/20068

  Projection on vertical or horizontal image axis · CPC title

• H04R29/001

  for loudspeakers (H04R29/007 takes precedence) · CPC title

• G10L25/51

  for comparison or discrimination · CPC title