Apparatus for determining a surface condition of an object

• US 20060029257A1
• Filed: 07/13/2005
• Published: 02/09/2006
• Est. Priority Date: 08/03/2004
• Status: Abandoned Application

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Murakami, Manabu, Eguchi, Junji
• Application Number:
  US11/181,643
• Publication Number:
  US 20060029257A1
• Time in Patent Office: Days
• US Class Current: 382/108
• CPC Class Codes: G06T 2207/30164 Workpiece; Machine component
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/60 Analysis of geometric attributes
• CPC Class Codes: G06T 7/64 of convexity or concavity

Abstract

The present invention provides an apparatus and a method for determining a surface condition of an inspected object. A potential region having an intensity different from the other regions by more than a predetermined threshold value in an object image captured by an imaging device is detected. An inspected region surrounding the potential region is identified. A feature for a parameter is extracted from the inspected region. The surface condition is determined based on the feature. The parameter includes one or more of an area of the potential region, a slope of the potential region, an intensity value entropy of the inspected region, an intensity value anisotropy of the inspected region, an average intensity value of an edge image of the inspected region, and a roundness of the potential region.

Claims

1. An apparatus using a computer for determining a surface condition of an inspected object, the apparatus comprising:

an imaging device for imaging the inspected object; and

an image processing unit configured to;

detect a potential region in the image of the inspected object, the potential region having an intensity different from the other regions in the image by more than a predetermined threshold value;

identify an inspected region surrounding the potential region;

extract a feature for a predetermined parameter from the inspected region; and

determine the surface condition based on the feature.

2. The apparatus of claim 1, wherein the parameter includes one or more of an area of the potential region, a slope of the potential region, an intensity value entropy of the inspected region, an intensity value anisotropy of the inspected region, an average intensity value of an edge image of the inspected region, and a roundness of the potential region.

3. The apparatus of claim 1, wherein the inspected object is a gear.

4. An apparatus using a computer for determining a surface condition of an inspected object, the inspected object having a plurality of consecutive units of similar texture on the surface, the apparatus comprising:

an imaging device for sequentially imaging the consecutive units on the surface of the inspected object while rotating the inspected object relatively to the imaging device; and

an image processing unit configured to;

determine a differential image between a current image and a previous image in the consecutive images;

detect a potential region having an intensity exceeding a predetermined threshold value in the differential image;

identify an inspected region surrounding the potential region;

extract a feature for a plurality of predetermined parameters from the inspected region; and

determine the surface condition based on the feature.

5. An apparatus using a computer for generating a data map to be used for determining a surface condition of an inspected object, the inspected object having a plurality of consecutive units of similar texture on the surface, the apparatus comprising an image processing unit configured to:

extract a feature vector from each of a plurality of samples of an inspected region of the inspected object;

self-organizing learn through use of the feature vectors to generate a self-organizing map;

group neurons that correspond to each same sample and are adjacent to each other into a cluster in the self-organizing map; and

classify one or more clusters as corresponding to a predetermined surface condition and the other one or more clusters as not corresponding to the predetermined surface condition to generate the data map.

6. The apparatus of claim 5, wherein the predetermined surface condition is a defect,

wherein the data map is generated by classifying one or more clusters as corresponding to the defect and the other one or more clusters as corresponding to an over-detection.

7. The apparatus of claim 5, further comprising an imaging device for sequentially imaging the consecutive units on the surface of the inspected object while rotating the inspected object relatively to the imaging device;

wherein the image processing unit is further configured to perform a determination using the data map, the determination including;

determining a differential image between a current image and a previous image in the images of the consecutive units;

detecting a potential region having an intensity exceeding a predetermined threshold value in the differential image;

identifying an inspected region surrounding the potential region;

extracting a feature vector for parameters from the inspected region; and

calculating a distance between the feature vector and a coupling coefficient vector of each neuron in the data map;

determining a region of adjacent neurons having a small distance; and

determining the surface condition based on the number of neurons that are in the determined region and belong to one of the clusters corresponding to the predetermined surface condition.

8. The apparatus of claim 7, wherein the image processing unit is further configured to, if a ratio of the number of neurons that are in the determined region and belong to one of the clusters corresponding to the predetermined surface condition to the number of all neurons in the determined region is greater than a predetermined threshold value, determine that the inspected region includes the predetermined surface condition.

9. The apparatus of claim 7, wherein the feature vector includes one or more of an area of the potential region, a slope of the potential region, an intensity value entropy of the inspected region, an intensity value anisotropy of the inspected region, an average intensity value of an edge image of the inspected region, and a roundness of the potential region.

10. The apparatus of claim 5, wherein the inspected object is a gear.

11. An apparatus for determining a surface condition of an inspected object, the apparatus comprising:

a storage device for storing a plurality of data maps provided for respective parts of the inspected object, each of the data maps learning a surface condition of the corresponding part of the inspected object; and

an imaging device for imaging the inspected object; and

an image processing unit configured to;

identify an inspected region in the image of the inspected object, the inspected region including a potential region that has an intensity different from the other regions in the image by more than a predetermined threshold value;

extract a feature vector from the inspected region;

identify a part of the inspected object to which the inspected region belongs;

select a data map corresponding to the identified part; and

input the feature vector into the selected data map to determine whether the inspected region has a predetermined surface condition.

12. The apparatus of claim 11, wherein the storage device stores a first data map that learns the predetermined surface condition in a first part of the inspected object, the first part being imaged over the entire inspected region, and a second data map that learns the predetermined surface condition in a second part of the inspected object, the second part being imaged in a portion of the inspected region.

13. The apparatus of claim 11, wherein the inspected object is a gear,

wherein the storage device stores a first data map that learns the predetermined surface condition in a center part of the gear, the center part being imaged over the entire inspected region, and a second data map that learns the predetermined surface condition in an end part of the inspected object, the second part being imaged in a portion of the inspected region.

14. The apparatus of claim of claim 11, wherein the image processing unit is further configured to generate a data map for each part of the inspected object,

wherein the generation of the data map includes:

(a) defining a self-organizing map for each part of the inspected object,

(b) preparing a plurality of samples of the inspected region, the preparing further including, for each of the samples of the inspected region,

(b1) extracting a feature vector from the inspected region;

(b2) identifying a part to which the inspected region belongs to select the self-organizing map corresponding to the part;

(b3) inputting the feature vector into the selected self-organizing map for learning;

(c) grouping adjacent neurons that correspond to each same sample into a cluster in the self-organizing map; and

(d) classifying one or more clusters as corresponding to the predetermined surface condition and the other one or more clusters as not corresponding to the predetermined surface condition to generate the data map.

15. An apparatus for determining a surface condition of an inspected object, the apparatus comprising:

an imaging device for sequentially imaging a surface of the inspected object while rotating the inspected object relatively to the imaging device; and

an image processing unit configured to;

detect a potential region having an intensity different from the other regions in each image by more than a predetermined threshold value;

identify an inspected region surrounding the detected potential region; and

determine that a predetermined surface condition is included in the inspected region when the inspected region is identified at the same position on the inspected object over a plurality of consecutive images captured by the imaging device.

16. The apparatus of claim 15, wherein the image processing unit is further configured to:

if the inspected region is identified at the same position on the inspected object over more than a predetermined number of consecutive images, determine that the predetermined surface condition is included in the inspected region;

for each inspected region other than the inspected region determined as including the predetermined surface condition, extract a feature from the inspected region for a plurality of parameters; and

determine whether the inspected region includes the predetermined surface condition based on the extracted feature.

17. The apparatus of claim 15, wherein the inspected object is a gear,

wherein the image processing unit is further configured to determine that the predetermined surface condition is included in the inspected region when the inspected region is identified at the same position in a tooth width direction of the gear over a plurality of consecutive images captured by the imaging device.

18. An apparatus for automatically determining a cause of a surface condition of an inspected object, the apparatus comprising:

a storage device for storing a cause seeking map in which neurons are clustered for each of causes of the surface condition in a form of a self-organizing map;

an imaging device for imaging the inspected object; and

an image processing unit configured to;

detect a potential region having an intensity different from the other regions in the image by more than a predetermined threshold value;

identify an inspected region surrounding the potential region;

determine whether the inspected region includes a predetermined surface condition;

identify position information of the inspected region determined as including the predetermined surface condition;

extract, from the position information, a position vector representing a position of the predetermined surface condition on the inspected object;

input the extracted position vector into the cause seeking map;

identify a neuron in the cause seeking map, the neuron having a minimum distance between the position vector that has been input into the cause seeking map and a coupling coefficient vector for the neuron; and

determine a cause of the predetermined surface condition in accordance with a cluster to which the identified neuron belongs.

19. The apparatus of claim 18, wherein the image processing unit is further configured to generate the cause seeking map by inputting a position vector prepared for learning, the position vector for learning identifying a position of the predetermined surface condition for each of causes of the predetermined surface condition.

20. The apparatus of claim 18, wherein the inspected object is a gear,

wherein the position vector is expressed by respective positions in a gear tooth width direction.