CN108428247B - Method and system for detecting direction of soldering tin point - Google Patents
Method and system for detecting direction of soldering tin point Download PDFInfo
- Publication number
- CN108428247B CN108428247B CN201810164009.4A CN201810164009A CN108428247B CN 108428247 B CN108428247 B CN 108428247B CN 201810164009 A CN201810164009 A CN 201810164009A CN 108428247 B CN108428247 B CN 108428247B
- Authority
- CN
- China
- Prior art keywords
- image
- soldering tin
- solder
- area
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
- G06T7/75—Determining position or orientation of objects or cameras using feature-based methods involving models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention relates to a method and a system for detecting the direction of a soldering tin point. The method comprises the following steps: acquiring a soldering tin image of a soldering tin point, and dividing the soldering tin image into a main body area and an external area according to the position of the soldering tin point in the soldering tin image; acquiring sub-images to be detected of the soldering tin images in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction; obtaining the probability of the sub-image to be tested being directional and non-directional by utilizing a pre-trained soldering tin point direction detection model; and if the probability that the sub-graph to be tested has the directivity is greater than the probability that the sub-graph to be tested does not have the directivity, judging that the direction of the soldering tin point is in the direction corresponding to the sub-graph to be tested. By the method and the system for detecting the direction of the soldering tin point, the accuracy and precision of the detection of the direction of the soldering tin point are improved.
Description
Technical Field
The invention relates to the technical field of automatic optical detection, in particular to a method and a system for detecting the direction of a soldering tin point.
Background
Automatic Optical Inspection (AOI) is a necessary step in an industrial manufacturing process, and detects foreign substances or surface defects by obtaining a surface state of a finished product in an optical manner and performing image processing. Soldering tin defect detection is a common application in the field of automatic optical detection, and a machine automatically scans a circuit board through a camera to obtain an image, automatically extracts a local image of each welding spot, judges the direction of a soldering tin point at the welding spot through an image processing technology, judges whether a defect exists according to the direction of the soldering tin point, and finally displays or marks the soldering tin suspected of the defect, so that the checking and the maintenance are convenient.
The defect that comparatively ripe AOI equipment is used for detecting the paster mostly in the market, it is still comparatively weak to plug-in components class soldering tin equipment, and traditional technique of detecting plug-in components class soldering tin point direction is based on ordinary image processing technique, and the effective soldering tin proportion of discernment each region judges soldering tin point direction, and the accuracy is lower, produces the erroneous judgement of soldering tin point direction easily.
Disclosure of Invention
Accordingly, it is necessary to provide a method and a system for detecting a solder dot direction with high accuracy and accurate judgment, in order to solve the problems of low detection accuracy and easy erroneous judgment.
A method for detecting the direction of a soldering tin point comprises the following steps:
acquiring a soldering tin image of a soldering tin point, and dividing the soldering tin image into a main body area and an external area according to the position of the soldering tin point in the soldering tin image;
acquiring sub-images to be detected of the soldering tin images in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction;
obtaining the probability of the sub-image to be tested being directional and non-directional by utilizing a pre-trained soldering tin point direction detection model;
and if the probability that the sub-graph to be tested has the directivity is greater than the probability that the sub-graph to be tested does not have the directivity, judging that the direction of the soldering tin point is in the direction corresponding to the sub-graph to be tested.
In one embodiment, the step of acquiring the to-be-tested subgraph of the solder image in at least one direction includes:
setting a detection area in at least one direction of the soldering tin image, and acquiring a to-be-detected subgraph of the soldering tin image in the direction corresponding to the detection area; wherein the detection region includes a partial image of the main region in one direction and a partial image of the outer region in the same direction.
In one embodiment, the step of providing the detection area in at least one direction of the solder image comprises:
selecting one direction as a reference direction of the soldering tin image;
and according to the reference direction, soldering point detection areas are respectively arranged in the directions which are separated from the reference direction by preset angles.
In one embodiment, the step of providing the solder point sensing areas in directions spaced apart from the reference direction by a predetermined angle includes:
8 solder point detection areas are provided in directions spaced apart from the reference direction by 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively.
In an embodiment, before the step of obtaining the probability that the sub-image to be tested has directionality and non-directionality by using the pre-trained solder point direction detection model, the method further includes the steps of:
and establishing an untrained solder point direction detection model based on a deep convolutional network theory and according to preset initial parameters, and performing classification training on the solder point direction detection model by using a plurality of training subgraphs of historical solder images to obtain a trained solder point direction detection model.
In one embodiment, the step of dividing the solder image into a body region and an outer region comprises:
determining the position of a soldering tin point in a soldering tin image, acquiring a first area taking the soldering tin point as a center, and acquiring a second area which covers the soldering tin point and is larger than the first area;
the first area is used as the main area of the solder image, and the area obtained by subtracting the first area from the second area is used as the outer area of the solder image.
In one embodiment, after the step of determining the direction of the solder point in the direction corresponding to the sub-image to be tested, the method further includes the steps of:
and if the direction of the soldering tin points is not matched with the direction of the soldering tin points preset in the soldering tin standard domain image, judging that the soldering tin points have the continuous soldering defect.
In one embodiment, before the step of obtaining the probability that the sub-image to be tested is directional or non-directional by using the pre-trained solder point direction detection model, the method further includes the steps of:
and screening the sub-graphs to be detected in the direction needing to be detected from the sub-graphs to be detected according to the direction of the preset soldering points of the soldering tin standard domain image.
A solder dot orientation detection system comprising:
the area dividing module is used for acquiring a soldering tin image of a soldering tin point and dividing the soldering tin image into a main area and an external area according to the position of the soldering tin point in the soldering tin image;
the subgraph acquisition module is used for acquiring subgraphs to be detected of the soldering tin image in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction;
the probability calculation module is used for acquiring the probability of the sub-image to be tested, which is directional or non-directional, by using a pre-trained soldering tin point direction detection model;
and the direction judgment module is used for judging the direction of the soldering tin points, and if the probability of directivity of the sub-image to be tested is greater than the probability of non-directivity, the direction of the soldering tin points is judged to be in the direction corresponding to the sub-image to be tested.
According to the method and the system for detecting the direction of the soldering tin point, the to-be-detected subgraphs of the soldering tin images in different directions are input into the pre-trained soldering tin point direction detection model, and the direction of the soldering tin point is judged according to directional and non-directional probability output of the to-be-detected subgraphs in all directions, so that the accuracy and precision of the direction detection of the soldering tin point are improved.
A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method for detecting the orientation of solder dots as described above when executing the computer program.
According to the computer equipment, through the computer program running on the processor, the directional and non-directional probability output of the sub-image to be detected in each direction is realized, the direction of the soldering tin point is judged, and the accuracy and precision of the direction detection of the soldering tin point are improved.
A computer storage medium has stored thereon a computer program which, when executed by a processor, implements the method of detecting the orientation of solder dots as described above.
The computer storage medium realizes the judgment of the direction of the soldering tin points according to the directional and non-directional probability output of the subgraph to be tested in each direction through the stored computer program, and improves the accuracy and precision of the direction detection of the soldering tin points.
Drawings
Fig. 1 is a flowchart of a solder dot orientation detection method according to an embodiment;
FIG. 2 is a schematic diagram of a division of a solder image into a main area and an outer area according to an embodiment;
FIG. 3 is a schematic view of one embodiment of solder dots having directionality in one direction;
FIG. 4 is a schematic view of an exemplary solder dot test area arrangement;
fig. 5 is a schematic configuration diagram of a solder dot direction detection system according to an embodiment.
Detailed Description
To facilitate an understanding of the invention, the invention will now be described more fully with reference to the accompanying drawings. Referring to fig. 1, fig. 1 shows a flowchart of a solder dot direction detection method according to an embodiment, which mainly includes the following steps:
step S10: the method comprises the steps of obtaining a soldering tin image of soldering tin points, and dividing the soldering tin image into a main body area and an external area according to the positions of the soldering tin points in the soldering tin image.
In this step, the solder image is a real-time image of the solder points to be detected, which is obtained by a camera of the detection device, and the main area and the external area of the solder image are divided according to the positions of the solder points in the solder image; the main area is the area where normal solder is located, and the outer area, if solder exists and is connected with the solder in the main area in a certain direction, indicates that the solder point has directionality.
Referring to fig. 2, fig. 2 is a schematic diagram illustrating division of a solder image into a main area and an outer area according to an embodiment. The step of dividing the solder image into the body region 200 and the outer region 300 includes: determining the position of a soldering tin point 100 in a soldering tin image, acquiring a first area taking the soldering tin point 100 as a center, and acquiring a second area which covers the first area and is larger than the first area; the first area is used as the main area 200 of the solder image, and the area obtained by subtracting the first area from the second area is used as the outer area 300 of the solder image.
Specifically, the position of the solder point 100 in the solder image is determined, according to the parameter setting of the standard format of the solder, a first area conforming to the preset parameter setting size is obtained by taking the solder point 100 as the center, the first area is determined as a main area 200, the main area is expanded outwards to obtain a second area covering the main area 200 and conforming to the preset parameter size, and the extension part of the second area excluding the main area is set as an outer area 300.
Referring to fig. 3, fig. 3 shows a schematic view of one embodiment in which the solder dots have directionality in one direction.
As shown in fig. 3, most of the solder 101 of the solder dot falls in the main area 200, and if the solder 102 exists in the outer area 300 in the 90 ° direction and the solder 101 in the main area 200 and the solder 102 in the outer area communicate with each other in this direction, it is determined that the direction of the solder dot 100 is in the 90 ° direction.
Step S20: acquiring sub-images to be detected of the soldering tin images in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction.
In this step, a local image of a main body region in a certain direction and a local image of an external region in the same direction as the local image of the main body region are taken as a to-be-detected subgraph, and the to-be-detected subgraph in at least one direction is obtained.
In one embodiment, the step of acquiring the to-be-tested subgraph of the solder image in at least one direction includes: setting a detection area in at least one direction of the soldering tin image, and acquiring a to-be-detected subgraph of the soldering tin image in the direction corresponding to the detection area; wherein the detection region includes a partial image of the main region in one direction and a partial image of the outer region in the same direction.
Specifically, on the whole solder image, at least one direction can be selected and a detection area can be set in the direction, the size of the detection area is not limited, and the detection area can be set according to the detection requirement and the precision requirement; the detection area comprises a local image of the main area in the direction and a local image of the outer area in the direction, and images corresponding to the detection areas are extracted from the whole soldering tin image according to the arrangement and the division of the detection areas to be used as sub-images to be detected in the direction.
In one embodiment, the step of providing the detection area in at least one direction of the solder image comprises: selecting one direction as a reference direction of the soldering tin image; and according to the reference direction, soldering point detection areas are respectively arranged in the directions which are separated from the reference direction by preset angles.
Specifically, since different directions have independence, independent directivity determination can be performed for different directions, that is, a plurality of detection regions can be set; a reference direction can be selected, and a plurality of soldering tin point detection areas are respectively arranged in the direction which is separated from the reference direction by a certain preset angle; according to the embodiment, the overlapping of the detection areas can be avoided by setting the reference direction and the interval angle, and the detection efficiency is improved.
In one embodiment, the step of providing the solder point sensing areas in directions spaced apart from the reference direction by a predetermined angle includes: 8 solder point detection areas are provided in directions spaced apart from the reference direction by 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively.
Referring to fig. 4, fig. 4 shows a schematic view of a solder point detection area arrangement according to an embodiment.
Specifically, the reference direction is set to be 0 degree, and a detection area is set at 45 degrees every anticlockwise or clockwise interval until the whole soldering tin image is completely covered; as shown in fig. 4, detection regions are provided in the directions at intervals of 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 ° counterclockwise from the reference direction, and a detection region of 0 °, 45 °, a detection region of 90 °, a detection region of 135 °, a detection region of 180 °, a detection region of 225 °, a detection region of 270 °, and a detection region of 315 are obtained, respectively, the detection regions in the respective directions cover the entire solder image, and each detection region includes a partial image of the main body region 200 and a partial image of the outer region 300 in the same direction.
In the embodiment, the detection areas are arranged in 8 directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, so that the whole solder image is covered comprehensively, and the directionality of the solder points can be determined in each independent direction, thereby realizing comprehensive detection of the whole solder image and improving the accuracy and precision of the detection.
In an embodiment, sub-images to be detected in the direction needing to be detected can be screened out from the sub-images to be detected according to the preset direction of the soldering points of the soldering tin standard domain image.
Specifically, the to-be-tested subgraph of the soldering tin point can be obtained in the preset direction according to the preset direction of the soldering tin standard plate image, so that the to-be-tested subgraph in the preset direction is detected, and whether the direction of the soldering tin point is in the preset direction or not is judged. By screening the subgraph to be detected, targeted detection can be realized, and the detection efficiency is improved while the workload is reduced.
Step S30: and obtaining the probability of the sub-image to be tested to be directional and non-directional by using a pre-trained soldering tin point direction detection model.
In this step, the acquired sub-graph to be tested of the solder image is input into the pre-trained solder point direction detection model, the pre-trained solder point direction detection model can perform category judgment on the sub-graph to be tested, and a probability value that the sub-graph to be tested is a directional sub-graph and a probability value that the sub-graph to be tested is a non-directional sub-graph can be respectively output.
In one embodiment, before the step of obtaining the probability of the sub-image to be tested being directional and non-directional by using the pre-trained solder point direction detection model, the method further includes the steps of: and establishing an untrained solder point direction detection model based on a deep convolutional network theory and according to preset initial parameters, and performing classification training on the solder point direction detection model by using a plurality of training subgraphs of historical solder images to obtain a trained solder point direction detection model.
Specifically, an untrained solder point direction detection model is established based on a deep convolution theory and according to preset initial parameters, and the solder point direction detection model can be provided with related model parameters such as the number of convolution layers and a convolution algorithm according to accuracy requirements; after the model is built, the soldering tin point direction detection model is classified and trained by utilizing a large number of training subgraphs of historical soldering tin images collected in advance, and after the training of the training subgraphs, the soldering tin point direction detection model can output two classified probabilities, namely a directional probability and a non-directional probability.
In one embodiment, the solder point direction detection model may include 3 convolutional layers, and a pooling layer and a full link layer, where the output result of each convolutional layer enters the pooling layer for pooling after performing nonlinear transformation using a ReLU (normalized Linear Unit) and passes through the full link layer, and finally outputs a classification probability through a Softmax function.
In one embodiment, a large number of subgraphs of historical solder images can be collected as training subgraphs, i.e. training samples, according to a preset sampling standard; the training subgraph comprises a directional subgraph and a non-directional subgraph, the non-directional training subgraph can also be called as a positive sample, and the directional training subgraph can also be called as a negative sample; the steps of collecting positive and negative samples are: acquiring a plurality of historical soldering tin images, and dividing the historical soldering tin images into a main body area and an external area according to positions of soldering tin points in the historical soldering tin images; training subgraphs of historical solder images are acquired in multiple directions respectively, and the training subgraphs comprise local images of a main body region in one direction and local images of an outer region in the same direction. And after a sufficient number of training subgraphs are obtained, carrying out classification training on the soldering point direction detection model by using the training subgraphs of the historical soldering image to obtain the trained soldering point direction detection model.
Step S40: and if the probability that the sub-graph to be tested has the directivity is greater than the probability that the sub-graph to be tested does not have the directivity, judging that the direction of the soldering tin point is in the direction corresponding to the sub-graph to be tested.
In the step, whether the probability that the sub-graph to be tested belongs to the directivity is larger than the probability that the sub-graph to be tested belongs to the non-directivity is judged according to the classification probability output by the pre-trained soldering tin point direction detection model; and if so, judging that the direction corresponding to the sub-graph to be tested is the direction of the soldering tin point.
In one embodiment, after the step of determining the direction of the solder point in the direction corresponding to the sub-image to be tested, the method further includes the steps of: and if the direction of the soldering tin points is not matched with the direction of the soldering tin points preset in the soldering tin standard domain image, judging that the soldering tin points have the continuous soldering defect.
In the embodiment, after the actual direction of the soldering point is determined to be in the direction corresponding to a certain sub-image to be tested, the actual direction is compared with the preset direction which the soldering point is preset in the soldering standard domain image and should have, and if the actual direction is not consistent with the preset direction, the soldering point is determined to have the continuous soldering defect. Through the actual direction of contrast soldering tin point and the direction of the soldering tin point that soldering tin standard layout image was preset, can further realize the detection of continuous tin defect, improve the precision that the defect detected.
Furthermore, the sub-graphs to be detected of the soldering tin point in 8 directions can be detected respectively, the tin connection defect of the soldering tin point in which direction or directions is/are judged, comprehensive detection of the soldering tin defect can be realized, and the detection accuracy is further improved.
According to the method for detecting the direction of the soldering tin point, the to-be-detected subgraphs of the soldering tin images in different directions are input into the pre-trained soldering tin point direction detection model, and the direction of the soldering tin point is judged according to the directional and non-directional probability output of the to-be-detected subgraphs in all directions, so that the accuracy and precision of the direction detection of the soldering tin point are improved.
An embodiment of the solder dot direction detecting system according to the present invention will be described in detail with reference to the accompanying drawings, and fig. 5 is a schematic configuration diagram of the solder dot direction detecting system according to an embodiment.
A solder dot orientation detection system comprising: the device comprises a region dividing module 10, a subgraph acquiring module 20, a probability calculating module 30 and a direction judging module 40.
The area dividing module 10 is configured to acquire a solder image of a solder dot, and divide the solder image into a main area and an external area according to a position of the solder dot in the solder image;
a subgraph obtaining module 20, configured to obtain a subgraph to be tested of the solder image in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction;
the probability calculation module 30 is used for acquiring the probability that the sub-graph to be tested has directionality and non-directionality by using a pre-trained soldering tin point direction detection model;
and the direction judging module 40 is configured to judge the direction of the solder point, and if the probability that the sub-image to be tested has directivity is greater than the probability that the sub-image to be tested does not have directivity, judge that the direction of the solder point is in the direction corresponding to the sub-image to be tested.
In one embodiment, the area dividing module 10 may be further configured to determine a solder point position in the solder image, obtain a first area centered on the solder point, and obtain a second area that covers and is larger than the first area; the first area is used as the main area of the solder image, and the area obtained by subtracting the first area from the second area is used as the outer area of the solder image.
In an embodiment, the sub-image obtaining module 20 may be further configured to set a detection area in at least one direction of the solder image, and obtain a sub-image to be tested of the solder image in a direction corresponding to the detection area; wherein the detection region includes a partial image of the main region in one direction and a partial image of the outer region in the same direction.
In an embodiment, for the sub-image obtaining module 20, it may further be configured to select a direction as a reference direction of the solder image; and according to the reference direction, soldering point detection areas are respectively arranged in the directions which are separated from the reference direction by preset angles. According to the embodiment, the overlapping of the detection areas can be avoided by setting the reference direction and the interval angle, and the detection efficiency is improved.
In an embodiment, the sub-pattern obtaining module 20 may further be configured to set 8 solder point detection areas in directions separated from the reference direction by 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively. In the embodiment, the detection areas are arranged in 8 directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, so that the whole solder image is covered comprehensively, and the directionality of the solder points can be determined in each independent direction, thereby realizing comprehensive detection of the whole solder image and improving the accuracy and precision of the detection.
In an embodiment, the subgraph obtaining module 20 may be further configured to screen out the subgraph to be detected in the direction to be detected from the subgraphs to be detected according to the direction of the preset soldering points of the soldering standard domain image. According to the embodiment, targeted detection can be realized by screening the subgraphs to be detected, so that the detection efficiency is improved while the workload is reduced.
In an embodiment, the probability calculation module 30 may be further configured to establish an untrained solder point direction detection model based on a deep convolutional network theory and according to preset initial parameters, and perform classification training on the solder point direction detection model by using training subgraphs of a plurality of historical solder images to obtain a trained solder point direction detection model.
In an embodiment, the direction determining module 40 may be further configured to determine whether the solder points have a continuous solder defect, and determine that the solder points have the continuous solder defect if the direction of the solder points is not matched with a direction of solder points preset in the standard solder layout image. In the embodiment, the detection of the continuous tin defect can be further realized by comparing the actual direction of the tin soldering points and the direction of the preset tin soldering points of the standard tin soldering domain image, and the defect detection precision is improved.
According to the detection system for the direction of the soldering tin points, the to-be-detected subgraphs of the soldering tin images in different directions are input into the pre-trained soldering tin point direction detection model, and the direction of the soldering tin points is judged according to the directional and non-directional probability output of the to-be-detected subgraphs in all directions, so that the accuracy and precision of the direction detection of the soldering tin points are improved.
In an embodiment, there is also provided a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the method for detecting the solder dot orientation as in any of the above embodiments.
In the computer device, when the processor executes the program, the accuracy and precision of the solder dot direction detection can be improved by implementing any one of the solder dot direction detection methods in the above embodiments.
Furthermore, it can be understood by those skilled in the art that all or part of the processes in the methods for implementing the above embodiments can be implemented by using a computer program to instruct related hardware, where the program can be stored in a non-volatile computer-readable storage medium, and in the embodiments of the present invention, the program can be stored in the storage medium of a computer system and executed by at least one processor in the computer system to implement the processes of the embodiments including the method for detecting the directions of solder dots as described above.
In an embodiment, there is also provided a storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements any of the methods of solder dot orientation detection as described in the above embodiments. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
The computer storage medium and the computer program stored therein can improve the accuracy and precision of the solder dot direction detection by realizing the flow of the embodiment including the method for detecting the solder dot direction as described above.
Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features. The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for detecting the direction of a soldering tin point is characterized by comprising the following steps:
acquiring a soldering tin image of a soldering tin point, and dividing the soldering tin image into a main body area and an external area according to the position of the soldering tin point in the soldering tin image;
acquiring sub-images to be detected of the soldering tin images in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction;
obtaining the probability of the sub-image to be tested being directional and non-directional by utilizing a pre-trained soldering tin point direction detection model;
if the probability that the sub-image to be tested has the directivity is larger than the probability that the sub-image to be tested does not have the directivity, the direction of the soldering tin point is judged to be in the direction corresponding to the sub-image to be tested;
the step of obtaining the to-be-detected subgraph of the soldering tin image in at least one direction comprises the following steps of:
setting a detection area in at least one direction of the soldering tin image, and acquiring a to-be-detected subgraph of the soldering tin image in the direction corresponding to the detection area;
the step of providing a detection area in at least one direction of the solder image includes:
selecting one direction as a reference direction of the soldering tin image;
and according to the reference direction, soldering point detection areas are respectively arranged in the directions which are separated from the reference direction by preset angles.
2. The method for detecting the orientation of solder dots as set forth in claim 1, wherein the detection area includes a partial image of the main area in one direction and a partial image of the outer area in the same direction.
3. The method for detecting the direction of solder dots according to claim 1, wherein the step of providing solder dot detection areas in the directions spaced apart from the reference direction by a predetermined angle comprises:
8 solder point detection areas are provided in directions spaced apart from the reference direction by 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively.
4. The method for detecting the direction of solder points as claimed in claim 1, wherein before the step of obtaining the probability of the sub-image to be tested being directional or non-directional by using the pre-trained solder point direction detection model, the method further comprises the steps of:
and establishing an untrained solder point direction detection model based on a deep convolutional network theory and according to preset initial parameters, and performing classification training on the solder point direction detection model by using a plurality of training subgraphs of historical solder images to obtain a trained solder point direction detection model.
5. The method for detecting the orientation of solder dots as set forth in claim 1, wherein the step of dividing the solder image into a main area and an outer area comprises:
determining the position of a soldering tin point in a soldering tin image, acquiring a first area taking the soldering tin point as a center, and acquiring a second area which covers the soldering tin point and is larger than the first area;
the first area is used as the main area of the solder image, and the area obtained by subtracting the first area from the second area is used as the outer area of the solder image.
6. The method for detecting the direction of a solder dot as set forth in claim 1, further comprising, after the step of determining the direction of the solder dot in the direction corresponding to the sub-image to be tested, the steps of:
and if the direction of the soldering tin points is not matched with the direction of the soldering tin points preset in the soldering tin standard domain image, judging that the soldering tin points have the continuous soldering defect.
7. The method for detecting the direction of solder points as set forth in claim 1, further comprising the steps of, before the step of obtaining the probability that the sub-image to be tested is directional or non-directional by using the pre-trained solder point direction detection model:
and screening the sub-graphs to be detected in the direction needing to be detected from the sub-graphs to be detected according to the direction of the preset soldering points of the soldering tin standard domain image.
8. A solder dot orientation detection system, comprising:
the area dividing module is used for acquiring a soldering tin image of a soldering tin point and dividing the soldering tin image into a main area and an external area according to the position of the soldering tin point in the soldering tin image;
the subgraph acquisition module is used for acquiring subgraphs to be detected of the soldering tin image in at least one direction; the sub-image to be detected comprises a local image of the main area in one direction and a local image of the external area in the same direction;
the probability calculation module is used for acquiring the probability of the sub-image to be tested, which is directional or non-directional, by using a pre-trained soldering tin point direction detection model;
the direction judgment module is used for judging the direction of the soldering tin point, and if the probability of directivity of the sub-image to be tested is greater than the probability of non-directivity, the direction of the soldering tin point is judged to be in the direction corresponding to the sub-image to be tested;
the subgraph acquisition module is also used for setting a detection area in at least one direction of the soldering tin image and acquiring a subgraph to be detected of the soldering tin image in the direction corresponding to the detection area;
the subgraph acquisition module is also used for selecting one direction as the reference direction of the soldering tin image; and according to the reference direction, soldering point detection areas are respectively arranged in the directions which are separated from the reference direction by preset angles.
9. The solder point direction detection system of claim 8, wherein the sub-pattern obtaining module is further configured to set 8 solder point detection areas in directions spaced from the reference direction by 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °, respectively.
10. A computer arrangement comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor realizes the method for detecting the orientation of solder dots as claimed in any one of claims 1 to 7 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810164009.4A CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810164009.4A CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108428247A CN108428247A (en) | 2018-08-21 |
| CN108428247B true CN108428247B (en) | 2020-06-12 |
Family
ID=63157336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810164009.4A Active CN108428247B (en) | 2018-02-27 | 2018-02-27 | Method and system for detecting direction of soldering tin point |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108428247B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146873B (en) * | 2018-09-04 | 2020-12-29 | 凌云光技术股份有限公司 | Learning-based intelligent detection method and device for defects of display screen |
| TWI745767B (en) * | 2019-10-18 | 2021-11-11 | 汎思數據股份有限公司 | Optical inspection secondary image classification method |
| TWI744999B (en) * | 2020-07-23 | 2021-11-01 | 和碩聯合科技股份有限公司 | Method for training solder joint detecting model, method for detecting solder joint, and solder joint detecting device |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104764752B (en) * | 2015-03-24 | 2017-09-12 | 广州视源电子科技股份有限公司 | A kind of solder joint connects tin detection method and equipment |
| CN105588840B (en) * | 2015-12-04 | 2019-01-25 | 广州视源电子科技股份有限公司 | A kind of electronic units fix method and device |
| CN105931226A (en) * | 2016-04-14 | 2016-09-07 | 南京信息工程大学 | Automatic cell detection and segmentation method based on deep learning and using adaptive ellipse fitting |
| CN105821538B (en) * | 2016-04-20 | 2018-07-17 | 广州视源电子科技股份有限公司 | The detection method and system of spun yarn fracture |
| CN106290388B (en) * | 2016-08-03 | 2018-09-28 | 国网山东省电力公司电力科学研究院 | A kind of insulator breakdown automatic testing method |
| CN106372630A (en) * | 2016-11-23 | 2017-02-01 | 华南理工大学 | Face direction detection method based on deep learning |
| CN107451997A (en) * | 2017-07-31 | 2017-12-08 | 南昌航空大学 | A kind of automatic identifying method of the welding line ultrasonic TOFD D scanning defect types based on deep learning |
-
2018
- 2018-02-27 CN CN201810164009.4A patent/CN108428247B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN108428247A (en) | 2018-08-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110047073B (en) | X-ray weld image defect grading method and system | |
| US10878283B2 (en) | Data generation apparatus, data generation method, and data generation program | |
| US10885618B2 (en) | Inspection apparatus, data generation apparatus, data generation method, and data generation program | |
| JP6285640B2 (en) | Integrated automatic and manual defect classification | |
| CN108428247B (en) | Method and system for detecting direction of soldering tin point | |
| CN109813717A (en) | Defect detecting method | |
| TWI669519B (en) | Board defect filtering method and device thereof and computer-readabel recording medium | |
| WO2017071406A1 (en) | Method and system for detecting pin of gold needle element | |
| CN113030121A (en) | Automatic optical detection method, system and equipment for circuit board components | |
| CN112418590B (en) | Printed circuit board component detection method and system | |
| CN113077416A (en) | Welding spot welding defect detection method and system based on image processing | |
| CN117274258A (en) | Method, system, equipment and storage medium for detecting defects of main board image | |
| CN112085708A (en) | Method and equipment for detecting defects of straight line edge in product outer contour | |
| CN115578318A (en) | Probe needle mark detection method and system based on image processing | |
| CN117455917B (en) | Establishment of false alarm library of etched lead frame and false alarm on-line judging and screening method | |
| CN117058411B (en) | Method, device, medium and equipment for identifying edge appearance flaws of battery | |
| CN113554645A (en) | Industrial anomaly detection method and device based on WGAN | |
| JP2014126445A (en) | Alignment device, defect inspection device, alignment method and control program | |
| CN113039631B (en) | Defect classification by fitting optical signals to point spread functions | |
| KR101659989B1 (en) | Apparatus and method for analyzing abnormal data using combination of multi-dimensional features | |
| CN116934732A (en) | Photovoltaic module detection method and device and electronic equipment | |
| JP6708695B2 (en) | Inspection equipment | |
| CN113222950B (en) | Surface defect detection model training method, surface defect detection method and system | |
| CN112529883A (en) | Patch detection method and device based on image edge recognition | |
| TWI689869B (en) | Image inspection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |