CN-119649184-B - Contact point detection method and device of contact pin electricity-taking connector based on visual nerve

Abstract

The application provides a contact point detection method and device of a contact pin electricity-taking connector based on visual nerves, which belong to the technical field of image processing and are used for guaranteeing the detection robustness. The method is applied to electronic equipment, the electronic equipment acquires at least two images, wherein the at least two images are obtained by shooting contact points of the contact pin power-taking connector by the shooting equipment in different shooting directions, each image of the at least two images comprises the contact points, the electronic equipment carries out correlation processing on the at least two images in a space dimension through a visual neural network model to obtain a processing result output by the visual neural network model, and the processing result indicates whether the contact points have poor contact.

Inventors

• LI KELU

Assignees

• 惠州市如意通电子科技有限公司

Dates

Publication Date

20260505

Application Date

20241125

Claims (6)

1. 1. A contact point detection method of a contact pin electricity taking connector based on visual nerves, which is characterized by being applied to electronic equipment, and comprising the following steps: The electronic equipment acquires at least two images, wherein the at least two images are obtained by shooting contact points of the contact pin power-taking connector in different shooting directions by shooting equipment, and each image in the at least two images contains the contact points; The electronic equipment carries out correlation processing on the at least two images in the space dimension through a visual neural network model to obtain a processing result output by the visual neural network model, wherein the processing result indicates whether the contact point has poor contact or not; Wherein the at least two images include a first image and a second image, the processing of the first image in the optical neural network model passing through a first fully connected layer of the optical neural network model, the processing of the second image in the optical neural network model passing through a second fully connected layer of the optical neural network model, wherein the first fully connected layer and the second fully connected layer share a portion of neurons for implementing the correlation processing in the spatial dimension; the first full-connection layer and the second full-connection layer share partial neurons, namely, the neurons which are not shared by the first full-connection layer in the cellular connection network structure are directly connected with the neurons which are not shared by the second full-connection layer in the first full-connection layer and are also directly connected with the neurons which are not shared by the first full-connection layer in the second full-connection layer, in the cellular connection network structure, the neurons which are not shared by the second full-connection layer in the first full-connection layer are not directly connected with the neurons which are not shared by the first full-connection layer in the second full-connection layer in the cellular connection network structure, the optical neural network model is configured to activate the neurons which are not shared by the second full-connection layer in the first full-connection layer when the optical neural network model processes the first image, and to activate the neurons which are not shared by the second full-connection layer when the first full-connection layer is not shared by the second full-connection layer in the optical neural network model, and neurons in the second fully connected layer that are not shared by the first fully connected layer are activated, and the portion of neurons in the first fully connected layer that are shared by the second fully connected layer are always activated when the optical neural network model processes the first image and the second image; the first full-connection layer is configured to train to convergence a training image set obtained by photographing the contact point in a first photographing direction using the photographing device, the second full-connection layer is configured to train to convergence a training image set obtained by photographing the contact point in a second photographing direction using the photographing device, or the at least two images include a first image, A second image and a third image, the processing of the first image in the visual neural network model passing through a first fully connected layer of the visual neural network model, the processing of the second image in the visual neural network model passing through a second fully connected layer of the visual neural network model, the processing of the third image in the visual neural network model passing through a third fully connected layer of the visual neural network model, wherein the first fully connected layer shares a portion of neurons with the second fully connected layer, and the second fully connected layer shares a portion of neurons with the third fully connected layer for effecting the associated processing in the spatial dimension, and the first fully connected layer comprises neurons, The neurons included in the second full-connection layer and the neurons included in the second full-connection layer form a honeycomb-connection network structure, and the first full-connection layer and the second full-connection layer share partial neurons, namely, the neurons in the honeycomb-connection network structure, which are directly connected with the neurons in the first full-connection layer and which are not shared by the second full-connection layer, are also directly connected with the neurons in the second full-connection layer and which are not shared by the first full-connection layer; the second fully-connected layer and the third fully-connected layer share partial neurons are neurons in the honeycomb-connected network structure which are directly connected with neurons in the second fully-connected layer which are not shared by the third fully-connected layer and are also directly connected with neurons in the third fully-connected layer which are not shared by the second fully-connected layer, in the honeycomb-connected network structure, the neurons in the first fully-connected layer which are not shared by the second fully-connected layer are not directly connected with neurons in the second fully-connected layer which are not shared by the first fully-connected layer, the neurons in the second fully-connected layer which are not shared by the third fully-connected layer are not directly connected with neurons in the third fully-connected layer which are not shared by the second fully-connected layer, the visual neural network model is configured such that when the visual neural network model processes the first image, the neurons in the first fully-connected layer which are not shared by the second fully-connected layer are not activated, and the neurons in the second fully-connected layer which are not shared by the second fully-connected layer are not activated when the first fully-connected layer is not shared by the second fully-connected layer are not activated, the neurons in the third fully connected layer which are not shared by the second fully connected layer are not activated, the neurons in the second fully connected layer which are not shared by the first fully connected layer and the second fully connected layer are activated, and when the third image is processed by the visual neural network model, the neurons in the second fully connected layer which are not shared by the third fully connected layer are not activated, the neurons in the third fully connected layer which are not shared by the second fully connected layer are activated, the first fully connected layer is configured to train a training image set obtained by shooting the contact point in the first shooting direction by using the shooting device to converge, the second fully connected layer is configured to train a training image set obtained by shooting the contact point in the second shooting direction by using the shooting device to converge, and the second fully connected layer is configured to train a training image set obtained by shooting the contact point in the third shooting direction by using the shooting device to converge.
2. 2. The method of claim 1, wherein the pin power connector comprises a base and a pin, the pin comprises a first end and a second end, the first end of the pin is disposed on the base, the contact points disposed between the first end and the second end for power comprise a first contact point and a second contact point, the first contact point and the second contact point are distributed on two sides of the pin, a first reference line connecting the first contact point and the second contact point is disposed, a second reference line connecting the first end and the second end is disposed, and the first reference line and the second reference line are located on a reference plane; The at least two images comprise a first image and a second image, the first image is an image obtained by shooting the first contact point and the second contact point by the shooting device in a first shooting direction, a first preset angle is formed between the first shooting direction and a reference direction, the first preset angle is not 0, the second image is an image obtained by shooting the first contact point and the second contact point by the shooting device in a second shooting direction, a second preset angle is formed between the second shooting direction and the reference direction, the second preset angle is not 0, and the reference direction is a direction perpendicular to the reference plane; The image of the first contact point in the first image is the same as the image part of the first contact point in the second image, and the image of the second contact point in the first image is the same as the image part of the second contact point in the second image.
3. 3. The method according to claim 2, wherein the electronic device correlates the at least two images in a spatial dimension through a neural network model to obtain a processing result output by the neural network model, and the processing result comprises: The electronic equipment convolves the first image through a convolution layer of the visual neural network model to obtain a first convolution vector set, then carries out pooling treatment on the first convolution vector set through a pooling layer of the visual neural network model to obtain a first pooling vector set, and finally inputs the first pooling vector set into a first full-connection layer of the visual neural network model through a first channel to obtain a first intermediate analysis result output by the first full-connection layer; The electronic equipment convolves the second image through the convolution layer of the visual neural network model to obtain a second convolution vector set, the pooling layer of the visual neural network model is used for pooling the second convolution vector set to obtain a second pooling vector set, and finally the second pooling vector set is input into a second full-connection layer of the visual neural network model through a second channel to obtain a second intermediate analysis result output by the second full-connection layer; Wherein the first fully-connected layer shares a portion of neurons with the second fully-connected layer for implementing the association process in the spatial dimension; And the electronic equipment processes the first intermediate analysis result and the second intermediate analysis result through an output layer of the optical neural network model to obtain the processing result output by the output layer.
4. 4. The method of claim 1, wherein the pin power connector comprises a base and a pin, the pin comprises a first end and a second end, the first end of the pin is disposed on the base, the contact points disposed between the first end and the second end for power comprise a first contact point and a second contact point, the first contact point and the second contact point are distributed on two sides of the pin, a first reference line connecting the first contact point and the second contact point is disposed, a second reference line connecting the first end and the second end is disposed, and the first reference line and the second reference line are located on a reference plane; The at least two images comprise the first image, the second image and the third image, wherein the first image is an image obtained by the shooting device shooting the first contact point and the second contact point in a first shooting direction, a first preset angle is formed between the first shooting direction and a reference direction, the first preset angle is not 0, the second image is an image obtained by the shooting device shooting the first contact point and the second contact point in a second shooting direction, the second shooting direction is the same as the reference direction, the third image is an image obtained by the shooting device shooting the first contact point and the second contact point in a third shooting direction, a second preset angle is formed between the third shooting direction and the reference direction, the second preset angle is not 0, and the reference direction is a direction perpendicular to the reference plane; The image of the first contact point in the first image, the image of the first contact point in the second image and the image of the first contact point in the third image are identical, and the image of the second contact point in the first image, the image of the second contact point in the second image and the image of the second contact point in the third image are identical.
5. 5. The method of claim 4, wherein the electronic device correlates the at least two images in a spatial dimension through a neural network model to obtain a processing result output by the neural network model, comprising: The electronic equipment convolves the first image through a convolution layer of the visual neural network model to obtain a first convolution vector set, then carries out pooling treatment on the first convolution vector set through a pooling layer of the visual neural network model to obtain a first pooling vector set, and finally inputs the first pooling vector set into a first full-connection layer of the visual neural network model through a first channel to obtain a first intermediate analysis result output by the first full-connection layer; The electronic equipment convolves the second image through the convolution layer of the visual neural network model to obtain a second convolution vector set, the pooling layer of the visual neural network model is used for pooling the second convolution vector set to obtain a second pooling vector set, and finally the second pooling vector set is input into a second full-connection layer of the visual neural network model through a second channel to obtain a second intermediate analysis result output by the second full-connection layer; The electronic equipment convolves the third image through the convolution layer of the visual neural network model to obtain a third convolution vector set, the pooling layer of the visual neural network model is used for pooling the third convolution vector set to obtain a third pooling vector set, and finally the third pooling vector set is input into a third full-connection layer of the visual neural network model through a third channel to obtain a third intermediate analysis result output by the third full-connection layer; Wherein the first fully-connected layer shares a portion of neurons with the second fully-connected layer, and the second fully-connected layer shares a portion of neurons with the third fully-connected layer for implementing the association process in the spatial dimension; And the electronic equipment processes the first intermediate analysis result, the second intermediate analysis result and the third intermediate analysis result through an output layer of the optical neural network model to obtain the processing result output by the output layer.
6. 6. A contact point detection device of a contact pin electricity taking connector based on visual nerves, characterized in that it is applied to an electronic apparatus, and is configured to perform the method according to any one of claims 1-5.

Description

Contact point detection method and device of contact pin electricity-taking connector based on visual nerve Technical Field The application relates to the technical field of image processing, in particular to a contact point detection method and device of a contact pin electricity-taking connector based on visual nerves. Background With the deep development of industrial automation and intelligent manufacturing, the performance and reliability of electrical equipment become important factors affecting production efficiency and safety. The contact pin electricity-taking connector is used as a key component of electricity-taking equipment, and the quality of contact points directly influences the stability of the equipment. At present, the traditional contact point detection method mainly depends on manual experience and equipment performance test, but has the problems that the manual detection is long in time consumption, the requirement of a large-scale production line is difficult to deal with, the manual detection depends on experience of detection personnel, misjudgment is easy to occur, and the like. Therefore, consider implementation using a neural network model. For example, the acquired data is first preprocessed to extract key features, ready for training of neural network models. Then, a proper neural network architecture is selected, and the model is trained and optimized to improve the accuracy and the robustness of detection. And finally, deploying the trained model into an actual production environment, realizing automatic detection, continuously monitoring the performance of the model, and ensuring the reliability of a detection result. However, how to guarantee the robustness of the neural network model is a problem of current research. Disclosure of Invention The embodiment of the application provides a contact point detection method and device of a contact pin electricity-taking connector based on visual nerves, which are used for guaranteeing the robustness of detection. In order to achieve the above purpose, the application adopts the following technical scheme: The contact point detection method of the contact pin electricity taking connector based on the optical nerve is applied to electronic equipment, and comprises the steps that the electronic equipment obtains at least two images, wherein the at least two images are obtained by shooting contact points of the contact pin electricity taking connector by shooting equipment in different shooting directions, each image of the at least two images comprises the contact points, the electronic equipment carries out correlation processing on the at least two images in the space dimension through an optical neural network model to obtain a processing result output by the optical neural network model, and the processing result indicates whether the contact points have bad contact conditions or not. Optionally, the pin electricity taking connector comprises a base and a pin, the pin comprises a first end and a second end, the first end of the pin is arranged on the base, a contact point for electricity taking arranged between the first end and the second end comprises a first contact point and a second contact point, the first contact point and the second contact point are distributed on two sides of the pin, a first reference line for connecting the first contact point and the second contact point is arranged, a second reference line for connecting the first end and the second end is arranged, the first reference line and the second reference line are located on a reference plane, at least two images comprise a first image and a second image, the first image is an image obtained by shooting the first contact point and the second contact point by the shooting device in a first shooting direction, a first preset angle is not 0 between the first shooting direction and the second contact point, the second image is an image obtained by shooting the first contact point and the second contact point in a second shooting direction, the second preset angle is not 0 between the second shooting direction and the reference direction, the first reference line and the second reference line are located on the reference plane, the first image and the second image is the same as the second image in the first contact point in the first image and the second image in the second image. The electronic equipment carries out convolution on a first image through a convolution layer of the visual neural network model to obtain a first convolution vector set, carries out pooling processing on the first convolution vector set through a pooling layer of the visual neural network model to obtain a first pooling vector set, and finally inputs the first pooling vector set into a first full-connection layer of the visual neural network model through a first channel to obtain a first intermediate analysis result output by the first full-connection layer, carries out convolution