Search

CN-121998952-A - Multi-dimensional visual detection method and system for mobile phone card holder

CN121998952ACN 121998952 ACN121998952 ACN 121998952ACN-121998952-A

Abstract

The application provides a multi-dimensional visual detection method and system for a mobile phone card holder, which relate to the field of visual detection and comprise the steps of acquiring first-class image information and second-class image information of the mobile phone card holder when the mobile phone card holder moves to a detection area, processing the first-class image information and the second-class image information to establish a pixel-level spatial corresponding relation between the first-class image information and the second-class image information, and comparing local information of the first-class image information and the second-class image information based on the pixel-level spatial corresponding relation, wherein the local information comparison comprises the steps of acquiring texture difference, brightness change and edge structure information of a local area, judging inherent design characteristics and actual surface defects of the mobile phone card holder according to the texture difference, the brightness change and the edge structure information, so as to obtain a local information comparison result, avoid time consumption caused by light source switching and repeated photographing in a traditional multiple exposure scheme, remarkably improve detection efficiency and meet the requirement of a production line.

Inventors

  • YUAN JIANXUN

Assignees

  • 东莞市鼎平精密五金科技有限公司

Dates

Publication Date
20260508
Application Date
20260128

Claims (10)

  1. 1. The multi-dimensional visual detection method for the mobile phone card holder is characterized by comprising the following steps of: When a mobile phone card holder moves to a detection area, acquiring first-type image information and second-type image information of the mobile phone card holder, wherein the first-type image information is acquired by an illumination mode configured to highlight inherent design features of the mobile phone card holder, and the second-type image information is acquired by an illumination mode configured to highlight potential surface defects of the mobile phone card holder; processing the first type image information and the second type image information to establish a pixel-level spatial correspondence between the first type image information and the second type image information; and based on the pixel-level spatial correspondence, carrying out local information comparison on the first-type image information and the second-type image information, wherein the local information comparison comprises the steps of obtaining texture difference, brightness change and edge structure information of a local area, and judging inherent design characteristics and actual surface defects of the mobile phone card holder according to the texture difference, the brightness change and the edge structure information to obtain a local information comparison result.
  2. 2. The method for multi-dimensional visual inspection of mobile phone card holder according to claim 1, wherein the step of comparing the first type of image information and the second type of image information based on the pixel-level spatial correspondence includes obtaining texture differences, brightness changes and edge structure information of local areas, and judging inherent design features and actual surface defects of the mobile phone card holder according to the texture differences, brightness changes and edge structure information, and obtaining local information comparison results includes: acquiring an average brightness value of a local area in the first type of image information; acquiring an average brightness value of the local area in the second type of image information; calculating the ratio of the average brightness value of the local area in the second type of image information to the average brightness value of the local area in the first type of image information to obtain a brightness ratio; detecting edge intensity of the local area in the second type of image information; When the average brightness value of the local area in the second type of image information is higher than a first preset brightness threshold value, the brightness ratio is lower than a first preset ratio threshold value, and meanwhile, the edge intensity of the local area in the second type of image information is higher than a first preset edge threshold value, determining that the local area has actual surface defects; And when the local area presents smooth color gradual change or continuous texture patterns in the first type of image information, and the average brightness value of the local area in the second type of image information is lower than a second preset brightness threshold value or the brightness ratio is higher than a second preset ratio threshold value, determining that the local area is a normal optical representation area.
  3. 3. The method for multi-dimensional visual inspection of mobile phone card holder according to claim 1, wherein the step of comparing the first type of image information and the second type of image information based on the pixel-level spatial correspondence includes obtaining texture differences, brightness changes and edge structure information of local areas, and judging inherent design features and actual surface defects of the mobile phone card holder according to the texture differences, brightness changes and edge structure information, and obtaining local information comparison results includes: Identifying transient foreign object features within a local region by analyzing whether the local region has high-frequency, discrete, irregularly shaped luminance or texture anomalies in the second type of image information, and in such a way that the luminance or texture anomalies exhibit relatively smooth or low-contrast features in the corresponding region in the first type of image information; According to the transient foreign body characteristics, continuously and synchronously acquiring images of the local area for a plurality of times at short time intervals in the mobile phone card movement process, comparing the position and the form change of the transient foreign body characteristics in continuous images, and evaluating the persistence of the transient foreign body characteristics through the position and the form change to obtain a persistence evaluation result; And correcting the judgment of local information comparison according to the persistence evaluation result of the transient foreign matter feature, wherein the correction comprises the steps of ignoring texture difference, brightness change and edge structure information caused by the transient foreign matter abnormality when the transient foreign matter feature is judged to be abnormal in the transient environment, determining the transient foreign matter feature as a normal actual surface, and determining the transient foreign matter feature as a potential permanent surface defect when the transient foreign matter feature shows the stability of the position and the form.
  4. 4. The method for multi-dimensional visual inspection of a mobile phone card holder according to claim 1, wherein the step of processing the first type of image information and the second type of image information to establish a pixel-level spatial correspondence between the first type of image information and the second type of image information comprises: Constructing an image pyramid of the first type of image information and the second type of image information; calculating translation deviation between the first type image information and the second type image information by using a phase correlation method at the coarsest layer of the image pyramid; According to the translation deviation, refining the translation deviation layer by layer, and carrying out sub-pixel interpolation on an original image layer to obtain sub-pixel level translation quantity; performing geometric transformation on the second type of image information according to the sub-pixel level translation amount to enable the second type of image information and the first type of image information to coincide on a pixel level; And after determining that the second type of image information is aligned with the first type of image information, performing alignment quality evaluation on the aligned images to obtain an alignment quality evaluation result.
  5. 5. The method for multi-dimensional visual inspection of mobile phone card holder according to claim 4, wherein the step of performing alignment quality evaluation on the aligned image after determining that the second type of image information is aligned with the first type of image information, and obtaining an alignment quality evaluation result comprises: Continuously recording the alignment quality fraction; Calculating the change trend of the alignment quality fraction; when the descending trend of the alignment quality fraction is identified, generating an alignment quality descending early warning; screening out representative image pairs for high-precision alignment to obtain high-precision alignment parameters; comparing the deviation of the high-precision alignment parameter and the historical stability parameter; and generating an alignment deviation accumulation early warning according to the accumulation increasing trend of the alignment parameter deviation.
  6. 6. The method for multi-dimensional visual inspection of a mobile phone card holder according to claim 1, wherein the step of processing the first type of image information and the second type of image information to establish a pixel-level spatial correspondence between the first type of image information and the second type of image information comprises: Preprocessing the first type image information and the second type image information to enhance image characteristics and inhibit noise, so as to obtain preprocessed first type image information and preprocessed second type image information; Extracting local feature points with scale invariance and rotation invariance from the preprocessed first-class image information and the preprocessed second-class image information; based on the local feature points, applying a feature point matching strategy to generate initial matching point pairs; screening the initial matching point pairs through an iterative optimization method, and removing abnormal matching points to obtain internal point matching; Calculating a geometric transformation matrix between the first type of image information and the second type of image information according to the interior point matching, wherein the geometric transformation matrix can describe translation, rotation, scaling and micro perspective distortion; and carrying out geometric correction on the second type of image information according to the geometric transformation matrix to enable the second type of image information to coincide with the first type of image information on a pixel level, and establishing a pixel level space corresponding relation between the first type of image information and the second type of image information.
  7. 7. The multi-dimensional visual inspection method of mobile phone card holder according to claim 6, wherein the step of screening the initial matching point pairs by an iterative optimization method to remove abnormal matching points and obtain inner point matching comprises the steps of: based on the descriptor similarity of the local feature points, carrying out preliminary screening on the initial matching point pairs, and removing the point pairs with the similarity lower than a preset threshold value to obtain the preliminarily screened matching point pairs; Projecting the preliminarily screened matching point pairs into an image space, and setting geometrical consistency constraint of a certain local area according to the geometrical shape of the mobile phone card holder, wherein the geometrical consistency constraint of the local area considers local geometrical characteristics of the repetitive texture or gradual change area of the surface of the mobile phone card holder; In each iteration, randomly selecting a certain subset of matching point pairs, and fitting based on the subset of matching point pairs to generate a geometric transformation model; Calculating the re-projection errors of all the matching point pairs under the geometric transformation model, and carrying out weighting treatment on the re-projection errors according to the geometric consistency constraint of the local area to obtain first weighted re-projection errors; and identifying and removing abnormal matching points with errors larger than a preset threshold according to the first weighted reprojection errors, and repeating iteration until convergence to obtain a certain group of internal point matching.
  8. 8. The method for multi-dimensional visual inspection of a mobile phone card holder according to claim 1, wherein the step of processing the first type of image information and the second type of image information to establish a pixel-level spatial correspondence between the first type of image information and the second type of image information comprises: Preprocessing the first type image information and the second type image information to enhance image characteristics and inhibit noise, so as to obtain preprocessed first type image information and preprocessed second type image information; extracting local feature points from the preprocessed first-type image information and the preprocessed second-type image information; generating an initial matching point pair based on the local feature points; screening the initial matching point pairs to obtain interior point matching; According to the interior point matching, calculating a non-rigid deformation field between the first type of image information and the second type of image information, wherein the non-rigid deformation field can describe local bending or twisting of the mobile phone card holder; And carrying out local deformation correction on the second type of image information according to the non-rigid deformation field to enable the second type of image information to be overlapped with the first type of image information on a pixel level, and establishing a pixel level space corresponding relation between the first type of image information and the second type of image information.
  9. 9. The method for multi-dimensional visual inspection of mobile phone card holder according to claim 8, wherein the step of screening the initial matching point pair to obtain the interior point matching comprises the steps of: Based on the descriptor similarity of the local feature points, carrying out preliminary screening on the initial matching point pairs, and removing point pairs with similarity lower than a preset threshold value to obtain preliminarily screened matching point pairs; Projecting the preliminarily screened matching point pairs into an image space, and setting non-rigid geometric consistency constraint of a certain local area according to the geometric shape and a preset deformation range of the mobile phone card holder, wherein the non-rigid geometric consistency constraint considers the local geometric characteristics of the surface repetitive texture or gradual change region of the mobile phone card holder and the geometric characteristics of a specular reflection, shadow or local shielding region; in each iteration, randomly selecting a certain subset of matching point pairs, and fitting based on the subset of matching point pairs to generate a non-rigid geometric transformation model, wherein the non-rigid geometric transformation model can describe local bending or twisting of the mobile phone card holder; Calculating the re-projection errors of all the matching points under the non-rigid geometric transformation model, and carrying out weighting treatment on the re-projection errors according to the non-rigid geometric consistency constraint of the local area to obtain second weighted re-projection errors; and identifying and removing abnormal matching points with errors larger than a preset threshold according to the second weighted reprojection errors, and repeating iteration until convergence to obtain inner point matching.
  10. 10. A multi-dimensional visual detection system of a mobile phone card holder is characterized in that the system comprises: The mobile phone card support comprises an image information acquisition module, a detection module and a display module, wherein the image information acquisition module is used for acquiring first-type image information and second-type image information of the mobile phone card support when the mobile phone card support moves to a detection area, the first-type image information is acquired by configuring an illumination mode for highlighting inherent design features of the mobile phone card support, and the second-type image information is acquired by configuring an illumination mode for highlighting potential surface defects of the mobile phone card support; The spatial correspondence establishing module is used for processing the first type of image information and the second type of image information to establish pixel-level spatial correspondence between the first type of image information and the second type of image information; The local information comparison module is used for comparing the local information of the first type of image information and the second type of image information based on the pixel-level spatial correspondence, the local information comparison comprises the steps of obtaining texture difference, brightness change and edge structure information of a local area, and judging inherent design characteristics and actual surface defects of the mobile phone card holder according to the texture difference, the brightness change and the edge structure information to obtain a local information comparison result.

Description

Multi-dimensional visual detection method and system for mobile phone card holder Technical Field The application relates to the technical field of visual detection, in particular to a multi-dimensional visual detection method and system for mobile phone card holders. Background On the intelligent manufacturing production line, quality detection of the mobile phone card holder is a key link for ensuring product quality and production efficiency. The traditional detection mode is difficult to cope with the requirements of high-precision and real-time quality control in mass production. In particular, in an automated production plant for mobile phone parts, a set of fully automatic vision inspection systems is typically deployed that acquire image information using a plurality of fixed-position industrial cameras. The image processing program in the computer analyzes the images to check whether surface flaws exist or not, and simultaneously measures the overall dimensional accuracy of the card holder. This complex surface property presents a significant challenge to the original visual inspection system. When a card holder with a mirror effect enters a detection station, the annular LED light source above the camera causes a large area of overexposed white patches to appear in the image, so that any fine scratches or depressions cannot be detected. In order to overcome the limitation of a single fixed illumination mode, an attempt is made to upgrade and reform a detection station through multiple exposure, and although the multiple exposure scheme is excellent in technical verification stage, the detection time is seriously out of standard after the actual production line is put into production. The image acquisition time of a single card holder is increased by a plurality of times, so that the total detection time of the single card holder is far longer than the takt time set by a production line, the production bottleneck is formed, and the overall production efficiency is seriously influenced. Disclosure of Invention The application discloses a multi-dimensional visual detection method and system for a mobile phone card holder, and aims to solve the problems of contradiction between detection precision and production beats, detection blind areas caused by a single illumination mode and low efficiency caused by a multiple exposure scheme when the existing visual detection system for the mobile phone card holder faces complex surface characteristics. The technical scheme of the application is as follows: In a first aspect, the application discloses a multi-dimensional visual detection method for a mobile phone card holder, which specifically comprises the following steps: When the mobile phone card holder moves to a detection area, acquiring first-class image information and second-class image information of the mobile phone card holder, wherein the first-class image information is acquired by an illumination mode configured to highlight inherent design features of the mobile phone card holder, and the second-class image information is acquired by an illumination mode configured to highlight potential surface defects of the mobile phone card holder; Processing the first type image information and the second type image information to establish a pixel-level spatial correspondence between the first type image information and the second type image information; Based on the pixel-level space corresponding relation, carrying out local information comparison on the first-type image information and the second-type image information, wherein the local information comparison comprises the steps of obtaining texture difference, brightness change and edge structure information of a local area, and judging inherent design characteristics and actual surface defects of the mobile phone card holder according to the texture difference, the brightness change and the edge structure information to obtain a local information comparison result. In a second aspect, the application also discloses a multi-dimensional visual detection system of the mobile phone card holder, which comprises: The mobile phone card support comprises an image information acquisition module, a detection module and a display module, wherein the image information acquisition module is used for acquiring first-class image information and second-class image information of the mobile phone card support when the mobile phone card support moves to a detection area, the first-class image information is acquired by configuring an illumination mode for highlighting inherent design characteristics of the mobile phone card support, and the second-class image information is acquired by configuring an illumination mode for highlighting potential surface defects of the mobile phone card support; The spatial correspondence establishing module is used for processing the first type of image information and the second type of image information to establish pixel-level spat