Search

CN-122016826-A - Flaw detection method and system for injection molding part of notebook shell

CN122016826ACN 122016826 ACN122016826 ACN 122016826ACN-122016826-A

Abstract

The invention relates to the technical field of machine vision, in particular to a flaw detection method and system for injection molding parts of notebook housings. The two cameras are fixed on initial rotation angles with symmetrical peak central angles, synchronously acquire one frame and extract linear characteristic response intensity, judge that no defect exists if the two images do not exceed a suspected defect threshold value, calculate peak excitation rotation angles through Gaussian double-point sampling closed solutions when the two images exceed the threshold value, rotate the first camera to the angle, rotate the second camera to a reference rotation angle which enables the defect response amplitude to be minimum and meets space constraint, carry out affine registration to carry out pixel-by-pixel difference after synchronous acquisition, and then screen according to the aspect ratio of a communication area to mark shallow linear defect areas. The invention solves the shallow linear defect of random directivity under the interference of the radiating holes and the problem of optimizing the observation angle under the duplex camera.

Inventors

  • LIU CHANGHAI
  • Zhu Daibi

Assignees

  • 重庆成田科技有限公司

Dates

Publication Date
20260512
Application Date
20260416

Claims (10)

  1. 1. The flaw detection method for the injection molding part of the notebook computer shell is applied to a detection device for adjusting an observation angle by respectively and independently rotating around a horizontal single shaft, and the injection molding part is horizontally placed on a detection table, and is characterized by comprising the following steps: Gradually rotating the first camera within a preset rotation angle range, collecting flawless injection molding calibration part images frame by frame, extracting linear characteristic response intensity from an arc transition area in each frame of images to obtain a response curve of the linear characteristic response intensity changing along with the rotation angle, and performing curve fitting on the response curve to obtain peak width parameters; synchronously acquiring one frame of image of the injection molding to be detected at each initial rotation angle by the first camera and the second camera, and respectively extracting linear characteristic response intensity of the arc transition region to obtain first response intensity and second response intensity; When at least one of the first response intensity or the second response intensity exceeds a preset suspected defect threshold, calculating a peak excitation rotation angle through Gaussian response curve double-point sampling closed solution according to the first response intensity, the second response intensity, a first camera initial rotation angle, a second camera initial rotation angle and peak width parameters; And after registering the first image and the second image, calculating the difference of the linear characteristic response intensities pixel by pixel to obtain a differential response diagram, and marking a continuous linear region with the differential value exceeding a preset judgment threshold value in the differential response diagram as a shallow linear defect region.
  2. 2. The method for detecting defects of injection molding parts of notebook housings according to claim 1, wherein the method for gradually rotating a first camera within a preset rotation angle range and collecting defect-free injection molding calibration part images frame by frame, extracting linear characteristic response intensity from a circular arc transition region in each frame of images to obtain a response curve of the linear characteristic response intensity changing along with the rotation angle, and performing curve fitting on the response curve to obtain peak width parameters comprises the following steps: The first camera is in the calibration scanning process of the defect-free injection molding calibration piece The rotation angle corresponding to the frame is recorded as To take the value from 1 to Is a positive integer of (a) and (b), To calibrate the total frame number of scanning The response intensity of the linear characteristic extracted from the arc transition region in the frame image is recorded as To take the following steps Maximum value as peak amplitude To the initial value of (1) Corresponding to the maximum value Peak center angle as response curve With a predetermined angle as a peak width parameter To the initial value of (1) 、 、 Minimizing an objective function by a nonlinear least squares method for undetermined parameters Obtaining peak width parameters by iterative solution Wherein For the peak amplitude to be the value of the peak, Is the peak center angle of the response curve.
  3. 3. The method for detecting flaws of injection-molded parts of notebook housings according to claim 2, wherein the method for synchronously acquiring each frame of image of the injection-molded parts to be detected at the initial rotation angles of the first camera and the second camera, respectively extracting linear characteristic response intensities of an arc transition region, and obtaining the first response intensity and the second response intensity comprises the following steps: determining Gabor filtering direction angle from tangential direction of contour line of arc transition zone in injection molding piece design file Determining Gabor filtering spatial frequency according to shallow layer concave width in injection molding design file Gabor filtering is carried out on the arc transition region in the current frame image, the convolution result of the real part of the filtering core and the image is marked as real part response, the convolution result of the imaginary part of the filtering core and the image is marked as imaginary part response, and the square root of the square sum of the real part response and the imaginary part response is taken to obtain the current frame Gabor filtering response amplitude In the arc transition zone The maximum value of the (B) is used as the linear characteristic response intensity of the current frame image, and each frame image acquired by the first camera and the second camera is respectively obtained in the arc transition area Obtain the first response intensity And a second response intensity 。
  4. 4. A method for detecting defects in injection molded parts of a notebook computer housing according to claim 3, wherein the first camera and the second camera are synchronously operated to acquire one frame of image of the injection molded part to be detected at respective initial rotation angles, the initial rotation angle of the first camera And a second camera initial rotation angle Wherein In order to respond to the peak center angle of the curve, Is a preset symmetrical offset angle, and Must be taken to be And Are within a preset rotation angle range of the first camera, and the initial rotation angles of the first camera and the second camera are related to Symmetrical.
  5. 5. The method of claim 4, wherein when at least one of the first response intensity or the second response intensity exceeds a predetermined suspected defect threshold, calculating the peak excitation rotation angle by a gaussian response curve double-point sampling closed-loop solution based on the first response intensity, the second response intensity, the first camera initial rotation angle, the second camera initial rotation angle, and the peak width parameter, comprises: To be used for Representing peak excitation rotation angle, through first response intensity And a second response intensity Obtaining the peak excitation rotation angle The closed solution of (2) is: ; Wherein the method comprises the steps of For the first camera initial rotation angle, For the initial rotation angle of the second camera, As a parameter of the peak width of the strip, When calculated When the preset rotation angle range of the first camera is exceeded, the nearest boundary value of the preset rotation angle range is taken as the nearest boundary value 。
  6. 6. The method for detecting defects in injection-molded parts of a notebook computer case according to claim 5, wherein the step of determining, as the reference rotation angle, a rotation angle having a difference from the peak excitation rotation angle not lower than a preset pitch threshold value comprises: the aperture of the radiating hole obtained from the design file of the injection molding part is recorded as The spacing between the holes is recorded as Calculating the linear response minimum offset angle of the edge of the radiating hole Traversing the candidate rotation angle within the preset rotation angle range of the first camera Screening simultaneously meets And (3) with Of all candidate values of (1), wherein For the peak excitation rotation angle, For the preset spacing threshold value, The central angle of the peak value of the response curve is selected from candidate values meeting the condition Obtaining the minimum value As reference rotation angle Wherein For the peak amplitude to be the value of the peak, Is a peak width parameter.
  7. 7. A method for detecting defects of injection molded parts of notebook computer housings according to claim 3, wherein the first camera is rotated step by step within a preset rotation angle range and the defect-free injection molded calibration part images are collected frame by frame, and the rotation angle difference between adjacent frames satisfies the following conditions 。
  8. 8. The method for detecting defects of injection molding parts of a notebook computer shell according to claim 3, wherein the method for synchronously acquiring the injection molding parts to be detected by the first camera and the second camera, and calculating the difference of linear characteristic response intensities pixel by pixel after registering the first image and the second image to obtain a differential response map comprises the following steps: Shooting the images of the injection molding to be detected by a first camera and a second camera respectively to be recorded as a first image and a second image, extracting the coordinates of corresponding characteristic points of the first image and the second image, and obtaining an affine transformation matrix by least square fitting And the first image is processed The homogeneous sitting marks of the pixels are as follows By means of And obtaining corresponding coordinates of each pixel of the first image under a second image coordinate system, carrying out inverse mapping resampling on the first image to enable the first image and the second image to correspond pixel by pixel under the second image coordinate system, respectively executing Gabor filtering on the corresponding first image and second image in an arc transition area to obtain a first amplitude value graph and a second amplitude value graph, and calculating a difference value pixel by pixel to obtain a differential response graph.
  9. 9. The method for detecting defects of injection molded parts of notebook housings according to claim 8, wherein the method for marking a continuous linear region with a difference value exceeding a preset determination threshold value in the difference response map as a shallow linear defect region comprises: for each connected region with the difference value exceeding the preset judging threshold value in the difference response diagram, taking the direction of the eigenvector corresponding to the maximum eigenvalue of the covariance matrix of the connected region point set as the main direction, and respectively calculating the projection range along the main direction as the length Projection range in the secondary direction as width Will (i) be Marking the connected region which is not lower than the preset linearity threshold value as a shallow linear defect region, and marking And judging that the communication area below the preset linearity threshold value is heat dissipation Kong Cancha interference and eliminating.
  10. 10. The utility model provides a flaw detection system of notebook shell injection molding, is applied to the detection device that first camera and second camera each independently rotate around horizontal unipolar in order to adjust the observation angle, and the injection molding is horizontal to be placed in the test bench, its characterized in that, the system includes: The calibration module is used for gradually rotating the first camera within a preset rotation angle range, collecting flawless injection molding calibration part images frame by frame, extracting linear characteristic response intensity from an arc transition area in each frame of images to obtain a response curve of the linear characteristic response intensity changing along with the rotation angle, and performing curve fitting on the response curve to obtain peak width parameters; the first acquisition module is used for synchronously acquiring each frame of image of the injection molding to be detected at the initial rotation angles of the first camera and the second camera, and respectively extracting the linear characteristic response intensity of the arc transition region to obtain a first response intensity and a second response intensity; The secondary acquisition module is used for calculating a peak excitation rotation angle through Gaussian response curve double-point sampling closed solution according to the first response intensity, the second response intensity, the first camera initial rotation angle, the second camera initial rotation angle and peak width parameters when at least one of the first response intensity or the second response intensity exceeds a preset suspected defect threshold value; And the analysis module is used for calculating the difference of the linear characteristic response intensities pixel by pixel after registering the first image and the second image to obtain a differential response diagram, and marking a continuous linear region with the differential value exceeding a preset judgment threshold value in the differential response diagram as a shallow linear defect region.

Description

Flaw detection method and system for injection molding part of notebook shell Technical Field The invention relates to the technical field of machine vision, in particular to a flaw detection method and system for injection molding parts of notebook shells. Background In the mold filling stage, molten resin is shunted by cores such as a heat dissipation hole when flowing in a cavity, and the shunted melt fronts are recombined near an arc transition area. At the junction interface, the molecular chains of the two melts cannot be fully diffused and fused due to the fact that the temperature of the melts is reduced and the partial pressure is insufficient, shallow linear depressions with the width of about tens to hundreds of micrometers and the depth of about several micrometers to tens of micrometers are formed on the surface of a manufactured part, and the defect is commonly called a weld mark in the injection molding industry. Besides the process parameter setting deviation such as injection speed, pressure maintaining, mold temperature and the like, the insufficient moisture content of the raw materials can cause local gasification when melt fronts meet to aggravate insufficient molecular chain diffusion, and moreover, the occasional factors such as viscosity difference among raw materials, fluctuation of return material proportion, degradation of residual materials of a charging barrel and the like can cause fluctuation of melt fluidity in batches or single batch, so that the weld mark still appears with certain probability under the condition of nominal fixed process parameters. Therefore, weld marks belong to sporadic appearance defects which are difficult to completely eliminate by process adjustment in injection molding production, and are required to be identified piece by piece in a finished product detection stage. The two sides of the welding mark groove respectively form a tiny inclined plane, and the existence of the inclined plane enables the welding mark area to generate directional scattering bright lines which can be captured by a camera under a specific observation angle to form optical reflection parameters for visual detection. In machine vision inspection, the detection of the shallow linear defects also faces visual disturbances. First, the slope angle of the weld mark groove is determined by the local temperature gradient and pressure distribution during injection and mold filling, and the individual differences exist along with the fluctuation of various contingency factors. The slope inclination determines the directional scattering direction and thus the optimal observation angle that maximizes the bright line response. Because the angle is different from one piece to another, when shooting is carried out at a fixed single observation angle, the directional scattering direction of the inclined plane of part of the piece to be detected deviates from the observation direction of the camera, and bright line response is insufficient to distinguish from background diffuse reflection, so that missed detection is caused. And secondly, the circular through hole radiating array near the arc transition area of the notebook computer shell also generates linear bright lines in the image, namely, when the camera is at a certain rotation angle, the periodic arrangement of the edges of the radiating holes presents linear response which is highly similar to the bright lines of the welding marks in the direction, width and local contrast in a response diagram of a linear feature extraction operator, the two form feature level superposition in a single frame image, and the fact that the linear bright lines are derived from the inclined planes of the welding marks or the edges of the radiating holes cannot be judged by means of single frame response intensity, so that erroneous judgment is caused. Disclosure of Invention (1) The technical problems to be solved are as follows: the invention aims to provide a flaw detection method and system for injection molding parts of notebook housings, which are used for solving random directivity shallow linear defects under the interference of radiating holes and solving the problem of optimizing the observation angle of double industrial cameras. (2) The technical scheme is as follows: In order to achieve the above object, in one aspect, the present invention provides a method for detecting flaws of an injection molded part of a notebook computer casing, which is applied to a detection device in which a first camera and a second camera rotate around a horizontal single axis independently to adjust an observation angle, wherein the injection molded part is horizontally placed on a detection table, and the method comprises: Gradually rotating the first camera within a preset rotation angle range, collecting flawless injection molding calibration part images frame by frame, extracting linear characteristic response intensity from an arc transition area in each fra