Search

CN-121612214-B - Scanning method for three-dimensional outline of electronic component package

CN121612214BCN 121612214 BCN121612214 BCN 121612214BCN-121612214-B

Abstract

The invention relates to the technical field of optical precision measurement and discloses a scanning method of a three-dimensional contour of an electronic component package body, which comprises the steps of executing sequence acquisition of non-uniform gradients to obtain an original light intensity array, and mapping a light intensity sequence into a complex frequency domain response vector; the invention eliminates the nonlinear interference of light intensity by complex domain vector synthesis, utilizes a vector argument difference mechanism to internally inhibit environmental vibration artifact and realizes high-precision geometric true value restoration under wide dynamic range and dynamic working condition.

Inventors

  • ZHOU YAXIONG
  • Min Gaoyang

Assignees

  • 四川盛弈电气设备有限公司

Dates

Publication Date
20260512
Application Date
20260202

Claims (9)

  1. 1. The scanning method of the three-dimensional outline of the electronic component packaging body is characterized by comprising the following steps of: controlling the structured light projection unit to project to the surface of the tested electronic component package body Step phase shift stripes and control the image acquisition unit to perform for each step phase shift Integrating acquisitions of different durations to obtain an original array of light intensities, wherein And Are integers greater than or equal to 3; For each pixel point in an image coordinate system, directly mapping light intensity sequences under different integration time into a plurality of initial response vectors on a complex plane according to a discrete transformation rule, wherein the modulus of the initial response vectors represents a fringe modulation degree under the integration time, and the amplitude angle of the initial response vectors represents phase information under the integration time; for pixel points with two or more effective initial response vectors, calculating amplitude angle deviation values among the effective initial response vectors with different integration times, and comparing the amplitude angle deviation values with a preset phase tolerance standard; when the amplitude angle deviation value exceeds the phase tolerance standard, judging that the pixel point is in an unsteady state caused by physical vibration, and resetting the normalized synthesized weight of all the other initial response vectors except the effective initial response vector with the shortest integral time to zero; Based on the normalized synthesis weight, performing a complex-domain vector superposition operation on all valid initial response vectors to generate a unique synthesized response vector, and resolving the argument of the synthesized response vector to obtain a truncated phase; Unwrapping the truncated phase and converting it into three-dimensional height data using a phase height map; The setting of the phase tolerance reference follows the self-adaptive noise floor constraint rule, namely, the system phase noise base value under the current measurement environment is obtained, the theoretical phase uncertainty is calculated according to the modulation modulus of the effective initial response vector, the superposition result of the system phase noise base value and the theoretical phase uncertainty is used as the phase tolerance reference, and the pixel point where the amplitude angle deviation value exceeds the superposition result is judged to be the non-steady state point.
  2. 2. The method for scanning the three-dimensional profile of the electronic component package according to claim 1, wherein the execution process of the linear response judgment further comprises a neighborhood crosstalk suppression sub-step, wherein for any integration time level, pixel points with original light intensity values higher than a sensor linear response threshold are identified to construct a saturated pixel set, surrounding potential overflow neighborhood is determined according to a preset topological rule based on the saturated pixel set, and when normalized synthesis weights of the pixel points in the potential overflow neighborhood are calculated, a preset suppression factor is introduced to reduce the weight value of an initial response vector corresponding to the integration time level, so that the synthesis operation tends to adopt the initial response vector under a shorter integration time level.
  3. 3. The method for scanning the three-dimensional profile of the electronic component package according to claim 2, wherein the specific operation of determining the surrounding potential overflow neighborhood according to the preset topology rule comprises the steps of performing morphological expansion operation on the saturated pixel set in the two-dimensional image space, setting the expansion radius according to the characteristic size of the point spread function of the optical system, subtracting the original saturated pixel set from the area covered after expansion to obtain the potential overflow neighborhood, wherein the area represents the spatial range which is not subjected to digital saturation but is polluted by photoelectron overflow.
  4. 4. The method of claim 1, wherein the vector superposition is performed on the basis of normalized luminous flux, which follows the following algorithm: , wherein, In order to synthesize the response vector, As a total number of steps of the integration time, Is the first The normalized composite weights corresponding to the level integration time, Is the first The effective initial response vector at the stage integration time, Is the first The duration value of the stage integration time.
  5. 5. The method of claim 1, wherein, in the step of obtaining the original light intensity array, The numerical value setting of the integration time of different levels of time length follows the light intensity dynamic coverage rule, wherein the first level of integration time is set as the minimum exposure time length which enables the metal area with highest reflectivity of the surface of the measured object not to generate light intensity saturation; first, the The level integral time is set as the maximum exposure time length for the stripe modulation degree of the substrate area with the lowest reflectivity of the surface of the measured object to be higher than the signal to noise ratio threshold value of the system, and the intermediate level integral time is set as the first level integral time and the second level integral time The stage integration time is distributed according to a preset gradient.
  6. 6. The method according to claim 1, wherein the three-dimensional reconstruction step further comprises a phase compensation sub-step for a highly reflective region, wherein after obtaining a truncated phase, a pixel region synthesized only by a shortest integration time vector is identified, a modulus of a synthesized response vector of the pixel region on a complex plane is retrieved, and when the modulus is lower than a preset reliability threshold, a phase of a steady-state pixel point synthesized by a multi-stage integration time vector in a spatial neighborhood of the pixel region is utilized to smooth repair the phase of the region by a preset interpolation algorithm.
  7. 7. The method of claim 1, wherein the method is performed entirely in a complex data stream, and all of the signal processing, weight distribution and noise suppression operations are directly applied to the real and imaginary parts of the initial response vector before the truncated phase is generated, and no intermediate gray scale or intensity images are generated and processed.
  8. 8. The method for scanning the three-dimensional contour of the electronic component package according to claim 1, further comprising a pre-calibration step for nonlinear response of the system, wherein a linearity mapping table of the sensor photoelectric response curve is established before the control structure light projection unit projects the fringes, and in the step of constructing the complex frequency domain response vector, the collected original light intensity array is subjected to inverse nonlinear compensation according to the linearity mapping table.
  9. 9. The method for scanning the three-dimensional contour of the electronic component package according to claim 1, wherein the step of constructing the complex frequency domain response vector specifically comprises extracting a light intensity sequence at the same integration time from an original light intensity array for each pixel point, projecting the light intensity sequence to a real axis and a virtual axis of a complex plane, respectively calculating a vector real part and a vector imaginary part, wherein the vector real part is a weighted sum of the light intensity sequence and a cosine function, the vector imaginary part is a weighted sum of the light intensity sequence and a sine function, and constructing an initial response vector by the vector real part and the vector imaginary part, wherein an amplitude angle of the initial response vector directly corresponds to a wrapped phase value of the pixel point at the current integration time.

Description

Scanning method for three-dimensional outline of electronic component package Technical Field The invention relates to a scanning method of a three-dimensional contour of an electronic component package, and belongs to the technical field of optical precision measurement. Background The current semiconductor encapsulation process utilizes stripe projection contour technique to detect three-dimensional morphology of ball grid array and wafer level encapsulation component and evaluate pin coplanarity and encapsulation body warpage, the technology projects sinusoidal grating to the surface of the tested object through projection unit, image sensor collects the highly modulated deformed stripe image of the tested object, and according to phase change solution three-dimensional point cloud data, the non-contact measurement means uses full-field data acquisition capability and submicron longitudinal resolution to establish dominant position on standardized production line, in the development process of automatic detection equipment, simple pursuing transmission efficiency and basic contour screening scheme can raise productivity, it is difficult to meet the extremely demanding requirement of advanced process on microscopic morphology, for example, chinese patent application patent publication number CN108692662B tries to utilize multitask rotating disk to cooperate with array type beam transceiver component, capture the signal position back-thrust electronic component contour size after the component is blocked, however, such method is based on physical blocking hard detection logic to stay two-dimensional projection or low precision physical boundary discrimination, the basic good elbow screening is advanced, the method is based on complex three-dimensional characteristic encapsulation body capture, especially the surface is difficult to meet the extremely severe requirements of microscopic morphology, and the ultra-contact state of the ultra-thin metal reflection sphere cannot be restored to the ultra-high-precision spherical reflection surface, and the ultra-low-precision is unable to measure the ultra-precision surface is high. The method comprises the steps of carrying out extremely binary differentiation on the surface optical characteristics of a measured object along with the improvement of packaging integration level, carrying out phase demodulation on the surface optical characteristics of the measured object along with the extremely binary differentiation, carrying out photoelectric response weighting or gray correction operation on the surface optical characteristics of the measured object along with the existence of a single packaging body, damaging strict linear characteristics of sinusoidal stripes due to the fact that the reflectivity span exceeds the dynamic range of a conventional linear image sensor, carrying out phase interception due to the fact that the light intensity of a tin ball area overflows under the condition of single integration time, carrying out physical phase step or false deformation on the surface optical characteristics of the substrate area due to weak signal interference caused by shot noise, solving the problem that the dynamic range is not enough, adopting a multiple exposure technology, respectively collecting a plurality of stripe images with different integration times, utilizing an image fusion algorithm to synthesize a light intensity distribution uniform stripe graph, and carrying out phase demodulation on the basis of synthesized images, carrying out principle deviation on the basis of a path of the synthesized image field optical intensity image synthesis technology, carrying out precise metering application, carrying out photoelectric response weighting or gray correction operation on the basis of the principle deviation, carrying out subsequent arc tangent phase calculation on the nonlinear distortion of the light intensity domain, and generating non-physical phase step or false deformation in the light intensity boundary severe change in the area due to the bright-dark boundary. Therefore, how to construct a scanning mechanism which avoids nonlinear interference of light intensity and suppresses physical environment from influencing a wide dynamic range becomes a technical problem to be solved by the invention. Disclosure of Invention In order to solve the problems in the background technology, the technical scheme of the invention is as follows, a scanning method of the three-dimensional outline of the electronic component package body comprises the following steps: controlling the structured light projection unit to project to the surface of the tested electronic component package body Step phase shift stripes and control the image acquisition unit to perform for each step phase shiftIntegrating acquisitions of different durations to obtain an original array of light intensities, whereinAndAre integers greater than or equal to 3; For each pixel point in an im