JPH1131227A - Measurement method by picture recognition and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a measurement method by picture recognition for quickly and stably extracting the isosbestic line or outline of an object to be measured in an arbitrary shape with irregular reflection at low costs by a software processing. SOLUTION: At first an external pressure model/internal pressure model is selected, and the concentration of an extracted isosbestic line or the like is designated. Next, a spring ring in which spring elements are linked is set for an objective picture. For example, in the case of the external pressure model, an uniform external pressure distribution equivalent to designated concentration is converted into a node force, and added from the outside of the ring to a spring element node. Also, internal pressure distribution equivalent to concentration distribution on the spring ring is converted into a node force, and added from the inside of the ring to the node. The spring element node is moved so that the inside and outside node forces can be balanced, and the isosbestic line of the designated concentration is obtained. The concentration distribution on the spring ring is integrated at the time of conversion, and a noise with concentration non-uniformity is averaged and smoothed. Also, the ring-shaped spring model is one-dimensional spring coupled system so that a processing can be attained at an extremely high speed. Thus, the measuring processing can be operated by the software processing of a computer.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention smoothes isodensity lines and shape contours of a measurement object having an irregular shape with irregular reflection using a spring model, and achieves high speed and noise reduction at low cost.
The present invention relates to a technology for detecting and measuring a strong resistance to disturbance.

[0002]

2. Description of the Related Art In a conventional technique of this kind, an object to be measured such as a cream solder on a printed circuit board, a solder after reflow, a mounted component, or the like is irradiated with illumination light to take an image, and filtering is performed to remove noise. Thereafter, the plane information (plane shape and area) of the target image is obtained, and the detection, measurement, and inspection of the measurement target are performed.

As a technique for measuring the height by a laser scanning method, there is a laser beam cutting method. By observing the light cutting line of the measurement object obtained by scanning with the laser beam by a triangulation method, a vertical cross section of the object can be measured, and height information can be obtained. Based on this height information, it is possible to calculate the shape data (planar shape, area, height, volume) of the object, and to detect, measure, and inspect the object.

A filter often used for the above filtering includes a median filter and a simple weighted average filter. In addition, a morphological CO filter (clos-opening-fi filter) by envelope processing is used.
liter), OC filter (open-closing)
filter) is used.

[0005]

In the above-mentioned conventional measurement technology, problems when the measurement object is, for example, cream solder printing used for mounting components on a printed circuit board will be described below.

[0006] The cream solder is obtained by kneading a small solder ball of about 10 μm, and is porous, so that the surface property is extremely uneven. For this reason, in the case of image processing, the irradiation light is irregularly reflected, and large unevenness in density occurs during imaging. Due to the intense noise, it is difficult to obtain a smooth concentration distribution of the cream solder, and it is difficult to extract a uniform flat shape, a smooth shape contour and an iso-density line of the cream solder without unevenness. Detection, measurement, and inspection were difficult. Similarly, in the laser beam cutting method, similarly, due to irregular reflection, the base height information includes significant noise, and it becomes difficult to calculate the shape data (planar shape, area, height, volume) of the object,
It was difficult to detect, measure, and inspect cream solder.

[0007] In the filtering using a median filter or a simple weighted average filter, if the noise is intense, noise variation appears even in a local region window frame where filtering processing is performed, and smoothing is difficult. Also, morphological CO filters (clos-opening)
-Filter), OC filter (open-clos)
ing filter) is an envelope processing, which does not average and intensify vibrating pulse-like noise. Therefore, since the smoothing is performed by taking in the upper limit value and the lower limit value of the noise, a large contour is formed. There is a problem of shifting. Edge and edge detection methods such as primary and secondary differentiation are used to extract contours and iso-density lines. However, if noise is severe, noise is amplified by the action of differentiation. Will be difficult.

Further, in image recognition, gray processing and binarization processing are used. However, there is a problem that the processing takes a long time because the pixel data handled by these processings is enormous. For this reason, an image processing board for gray processing and binarization processing is generally used. However, the degree of freedom in design is reduced, and there is a technical restriction in achieving compatibility with a measurement object, and cost increases. It becomes a factor.

The present invention is intended to solve the above-mentioned problems. (1) Even if an object to be measured is an image of an arbitrary shape having diffuse reflection, an arbitrary isodensity line which is stable by smoothing diffuse reflection noise Obtain an arbitrary shape contour.

(2) The pixel data to be handled is reduced, and the smoothing / extraction processing of the iso-density line / shape contour is performed at high speed.

(3) It is possible to develop an algorithm having good adaptability, simplify the system configuration, and reduce the cost by using a software processing method in the CPU without using a dedicated image processing board.

An object of the present invention is to provide a measuring method by image processing which can realize the above.

[0013]

Means for Solving the Problems The present invention solves the above problems by the following means <1> to <4>.

<1> A procedure of connecting a spring element having nodes at both ends in a ring shape in association with an isodensity line or a shape contour of an object on an image, and forming a ring outside the ring. And applying a pressure distribution corresponding to the designated concentration to one side of the ring and applying a pressure distribution corresponding to the concentration distribution on the ring to the other side outside and inside the ring; Moving the node of the spring element so that the pressure distribution applied to the spring element is balanced through the node of the spring element; and changing the shape of the ring at the time of equilibrium in all of the spring elements to an equal density corresponding to the designated density. A measurement method based on image recognition, comprising: a step of forming a line or a shape contour.

<2> In the procedure for moving the node, the pressure distribution applied to the outside of the ring and the pressure distribution applied to the inside of the ring on the spring element are converted into equivalent node forces, and the converted node force is A measurement method using image recognition, characterized in that nodes of a spring element are moved so as to be balanced.

<3> In the step of moving the node, a simultaneous equation of a spring is solved so that the converted node force and the node stress of the spring element are balanced at the node, and a node displacement is obtained to obtain the spring element. A measuring method by image recognition, characterized by moving a nodal point.

<4> A recording medium characterized in that a program for causing a computer to execute the procedure in the measurement method by image recognition from <1> to <3> is recorded on a medium readable by the computer.

In the present invention, the isodensity line of the object on the image
A ring-shaped spring model expressed in correspondence with the shape contour is set, and the pressure distribution corresponding to the specified concentration and the pressure distribution corresponding to the concentration distribution on the ring are converted into nodal forces from inside and outside at the nodes of the spring element of this spring model. In addition, by moving the spring element nodes so as to balance them, an iso-density line or contour shape of the designated density is obtained. Here, the density distribution on the ring is integrated on the spring element at the time of conversion to the nodal force, and the noise of density unevenness is averaged and smoothed. Further, since the ring-shaped spring model is a one-dimensional spring-coupled model, the processing can be performed at extremely high speed. For this reason,
The processing procedure of the method of the present invention can be performed by software processing of a computer, so that it is possible to develop an algorithm that is suitable for a measurement target, and to simplify the system configuration and reduce costs.

[0019]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing an example of a configuration for executing an embodiment of a measuring method according to the present invention. In the present embodiment, an example is described in which a contour line and a shape contour of a land on a printed circuit board to which cream solder having irregular reflection and large unevenness of density is applied are extracted.

In FIG. 1, reference numeral 1 denotes an illuminating device for illuminating an object to be measured, 2 an image capturing device such as a camera for capturing an image of an object to be measured, and 3 a storage device (image memory, memory) for capturing and storing or storing a captured image. , A hard disk, etc.) and 4 is a processing device (C) that performs measurement on stored or stored images by image processing by software processing.
PU) and 5 are display devices (displays) for displaying images, messages, and the like in the main measurement process, processing results, and the like.

An operation example according to the method of the present embodiment with the above configuration will be described. FIG. 2 is a processing flowchart showing the processing procedure of the present method. FIGS. 3A and 3B are explanatory diagrams of the method. FIG. 3A is a diagram illustrating a cross-sectional example of a density distribution of an image of a measurement object, and FIG. 3B is a ring-shaped spring model (an example of an external pressure model). FIG.

In this method, the iso-density lines and shape contours of an arbitrary-shaped cream solder having a large density unevenness as shown in the cross-sectional example of the density distribution in FIG. A closed ring-shaped spring element model as shown in FIG. 3B is used. This ring-shaped spring model
A ring is formed by connecting several spring elements having nodes at both ends with nodes. This ring is set along the contour of the land on the image.

In this method, the pressure distribution corresponding to the concentration is applied to this ring to extract the iso-density line and the contour shape. Depending on the direction in which the pressure distribution acts, the external pressure model and the internal pressure model are extracted. Divided into The external pressure model applies a uniform external pressure distribution (converted to a nodal force P) corresponding to the specified density of the isodensity line from the outside of the ring as shown in FIG. An internal pressure distribution (converted to nodal forces P _i , i = 1 to n) corresponding to the distribution is applied. The internal pressure model is dual with the external pressure model described above, in which a uniform internal pressure distribution corresponding to the specified concentration of the isoconcentration line acts from the inside of the ring, and an external pressure distribution corresponding to the concentration distribution on the ring is applied from the outside of the ring. Let it work.

First, the cream solder of the object on the printed circuit board is irradiated by the illumination device 1, the image of the object is picked up by the image pickup device 2, and stored or stored in the storage device (image memory, memory, hard disk, etc.) 3. .

Next, the CPU 4 performs processing for smoothing and extracting isodensity lines (or shape contours) of the cream solder having large shading in accordance with the processing flow of FIG. At this time, the above-mentioned external pressure model / internal pressure model is selected, and the density value of the iso-density line to be taken out is specified with reference to the grayscale image and the histogram. Here, it is assumed that the external pressure model is selected.

Next, in order to extract the iso-density line / shape contour of the solder having a large density unevenness while smoothing the density unevenness,
Closed ring-shaped spring element model (spring ring)
It is installed along the contour of the land on the image stored or stored in the storage device 3.

Next, in order to set the individual spring elements constituting the spring ring, the number of nodes and the positions of the nodes are specified. Note that the node position designation here can be set to a default equal interval.

Next, in the external pressure model, from inside the ring,
In order to apply an internal pressure distribution corresponding to the concentration distribution on the ring, the concentration distribution on the set spring element is extracted and converted into a nodal force. This nodal force (internal pressure) is obtained by integrating the internal pressure distribution on the spring element of the ring, and acts on both nodes of the spring element. As described above, since the density distribution is integrated on the spring element of the ring on the internal pressure side, the noise of shading is averaged and smoothed. Also, usually, the two nodal forces of the spring are not equal due to the complex density distribution.

An example of a method of converting the density distribution into a nodal force will be described with reference to FIG. The node forces on both sides of the spring element are P ₁ and P ₂ , the direction of the spring element is x direction, the density distribution on the spring element is p (x), and the distance from the center of gravity of the spring element to the node is d ₁ , d ₂ . Assuming that the length of the spring element is d and the moment due to both nodal forces P ₁ and P ₂ from the center of gravity of the spring element is M,

[0031]

(Equation 1)

## EQU1 ## Since the above holds, the nodal forces P ₁ , P ₂
Is found. As is clear from the above equation, the density unevenness is averaged and smoothed in order to integrate the density distribution on the spring element.

Next, in the external model, from the outside of the ring,
To apply a uniform external pressure distribution corresponding to the specified concentration of the isoconcentration line, the specified concentration is converted to a nodal force. This node force (external pressure) acts on both nodes of the spring as equal node force.

Next, returning to the processing flow of FIG. 2, a difference between the nodal force on the external pressure side and the nodal force on the internal pressure side is obtained for each node, and it is determined whether or not both node forces are balanced and balanced. If unbalanced, the nodes are moved so that the node forces are balanced and balanced. The nodal movement vector can be obtained by synthesizing nodal displacements obtained by solving the following simultaneous equations of springs so that the nodal force and the nodal stress of the spring are balanced at the nodal points. If all nodes are balanced, the process ends. The spring ring shape (isobar) at the end point is an isodensity line or contour shape of the specified density to be obtained. Here, the spring prevents the ring from crossing or the closed region in the ring from collapsing, and plays a role of forming a stable ring.

An example of the method of calculating the balance will be described with reference to FIG. Here, attention is paid to the spring element in the center of the figure. i and j are nodes on both sides of this spring element. The spring force acting on these two nodes is f _ij , the natural length of the spring is L ₀ ,
The length displaced by the balance of the forces applied to the nodes is L, the nodal forces applied to the node i from the external pressure side are P and P ', the nodal forces applied to the node j from the external pressure side are P and P ″, and the internal pressure side is the node i. Assuming that the applied node forces are P _i and P _i ′ and the node forces applied to the node j from the external pressure side are P _j and P _j ′, the displaced spring length L can be obtained by solving the following simultaneous equations. Note that f _ij and P, P ′, P ″, P _i , P _i ′, P _j , and P _j ′, which are not _abbreviated as (scalar amount), represent vectors. The nodal forces P ′, P ″, P _i ′, and P _j ′ are nodal forces similarly applied to adjacent spring elements.

F _ij (scalar quantity) = k (L−L ₀ ) At node i, P + P′−f _ij + P _i + P _i ′ = 0 At node j, P + P ″ + f _ij + P _j + P _j ′ = 0 Next, , After the displacement of the nodes, the spring ring is reconfigured, and the length of each inter-node spring is determined as the natural length L ₀ of the new spring, where the density distribution on the ring is newly obtained. Returning to the process of extracting the concentration distribution on the spring element, the calculation of the nodal force and the calculation of the simultaneous equations of the spring are repeated, and the nodal force on the external pressure side and the nodal force on the internal pressure side become equal by the above-described equilibrium determination. By the way, an equal pressure line = an equal concentration line is obtained.

Since this ring-shaped spring model is a one-dimensional spring continuous model, the number of elements of the simultaneous equations is small and the processing is extremely fast. For this reason, processing can be performed by software processing, and a dedicated image processing board is not required, and simplification of the system configuration and low cost can be realized. In addition, since the software processing method allows a degree of design freedom, it is possible to develop an algorithm that is optimal for the measurement target, and improve the measurement accuracy.

The internal pressure model is dual with the external pressure model, and can be processed in the same manner as the external pressure model.

Further, since the present invention can be processed by a software processing method, a program for causing a computer to execute the above-described procedure of the present method is stored in a computer-readable recording medium (for example, a floppy disk or a floppy disk). It can be recorded on a CD-ROM or the like and distributed.

Further, in the above embodiment, the case where the arbitrary shape of the cream solder on the printed circuit board is detected has been described as an example. However, the present invention can be applied to various fields as exemplified below.

(1) Field of Printed Circuit Boards, etc. Detection, measurement, and inspection of solder and mounted components on printed circuit boards, and detection and positioning of reference marks and lands on printed circuit boards. In particular, detection, measurement and inspection of other types of irregularly-reflected solder (cream solder, solder after reflow, etc.) on printed circuit boards, and detection and positioning of irregularly-reflected solder levelers, solder plating reference marks and lands .

(2) Fields such as screen printing masks Detection of reference marks on masks and openings of circuit patterns
Positioning.

(3) Field of mounted components Detection and positioning of reference marks and lands of mounted components such as flip chips, CSPs, BGAs and MCMs, and detection, measurement and inspection of bumps of flip chips, CSPs, BGAs and MCMs .

[0044]

As is clear from the above description, according to the present invention, the unevenness of the irregular reflection of a measurement object having an arbitrary shape such as a cream solder or a solder after reflow is smoothed to perform measurement and inspection. Easy and smooth contour lines and contours can be extracted. In the extraction processing, since the one-dimensional spring coupled system is solved, the processing can be performed at high speed. Therefore, it can be processed by software processing,
A dedicated image processing board is not required, which simplifies the system configuration and lowers costs, and also provides design flexibility, which makes it possible to develop the most suitable algorithm for the measurement target and improve measurement accuracy. it can.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an exemplary configuration for performing an exemplary embodiment of a method according to the present invention.

FIG. 2 is a flowchart showing one embodiment of the method of the present invention.

FIG. 3A is a view showing a cross section of a cream solder concentration distribution (shading unevenness) in the embodiment, and FIG.
It is a figure showing the example (external pressure model) of the ring-shaped spring model in the above-mentioned embodiment.

FIGS. 4A and 4B are diagrams for explaining the nodal force conversion of the density distribution in the embodiment, wherein FIG. 4A is a cross-sectional view and FIG.

FIG. 5 is a diagram illustrating a balance calculation of a spring element in the embodiment.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 lighting device 2 imaging device 3 storage device 4 processing device 5 display device

Claims (4)

[Claims] A step of connecting a spring element having nodes at both ends in a ring shape in association with an isodensity line or a shape contour of an object on an image to form a ring; A step of applying a pressure distribution corresponding to a specified concentration to one inner side, and applying a pressure distribution corresponding to a concentration distribution on the ring to the other side outside and inside the ring; Moving the nodes of the spring element so that the applied pressure distribution is balanced through the nodes of the spring element; and iso-density lines corresponding to the specified density by changing the shape of the ring at the time when all of the spring elements are balanced. Or a procedure for forming a shape contour, and a method for measuring by image recognition, comprising: 2. In the step of moving the node, the pressure distribution applied to the outside of the ring and the pressure distribution applied to the inside of the ring to the spring element are converted into equivalent node forces, and the converted node forces are balanced. The method according to claim 1, wherein the node of the spring element is moved as described above. 3. In the step of moving the node, a simultaneous equation of a spring is solved so that the converted node force and the node stress of the spring element are balanced at the node. The method according to claim 2, wherein the node is moved. 4. A program for causing a computer to execute a procedure in the measurement method based on image recognition according to claim 1, wherein the program is recorded on a medium readable by the computer. recoding media.