CN108510513A - A kind of PCB image circle detection method based on PCA and segmentation RHT - Google Patents

Abstract

The invention discloses a PCB image circle detection method based on PCA and segmented RHT, comprising the following steps: S1. loading the original color PCB board image; S2. performing grayscale binarization on the image, and extracting the image by using an edge detection algorithm Edge, remove points with more crossing times; S3. Mark the line segments in the grayscale binarized image, find out the line segments whose length is greater than the set threshold t; S4. Perform principal component analysis on each line segment to obtain the eigenvalue , keep the circle-like curve segment; S5. Carry out circle fitting on the circle-like curve segment to obtain the rough circle parameters, and filter out the effective curve segment from the circle-like curve segment; S6. Perform segmented circle detection on the effective curve segment to obtain the accurate circle parameters. The invention has the advantages of more accurate detection of circles, more effective removal of non-circular curves, smaller parameter errors of circles, and advantages in processing speed.

Description

A PCB Image Circle Detection Method Based on PCA and Segmented RHT

technical field

The present invention relates to the field of image processing and machine vision, and more specifically, to a PCB image circle detection method based on PCA and segmented RHT.

Background technique

At present, the commonly used method for printed circuit board (PCB) defect detection is the reference method, which registers the image of the board to be matched with the reference template, and the detection and positioning of circular holes is the most important. Among the existing inventions and technologies, some are only suitable for printed circuit board (PCB) images with relatively single content, some cannot guarantee better accuracy, and some processing will introduce a lot of noise, so the robustness is poor.

Contents of the invention

In order to overcome the above-mentioned defects in the prior art, the present invention provides a PCB image circle detection method based on PCA and segmented RHT.

In order to solve the problems of the technologies described above, the technical solution of the present invention is as follows:

A PCB image circle detection method based on PCA and segmented RHT, comprising the following steps:

S1. Load the original color PCB board image;

S2. Carry out grayscale binarization to the image, adopt the edge detection algorithm to extract the edge of the image, and remove the points with more crossing times;

S3. Mark the line segments in the grayscale binarized image, and find out the line segments whose length is greater than the set threshold t;

S4. Perform Principal Component Analysis (PCA) on each line segment to obtain eigenvalues and retain the circle-like curve segment;

S5. Carry out circle fitting to the circle-like curve segment to obtain rough circle parameters, and select effective curve segments in the circle-like curve segment;

S6. Perform segmented circle detection on the effective curve segment to obtain precise circle parameters.

The above working principle is as follows: firstly, grayscale binarization is performed on the image, and the edge of the image is extracted through the edge detection algorithm and points with a large number of intersections are removed; all line segments in the image are marked, and the circle-like curve is screened out through principal component analysis segment; the circle fitting is performed on the circle-like curve segment to obtain rough circle parameters, and the effective curve segment is screened out; finally, the circle detection is performed on the effective curve segment in a segmented manner to obtain accurate circle parameters.

Preferably, the specific steps of the PCA direction analysis in step S4 are as follows:

For each line segment, let the pixel be N, any point coordinates ( _xi , y _i ), according to the following formula:

S ₁₁ , S ₁₂ , S ₂₁ and S ₂₂ are obtained to form the following covariance matrix:

Get the characteristic roots λ ₁ and λ ₂ of S;

Determine whether the line segment is a circle-like curve segment satisfying the condition 1≤λ ₁ /λ ₂ ≤t, if yes, keep it, otherwise remove it.

Preferably, the specific process of circle fitting in the step S5 is as follows:

Using the least squares circle fitting method with constraints, the center and radius of the circle are approximated for screening curve segments.

Preferably, the segmented circle detection in the step S6 adopts Random Hough Transform (Random Hough Transform, RHT), and the specific process is as follows:

Let D be the retained image edge point set, mark the effective curve segment, and perform random Hough transform in sequence according to the marking order:

Randomly select 3 points from the curve segment point set, determine a candidate circle parameter, and calculate the number of points in the curve segment point set that fall on the candidate circle through evidence accumulation. If it is greater than the minimum number of points necessary for a circle, then confirm that the candidate circle is real circle, delete the points on the circle from D, and then proceed to the detection of the next circle until all the curve segments are detected.

Preferably, in the step S6, after the precise circle parameters are obtained, the corresponding pixel points are deleted in the image space until all the curve segments are detected.

Preferably, the threshold t is 1.5.

Preferably, the edge detection algorithm is a canny operator.

Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

On the basis of principal component analysis, the present invention adopts random Hough transform for segmented circle detection, more accurate detection of circles, more effective removal of non-circular curves, smaller parameter errors of circles, and advantages in processing speed .

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Figure 1 is a schematic diagram of a PCB image circle detection method based on PCA and segmented RHT.

Detailed ways

The accompanying drawings are for illustrative purposes only and cannot be construed as limiting the patent;

For those skilled in the art, it is understandable that some well-known structures and descriptions thereof may be omitted in the drawings.

The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

A kind of PCB image circle detection method based on PCA and segmentation RHT, as shown in Figure 1, comprises the following steps:

S1. Load the original color PCB board image;

S2. Carry out grayscale binarization to the image, adopt the edge detection algorithm to extract the edge of the image, and remove the points with more crossing times;

S3. Mark the line segments in the grayscale binarized image, and find out the line segments whose length is greater than the set threshold t;

S4. Carry out PCA direction analysis to each line segment, obtain eigenvalue, keep the circle-like curve segment;

S5. Carry out circle fitting to the circle-like curve segment to obtain rough circle parameters, and select effective curve segments in the circle-like curve segment;

S6. Perform segmented circle detection on the effective curve segment to obtain precise circle parameters.

In this embodiment, the specific steps of the PCA direction analysis in step S4 are as follows:

For each line segment, let the pixel be N, any point coordinates ( _xi , y _i ), according to the following formula:

S ₁₁ , S ₁₂ , S ₂₁ and S ₂₂ are obtained to form the following covariance matrix:

Get the characteristic roots λ ₁ and λ ₂ of S;

Determine whether the line segment is a circle-like curve segment satisfying the condition 1≤λ ₁ /λ ₂ ≤t, if yes, keep it, otherwise remove it.

In this embodiment, the specific process of circle fitting in step S5 is as follows:

Using the least squares circle fitting method with constraints, the center and radius of the circle are approximated for screening curve segments.

In this embodiment, the segmented circle detection in step S6 adopts random Hough transform, and the specific process is as follows:

Let D be the retained image edge point set, mark the effective curve segment, and perform random Hough transform in sequence according to the marking order:

Randomly select 3 points from the curve segment point set, determine a candidate circle parameter, and calculate the number of points in the curve segment point set that fall on the candidate circle through evidence accumulation. If it is greater than the minimum number of points necessary for a circle, then confirm that the candidate circle is real circle, delete the points on the circle from D, and then proceed to the detection of the next circle until all the curve segments are detected.

In this embodiment, after the accurate circle parameters are obtained in step S6, the corresponding pixel points are deleted in the image space until all the curve segments are detected.

In this embodiment, the threshold t is 1.5.

In this embodiment, the edge detection algorithm is a canny operator.

Apparently, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, on the basis of the above description, other changes or changes in different forms can also be made. It is not necessary and impossible to exhaustively list all the implementation manners here. All modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (7)

1. a kind of PCB image circle detection method based on PCA and segmentation RHT, which is characterized in that include the following steps：

S1. it is loaded into original color pcb board image；

S2. binarization of gray value is carried out to image, image border is extracted using edge detection algorithm, it is more to remove number of crossings Point；

S3. the line segment in binarization of gray value image is marked, finds out the line segment that length is more than set threshold value t；

S4. PCA Orientations are carried out to every line segment, obtains characteristic value, reserved category circular curve segment；

S5. round fitting is carried out to class circular curve segment, obtains rough Circle Parameters, efficiency curve section is filtered out in class circular curve segment；

S6. segmentation loop truss is carried out to efficiency curve section, obtains accurate Circle Parameters.

2. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described The PCA Orientations of step S4 are as follows：

For every line segment, if pixel is N, any point coordinates (x_i,y_i), according to following formula：

Obtain S₁₁,S₁₂,S₂₁And S₂₂, for constituting following covariance matrix：

Obtain the characteristic root λ of S₁And λ₂；

Judge whether line segment is to meet 1≤λ of condition₁/λ₂Otherwise the class circular curve segment of≤t just removes if it is, retaining.

3. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described The detailed process of circle fitting is as follows in step S5：

Using the Least Square Circle fitting process of belt restraining, approximation obtains the round center of circle and radius, is used for screening curve section.

4. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described Loop truss is segmented in step S6 uses randomized hough transform, detailed process as follows：

If D is the image border point set remained, efficiency curve section is marked, is carried out successively according to flag sequence random Hough transformation：

3 points are randomly selected from curved section point concentration, a candidate Circle Parameters is determined, passes through accumulation of evidence calculated curve section point set Decline the points on candidate's circle, if more than minimal point necessary to circle, then confirms candidate circle for true circle, from D The point on the circle is deleted, the detection of next circle is then proceeded by, is finished until all curved sections all detect.

5. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described Step S6 deletes corresponding pixel after obtaining accurate Circle Parameters, in image space, until all curved section detections finish.

6. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described Threshold value t is 1.5.

7. the PCB image circle detection method according to claim 1 based on PCA and segmentation RHT, which is characterized in that described Edge detection algorithm is canny operators.