CN104574408A - Industry transparent film package detecting method and device based on shape feature extraction - Google Patents

Abstract

The invention relates to an industry transparent film package detecting method and device based on shape feature extraction. The method comprises the steps that preprocessing is conducted on an obtained packaging product image, super pixel segmentation is conducted on the preprocessed image, seven Hu invariant moments are used for defining a shape feature to achieve the shape feature extraction, and new parameters are introduced; a multivariable parameter matrix is processed, and then principal components are obtained; a least-square support vector machine is used for conducting training based on the shape feature, and then superpixel blocks are classified. The device comprises a conveyor belt module, an image shooting module, an algorithm module and a result analysis module. The industry transparent film package detecting method and device based on the shape feature extraction can reduce the detecting time and maintain higher classification performance at the same time.

Description

The industrial transparent film packaging detection method of Shape-based interpolation feature extraction and device

Technical field

The present invention relates to transparent film packaging detection technique field, particularly relate to a kind of industrial transparent film packaging detection method and device of Shape-based interpolation feature extraction.

Background technology

Often there is defects such as encapsulating bubble, fold, relax in product external packaging film, has a strong impact on the image of product.Because packing film transmittance is high, the artificial naked eyes of existing many employings detect, and Detection results is undesirable, and the experience of operator and technical ability affect comparatively large, automatic production line cannot realize carry out Aulomatizeted Detect to the heat-sealing defect of product external packaging film.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of industrial transparent film packaging detection method and device of Shape-based interpolation feature extraction, can reduce detection time, and keep higher classification performance simultaneously.

The technical solution adopted for the present invention to solve the technical problems is: the industrial transparent film packaging detection method providing a kind of Shape-based interpolation feature extraction, comprises the following steps:

(1) pre-service is carried out to the packaging product image obtained;

(2) pretreated image is carried out to the segmentation of super-pixel;

(3) utilize seven Hu not bending moment definition shape facility realize Shape Feature Extraction, and introduce new parameter;

(4) multivariable parameter matrix is processed, obtain major component;

(5) use least square method supporting vector machine to train according to shape facility, then super-pixel block is classified.

Described packaging product image uses bowl-shape light source and wide-angle lens to obtain.

In described step (1), employing first utilizes Canny rim detection, and recycling expansive working carries out pre-service to the packaging product image obtained.

Described step (2) adopts the mode of simple linear iteration cluster to carry out super-pixel segmentation.

In described step (3) seven Hu not bending moment be respectively:

φ ₁＝η ₂₀+η ₀₂；

${\mathrm{\φ}}_2={({\mathrm{\η}}_{20}-{\mathrm{\η}}_{02})}^2+4{\mathrm{\η}}_{11}^2;$

φ ₃＝(η ₃₀-3η ₁₂) ²+(3η ₂₁-η ₀₃) ²；

φ ₄＝(η ₃₀+η ₁₂) ²+(η ₂₁+η ₀₃) ²；

φ ₅＝(η ₃₀-3η ₁₂)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]

+(3η ₂₁-η ₀₃)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]；

φ ₆＝(η ₂₀-η ₀₂)[(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]+4η ₁₁(η ₃₀+η ₁₂)(η ₂₁+η ₀₃)；

φ ₇＝(3η ₂₁-η ₀₃)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]+

(η ₀₃-η ₁₂)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]；

η _ijrepresent normalization center, (i+j) rank square of image.The center square of (i+j) of the f (x+y) of image function is defined as wherein Ω is the interval of x, y.For the digital picture of N*M, utilize summation to replace integration, then (i+j) center, rank can be expressed as then format center, (i+j) rank square can be expressed as ${\mathrm{\η}}_{\mathrm{ij}}=\frac{{\mathrm{\μ}}_{\mathrm{ij}}}{{{\mathrm{\μ}}^{\mathrm{\γ}}}_{00}},$ Wherein $\mathrm{\γ}=\frac{i+j}{2}+1(i+j=\mathrm{2,3},...);$

Wherein, the normalized moments of seven Hu not bending moment is constant to translation, convergent-divergent, stretching, extension and extruding change; The normalization center square of the first six Hu not bending moment is to invariable rotary; The normalization center square of the 7th Hu not bending moment is also constant to distortion to invariable rotary.

Introduce new parameter in described step (3) to comprise: area, girth, density, hole number, hole number and area ratio; Wherein, area: be used for calculating the pixel count that comprises of hole; Girth: on the outline line of hole, pel spacing is measured from sum; Density: wherein S is area, and L is girth; Hole number: the hole number in a packaging; Hole number and area ratio: be used for distinguishing macroscopic void and small holes.

Described step (4) middle use least square method supporting vector machine chooses 1000 holes and 2000 non-holes are trained, and uses the sorter of training out to classify to the super-pixel split.

The technical solution adopted for the present invention to solve the technical problems is: the industrial transparent film packaging pick-up unit also providing a kind of Shape-based interpolation feature extraction, comprising: conveyor belt module, for transmitting packaging product; Picture shooting module, is positioned at directly over conveyor belt module, for obtaining the image transmitting packaging product; Algoritic module, is connected with described picture shooting module, for carrying out picture processing according to above-mentioned detection method; Results analyses module, for carrying out analysis beneficial effect to the result of picture processing

Owing to have employed above-mentioned technical scheme, the present invention compared with prior art, there is following advantage and good effect: the Shape Feature Extraction that the present invention adopts not only make use of seven Hu not bending moment definition shape facility, and introduce new parameter, also use principal component analysis (PCA) and carry out dimension-reduction treatment, and utilize LSSVM to train, then the sorter trained is classified to super-pixel block.Method of the present invention not only reduces the time, and effect is also relatively good, industrially has feasibility.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention;

Fig. 2 uses apparatus structure schematic diagram of the present invention.

Embodiment

Below in conjunction with specific embodiment, set forth the present invention further.Should be understood that these embodiments are only not used in for illustration of the present invention to limit the scope of the invention.In addition should be understood that those skilled in the art can make various changes or modifications the present invention, and these equivalent form of values fall within the application's appended claims limited range equally after the content of having read the present invention's instruction.

Embodiments of the present invention relate to a kind of industrial transparent film packaging detection method of Shape-based interpolation feature extraction, as shown in Figure 1, comprise the following steps:

(1) pre-service is carried out to the packaging product image obtained;

(2) pretreated image is carried out to the segmentation of super-pixel;

(3) utilize seven Hu not bending moment definition shape facility realize Shape Feature Extraction, and introduce new parameter;

(4) multivariable parameter matrix is processed, obtain major component;

(5) use least square method supporting vector machine to train according to shape facility, then super-pixel block is classified.

Below each step is described in detail.

Pre-service:

First Canny rim detection is utilized, recycling expansive working.

Super-pixel is split:

Be the two-dimensional matrix of pixel composition to the understanding of image, so segmentation is also based on pixel, but the segmentation based on pixel can cause treatment effeciency too low in the past, and what therefore the present invention adopted is super-pixel segmentation.Super-pixel refers to that be communicated with in regional area in image, brightness or the close pixel of color set.What the present invention adopted is the segmentation of SLIC (Simple LinearIterative Clustering) super-pixel.

Feature extraction

What the present invention adopted method is seven Hu not bending moment definition shape facilities, and the new parameter introduced.The expression of seven Hu not bending moment is as follows:

φ ₁＝η ₂₀+η ₀₂

${\mathrm{\φ}}_2={({\mathrm{\η}}_{20}-{\mathrm{\η}}_{02})}^2+4{\mathrm{\η}}_{11}^2$

φ ₃＝(η ₃₀-3η ₁₂) ²+(3η ₂₁-η ₀₃) ²

φ ₄＝(η ₃₀+η ₁₂) ²+(η ₂₁+η ₀₃) ²

φ ₅＝(η ₃₀-3η ₁₂)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]+

(3η ₂₁-η ₀₃)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]

φ ₆＝(η ₂₀-η ₀₂)[(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]+4η ₁₁(η ₃₀+η ₁₂)(η ₂₁+η ₀₃)

φ ₇＝(3η ₂₁-η ₀₃)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]+

(η ₀₃-η ₁₂)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]

Wherein, η _ijrepresent normalization center, (i+j) rank square of image.The center square of (i+j) of the f (x+y) of image function is defined as wherein Ω is the interval of x, y.For the digital picture of N*M, utilize summation to replace integration, then (i+j) center, rank can be expressed as then format center, (i+j) rank square can be expressed as ${\mathrm{\η}}_{\mathrm{ij}}=\frac{{\mathrm{\μ}}_{\mathrm{ij}}}{{{\mathrm{\μ}}^{\mathrm{\γ}}}_{00}},$ Wherein $\mathrm{\γ}=\frac{i+j}{2}+1(i+j=\mathrm{2,3},...).$

Through calculating invariant moment features be F _m=φ ₁, φ ₂.... φ ₇, wherein the value of High Order Moment is very little, therefore needs to carry out standardization when coupling, and normalized moments is constant to translation, convergent-divergent, stretching, extension and extruding change.In addition, front 6 normalization center squares are to invariable rotary, and the 7th also constant to distortion.

Although these seven squares that well shape facility can be described, when image data base is larger, only these seven scalars are inadequate, and the present invention has introduced new parameter: area, girth, density, hole number, hole number and area ratio.Shown in being expressed as follows of area, girth, density, hole number, hole number and area ratio:

1) area: be used for calculating the pixel count that comprises of hole;

2) girth: on the outline line of hole, pel spacing is measured from sum, the distance between pixel arranged side by side is 1 pixel, and between vergence direction, the distance of pixel is when carrying out circumferential measurements, needs to calculate distance respectively according to connected mode between pixel;

3) density: wherein S is area, and L is girth;

4) hole number: calculate the hole number in a packaging;

5) hole number and area ratio: be mainly used to distinguish macroscopic void and small holes.

Principal component analysis (PCA)

The main thought of principal component analysis (PCA) carries out matrix disposal to multivariable parameter matrix, and what obtain is the linear combination of original variable, and two pairwise uncorrelateds, the information that original variable comprises can be reacted to greatest extent.

LSSVM classifies

Use least square method supporting vector machine (Least Squares Support Vector Machine, LSSVM) that the super-pixel of segmentation is categorized as hole and non-hole.For least square method supporting vector machine, optimization problem can be expressed as:

$\min J(w,\mathrm{\ξ})=\frac{1}{2}w\·w+\mathrm{\γ}\underset{i=1}{\overset{t}{\mathrm{\Σ}}}{\mathrm{\ξ}}_i^2---\left(4.9\right)$

Use Lagrange to solve above-mentioned optimization problem, be converted to and solve a linear equation problem.

In the training of LSSVM, choose 1000 holes, 2000 non-holes are trained, and namely train 3000 groups of data.Use the sorter of training out to classify to the super-pixel split, each super-pixel is divided into hole and non-hole.

Second embodiment of the present invention relates to the industrial transparent film packaging pick-up unit of Shape-based interpolation feature extraction, as shown in Figure 2, comprising: conveyor belt module 1, picture shooting module 2, algoritic module 3, results analyses module 4.Conveyor belt module 1 is made up of travelling belt, its effect at the uniform velocity transmits industrial packaging product, picture shooting module 2 is positioned at directly over conveyor belt module, effect uses bowl-shape light source and wide-angle lens carry out taking thus obtain the image of packaging product, the effect of algoritic module 3 carries out picture processing according to above-mentioned detection method, and the effect of results analyses module 4 analyzes the result of picture processing.

Be not difficult to find, the Shape Feature Extraction that the present invention adopts not only make use of seven Hu not bending moment definition shape facility, and introduce new parameter, also use principal component analysis (PCA) and carry out dimension-reduction treatment, and utilize LSSVM to train, then the sorter trained is classified to super-pixel block.Method of the present invention not only reduces the time, and effect is also relatively good, industrially has feasibility.

Claims (8)

1. an industrial transparent film packaging detection method for Shape-based interpolation feature extraction, is characterized in that, comprise the following steps:

(1) pre-service is carried out to the packaging product image obtained;

(2) pretreated image is carried out to the segmentation of super-pixel;

(3) utilize seven Hu not bending moment definition shape facility realize Shape Feature Extraction, and introduce new parameter;

(4) multivariable parameter matrix is processed, obtain major component;

(5) use least square method supporting vector machine to train according to shape facility, then super-pixel block is classified.

2. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, is characterized in that, described packaging product image uses bowl-shape light source and wide-angle lens to obtain.

3. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, it is characterized in that, in described step (1), employing first utilizes Canny rim detection, and recycling expansive working carries out pre-service to the packaging product image obtained.

4. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, is characterized in that, described step (2) adopts the mode of simple linear iteration cluster to carry out super-pixel segmentation.

5. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, is characterized in that, in described step (3) seven Hu not bending moment be respectively:

φ ₁＝η ₂₀+η ₀₂；

φ ₂＝(η ₂₀-η ₀₂) ²+4η ₁ ² ₁；

φ ₃＝(η ₃₀-3η ₁₂) ²+(3η ₂₁-η ₀₃) ²；

φ ₄＝(η ₃₀+η ₁₂) ²+(η ₂₁+η ₀₃) ²；

φ ₅＝(η ₃₀-3η ₁₂)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]

+(3η ₂₁-η ₀₃)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]；

φ ₆＝(η ₂₀-η ₀₂)[(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]+4η ₁₁(η ₃₀+η ₁₂)(η ₂₁+η ₀₃)；

φ ₇＝(3η ₂₁-η ₀₃)(η ₃₀+η ₁₂)[(η ₃₀+η ₁₂) ²-3(η ₂₁+η ₀₃) ²]+

(η ₀₃-η ₁₂)(η ₂₁+η ₀₃)[3(η ₃₀+η ₁₂) ²-(η ₂₁+η ₀₃) ²]；

η _ijrepresent normalization center, (i+j) rank square of image;

Wherein, the normalized moments of seven Hu not bending moment is constant to translation, convergent-divergent, stretching, extension and extruding change; The normalization center square of the first six Hu not bending moment is to invariable rotary; The normalization center square of the 7th Hu not bending moment is also constant to distortion to invariable rotary.

6. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, it is characterized in that, introduce new parameter in described step (3) and comprise: area, girth, density, hole number, hole number and area ratio; Wherein, area: be used for calculating the pixel count that comprises of hole; Girth: on the outline line of hole, pel spacing is measured from sum; Density: wherein S is area, and L is girth; Hole number: the hole number in a packaging; Hole number and area ratio: be used for distinguishing macroscopic void and small holes.

7. the industrial transparent film packaging detection method of Shape-based interpolation feature extraction according to claim 1, it is characterized in that, described step (4) middle use least square method supporting vector machine chooses 1000 holes and 2000 non-holes are trained, and uses the sorter of training out to classify to the super-pixel split.

8. an industrial transparent film packaging pick-up unit for Shape-based interpolation feature extraction, is characterized in that, comprising: conveyor belt module, for transmitting packaging product; Picture shooting module, is positioned at directly over conveyor belt module, for obtaining the image transmitting packaging product; Algoritic module, is connected with described picture shooting module, for carrying out picture processing according to the detection method as described in claim arbitrary in claim 1-7; Results analyses module, for analyzing the result of picture processing.