CN104772880B - Injection molding mechanical arm mold anomaly detection method based on LMDO (Local Multilayered Difference Operator) - Google Patents

Abstract

The invention provides an injection molding mechanical arm mold anomaly detection method based on an LMDO (Local Multilayered Difference Operator). The anomaly detection method comprises the following steps: (1) acquiring a standard template image when an injection molding machine opens a mold in place, and pre-processing to obtain a later difference background image; (2) waiting for working state information of the injection molding machine; upon detection of a situation that the injection molding machine is operated until the mold is opened in place, continuously acquiring the image of a mold cavity by a camera, extracting an average image of the plurality of images, and pre-processing the average image to do preparation for subsequent image processing, thereby obtaining a later difference foreground image; and (3) carrying out an anomaly detection algorithm based on the LMDO on the difference foreground image and the difference background image to obtain an abnormal region without a light illumination interference part. The injection molding mechanical arm mold anomaly detection method based on the LMDO, provided by the invention, has the characteristics of good instantaneity, strong robustness on illumination variation and the like; and whether the mold has an abnormal state or not can be monitored through mold opening information of the injection molding machine.

Description

A kind of injection molding mechanical arm mould method for detecting abnormality based on LMDO

Technical field

It is a kind of injection machine production process equipped with manipulator the present invention relates to injection machine industry in industrial automation In to the intelligent surveillance device of mould.

Background technology

Recently as the continuous expansion of plastic products application field, global manufacturing shows to the demand of injection machine holds Continuous unprecedented soaring trend.The injection machine of modernization is equipped with manipulator, and manipulator is the part work(that can imitate human arm Can, injecting products can be automatically taken out, it stacked to product according to pre-provisioning request, arranged and put.Machinery Hand is the automated production equipment exclusively for Injection Molding Machine Design exploitation, and it can mitigate the heavy manual labor of workman, improve labor Dynamic condition ensures safety in production, and improve production efficiency plays extremely important work the aspects such as the market competitiveness of enterprise are strengthened With.Nowadays, using generally, safeguard protection and module protection during manipulator automatic clamping material are all the more for manipulator of injection machine It is important.In order to the residual that exists to mould in the injection machine production process with manipulator, sliding block dislocation, the demoulding are bad etc. abnormal Situation carries out monitor in real time to realize the automatic protection to mould, and mould protector introduces the injection machine with manipulator by a large amount of In industry.

This monitoring system based on image processing techniques greatly improves the security in injection machine mould production process And operating efficiency, largely reduce the labour intensity of repair a die cost and staff.Mould protection in the market Device function is all relatively simple, and the image processing algorithm of the software that mold protecting device is used is relatively easy, intelligence degree phase To relatively low, the system of result in generally requires the manual calibration and experience of complexity to ensure the correctness of its abnormality detection result.But Be generally with manipulator injection machine where working environment can be sufficiently complex, largely illumination is had often Change, and use traditional algorithm mould protecting can not in this changeable environment normal work, easily frequently go out The problems such as existing abnormal area false alarm.At this time user needs the relevant parameter of frequently adjustment system, or directly again Study, the problem for causing cumbersome and operating efficiency low.

The content of the invention

In order to overcome the real-time of existing injection molding mechanical arm mould monitoring method not enough, poor to illumination variation robustness etc. Deficiency, the present invention provide it is a kind of with real-time it is good, to illumination variation strong robustness the features such as it is poor based on Local Multilevel time Divide the method for detecting abnormality of operator (LMDO), whether it there can be abnormal shape by injection moulding machine mould open image information supervision molding State.

The technical solution adopted for the present invention to solve the technical problems is：

A kind of injection molding mechanical arm mould method for detecting abnormality based on LMDO, the method for detecting abnormality includes following mistake Journey：

1) standard form image when gathering injection moulding machine mould open in place, and pre-processed, the substracting background after Image；

2) injection machine work state information is waited, is run when molding in place, by video camera pair when injection machine is detected The continuous capture of mould cavity surface, extracts the average image of a few width images, and the average image is pre-processed, it is at successive image Reason is prepared, the difference foreground image after；

3) Outlier Detection Algorithm based on LMDO is performed to difference foreground image and substracting background image, obtains excluding illumination The abnormal area of interference sections；Process is as follows：

3.1) Local Multilevel time difference operator

It is R in radius, neighbours' point is the image local area of P, the contrast level corresponding to each neighbours' point pixel value t_pGained is calculated by following formula：

g_s=g_p-g_c (1)

In formula, g_pIt is the P gray value of neighbours' point, g_cIt is the gray value of regional area central pixel point, maxC, minC point The contrast maximum and minimum value between neighbours' point and central point are not represented, and T represents contrast layering quantity；For each Contrast level i, the contrast value of how many neighbours point fall into the layer around observation central point, to obtain every level LMDO：

By each levelIt is concatenated together constituting LMDO_P.R:

3.2) difference is carried out to background and foreground image using LMDO values

According to T levelValue judges the phase knowledge and magnanimity between pixel, i-th layer of note background image certain point It is secondary8bit binary values be Blmdo (j), j=0...7, i-th layer of the foreground image point's 8bit binary values are Flmdo (j), and both difference obtain the similarity of the i-th level：

Wherein,Be with or；

Seeking the similarity between pixel must obtain the weight of each level, then by the similarity S of each level_iWith it is respective Multiplied by weight, obtain overall similarity：

Wherein, a_iI-th weight of level is represented, is shown by great number tested data, work as a₁=a₂=... a_TWhen, compare The similarity best results of two images；If similarity S is less than a given threshold Q, the pixel is before being identified as Scape, then value is gray level 0, is otherwise background, and its value is gray level 255, so as to realize the segmentation of abnormal area；

4) continuous open and close computing is carried out to abnormal area image；

5) abnormal area is marked by region-growing method and area measurement；

6) extracted finally by morphological Edge and obtain abnormal area profile and alarm, injection machine stops pressing mold and enters chain guarantor Shield measure.

Beneficial effects of the present invention are mainly manifested in：Using the method for detecting abnormality based on Local Multilevel time difference operator, Efficiently solve the problems, such as the system flase drop that illumination variation causes.Meanwhile, LMDO has less computation complexity and good Texture features, major part is comparison operation, realizes simple, improves image abnormity detection efficiency.These are all prior arts The key factor of the influence system not accounted for, this Outlier Detection Algorithm greatly strengthen the real-time and robust of system Property.

Brief description of the drawings

Fig. 1 is the structured flowchart of the die sinking method for detecting abnormality based on LMDO.

Fig. 2 is the acquisition process figure of foreground image LMDO values.

Fig. 3 is the acquisition process figure of background image LMDO values.

Fig. 4 is the learning process figure of background image.

Fig. 5 is the die sinking method for detecting abnormality flow chart based on LMDO.

Fig. 6 is the image after background LMDO figures and prospect LMDO figures difference and Threshold segmentation.

Specific embodiment

The invention will be further described below in conjunction with the accompanying drawings.

1~Fig. 6 of reference picture, a kind of injection molding mechanical arm mould method for detecting abnormality based on LMDO, the abnormality detection side Method includes procedure below：

1) standard form image when gathering injection moulding machine mould open in place, and pre-processed, the substracting background after Image；

2) injection machine work state information is waited, is run when molding in place, by video camera pair when injection machine is detected The continuous capture of mould cavity surface, extracts the average image of a few width images, and the average image is pre-processed, it is at successive image Reason is prepared, the difference foreground image after；

3) Outlier Detection Algorithm based on LMDO is performed to difference foreground image and substracting background image, obtains excluding light According to the abnormal area of interference sections；Process is as follows：

3.1) Local Multilevel time difference operator

LMDO is a kind of a certain pixel of comparing and its neighborhood territory pixel value, and then produces the contrast between pixel, then Contrast value is mapped to certain level up, the LMDO values of many levels are obtained, using LMDO values as characteristics of image, is used for Recognize that the image difference in the environment of illumination variation has extraordinary effect.

It is R in radius, neighbours' point is the image local area of P, the contrast level corresponding to each neighbours' point pixel value t_pGained is calculated by following formula：

g_s=g_p-g_c (1)

In formula, g_pIt is the P gray value of neighbours' point, g_cIt is the gray value of regional area central pixel point, maxC, minC point The contrast maximum and minimum value between neighbours' point and central point are not represented, and T represents contrast layering quantity.For each Contrast level i, the contrast value that can observe how many neighbours point around central point fall into the layer, to obtain every level LMDO：

By each levelIt is concatenated together constituting LMDO_P.R:

3.2) difference is carried out to background and foreground image using LMDO values

According to T levelValue judges the phase knowledge and magnanimity between pixel, i-th layer of note background image certain point It is secondary8bit binary values be Blmdo (j) (j=0...7), i-th layer of the foreground image point's 8bit binary values are Flmdo (j), and both difference obtain the similarity of the i-th level：

Wherein,Be with or.

Because the pixel value of each level occupies certain contrast in regional area, the similarity between pixel is sought Must obtain the weight of each level, then by the similarity S of each level_iWith respective multiplied by weight, obtain overall similar Degree：

Wherein, a_iI-th weight of level is represented, is shown by great number tested data, work as a₁=a₂=... a_TWhen, compare The similarity best results of two images.If similarity S is less than a given threshold Q, the pixel is before being identified as Scape, then value is gray level 0, is otherwise background, and its value is gray level 255, so as to realize the segmentation of abnormal area；

4) continuous open and close computing is carried out to abnormal area image to eliminate influence of noise；

5) abnormal area is marked by region-growing method and area measurement；

6) extracted finally by morphological Edge and obtain abnormal area profile and alarm, injection machine stops pressing mold and enters chain guarantor Shield measure；If without abnormal area, continuing waiting for the injection moulding machine mould open status information in next cycle.

Whole system device it is main by：Analog video camera, Video Decoder, core dsp processor, touch-screen, liquid crystal Show that device, keyboard etc. are constituted.Whole exception handling procedure：First, in the case of injection moulding machine mould open state, held by keyboard button Row study instruction, makes camera gather standard form image, and is pre-processed, to eliminate unrelated information in image, as It is used for the background image of difference afterwards, and stores information in the memory module of DSP core control panel (TMS320DM6437), As shown in Figure 4；Secondly, as shown in figure 5, until molding in place, system is in the monitoring state, by analog video camera to mould The continuous capture of cavity surface, by DSP core control panel to camera acquisition to a few width image zooming-out the average images, and to mean chart It is that successive image treatment is prepared, as the foreground image for difference as pre-processing；Again by background image and foreground image Difference is carried out, and the image after difference is realized into binaryzation using Threshold segmentation；And continuous open and close fortune is carried out to image Calculate to eliminate influence of noise；Abnormal area is marked by region-growing method and area measurement；Finally by air cavity detection Check whether there is exception, if exception, in touch-screen display alarm information, injection machine stops pressing mold arranging into interlock protection Apply；Injection machine work state information when otherwise continuing waiting for die sinking in place.

The method for detecting abnormality based on LMDO, specific steps：

As shown in Fig. 2 being 1 in radius, neighbours' point is 8 image local area, pixel value g_cCentered on=130 point Point, each pixel value of neighbours' point pixel value from the upper left corner as corresponding to rotate counterclockwise is respectively 134,27,60,4,127, 221,82,187.Contrast layering quantity T=4 is taken, the LMDO values of background image are asked for, obtained first corresponding to each neighbours' point Contrast level t_pCalculated by following formula：

For each contrast level i (i=1 ... 4), the contrast of how many neighbours point around central point can be observed Angle value fall into the layer, to obtain the LMDO values of every level：

By each levelIt is concatenated together constituting LMDO_8.1:

As shown in figure 3, being 1 in radius, neighbours' point is 8 image local area, pixel value g_cCentered on=120 point Point, each pixel value of neighbours' point pixel value from the upper left corner as corresponding to rotate counterclockwise is respectively 134,47,58,6,140, 231,75,160.Contrast layering quantity T=4 is equally taken, the LMDO values of foreground image are asked for, each neighbours' point institute is obtained first Corresponding contrast level t_pCalculated by following formula：

For each contrast level i (i=1 ... 4), the contrast of how many neighbours point around central point can be observed Angle value fall into the layer, to obtain the LMDO values of every level：

By each levelIt is concatenated together constituting LMDO_8.1:

Below according to 4 levelsValue judges the phase knowledge and magnanimity between pixel, the of note background pixel value 130 I levels8bit binary values be Blmdo (j) (j=0...7), i-th layer of foreground pixel value 120 8bit binary values be Flmdo (j), both difference obtain the similarity of the i-th level：

Because the pixel value of each level occupies certain contrast in regional area, the similarity between pixel is sought Must obtain the weight of each level, then by the similarity S of each level_iWith respective multiplied by weight, obtain overall similar Degree：

Take a₁=a₂=... a_T=1

If similarity S is less than a given threshold Q, the pixel is to be identified as prospect, then value is gray level 0, Otherwise it is background, its value is gray level 255, so as to realize the segmentation of abnormal area；

Under mould die opening state, the template image gathered by camera and is pre-processed, and make an uproar in image with being eliminated The image of sound and unrelated information.When mould is molded in place, under system is in the monitoring state, by video camera to mould type The continuous capture of Cavity surface, extracts the average image of a few width images, and pretreated image is done to the average image.Arrived in mould die sinking Many moulds do not fall off down during position, can influence the efficiency of system.

As shown in fig. 6, be background image and foreground image by the result figure obtained by LMDO abnormality processings, black part in figure It point is exactly result that abnormal area is split.

This LMDO method for detecting abnormality can effectively remove the influence of illumination variation, and computing is simple, accelerates exception The speed of detection, effectively improves operating efficiency.

Claims (1)

1. a kind of injection molding mechanical arm mould method for detecting abnormality based on LMDO, it is characterised in that：The method for detecting abnormality bag Include procedure below：

1) standard form image when gathering injection moulding machine mould open in place, and pre-processed, the substracting background image after；

2) injection machine work state information is waited, is run when molding in place, by video camera to mould when injection machine is detected The continuous capture of cavity surface, extracts the average image of a few width images, and the average image is pre-processed, and is that standard is done in successive image treatment It is standby, the difference foreground image after；

3) Outlier Detection Algorithm based on LMDO is performed to difference foreground image and substracting background image, obtains excluding illumination interference Partial abnormal area；Process is as follows：

3.1) Local Multilevel time difference operator

It is R in radius, neighbours' point is the image local area of P, the contrast level t corresponding to each neighbours' point pixel value_pUnder Formula calculates gained：

g_s=g_p-g_c (1)

In formula, g_pIt is the P gray value of neighbours' point, g_cIt is the gray value of regional area central pixel point, maxC, minC generation respectively Contrast maximum and minimum value between table neighbours point and central pixel point, T represent contrast layering quantity；For each Contrast level i, the contrast value of how many neighbours point fall into the layer around observation central pixel point, to obtain every level LMDO：

${\mathrm{LMDO}}_{P.R}^i=\underset{p=0}{\overset{P-1}{\mathrm{\Σ}}}{2}^p\·C_p,C_p=\left\{\begin{array}{c}1,t_p=i\\ 0,t_p\≠i\end{array}\right.,i=1...T---\left(3\right)$

By each levelIt is concatenated together constituting LMDO_P.R:

${\mathrm{LMDO}}_{P.R}={\mathrm{LMDO}}_{P.R}^1/{\mathrm{LMDO}}_{P.R}^2/.../{\mathrm{LMDO}}_{P.R}^T$

3.2) difference is carried out to background and foreground image using LMDO values

According to T levelValue judges the similarity between pixel, the i-th level of note background image certain point8bit binary values be Blmdo (j), j=0...7, i-th layer of the foreground image point8bit bis- Hex value is Flmdo (j), and both difference obtain the similarity of the i-th level：

$S_i=\underset{j=0}{\overset{P-1}{\mathrm{\Σ}}}\left(Blmdo\right(j)\⊗Flmdo(j\left)\right)---\left(4\right)$

Wherein,Be with or；

Seeking the similarity between pixel must obtain the weight of each level, then by the similarity S of each level_iWith respective power Heavy phase multiplies, and obtains overall similarity：

$S=\underset{i=1}{\overset{T}{\mathrm{\Σ}}}(a_i\×S_i)---\left(5\right)$

Wherein, a_iRepresent i-th weight of level；If similarity S is less than a given threshold Q, the pixel is to be identified It is prospect, then value is gray level 0, is otherwise background, its value is gray level 255, so as to realize the segmentation of abnormal area；

4) continuous open and close computing is carried out to abnormal area image；

5) abnormal area is marked by region-growing method and area measurement；

6) extracted finally by morphological Edge and obtain abnormal area profile and alarm, injection machine stops pressing mold arranging into interlock protection Apply.