CN111965195B - Monocular-based double-row needle bearing needle missing detection method - Google Patents

Abstract

The invention discloses a double-row needle bearing needle missing detection method based on a single eye. The machine vision is applied to the needle missing detection of the double-row needle roller bearing, so that the problems of high working strength, time consumption and labor consumption of a manual measurement mode are solved, and the problem of poor universality of contact measurement is also solved. The method can be used as a needle missing detection means for the double-row needle bearing, is suitable for the problems of needle missing detection, size, accurate positioning and the like of other needle bearings, and has good universality. As a flexible detection means, the double-row needle bearing needle missing detection device and method based on machine vision can be very conveniently transplanted to other related application fields such as detection, identification and classification.

Description

Monocular-based double-row needle bearing needle missing detection method

Technical Field

The invention relates to a method for detecting needle missing condition of a needle bearing, in particular to a method for detecting needle missing of a double-row needle bearing based on vision.

Background

Needle roller bearings are needle roller bearings with cylindrical rollers, and generally the diameter of the bearing itself is much larger than the diameter of the rollers. The double row roller bearing is a cylindrical roller comprising an upper layer and a lower layer (as shown in fig. 1), and the production and installation process is more complicated compared with the single row needle roller bearing. During the production process, the rollers are usually aligned in fixed positions on the upper and lower layers by mechanical means. However, due to uncertainties in the manufacturing and assembly process, it sometimes happens that a roller is not installed in a given position or that a roller or even some rollers are missing (so-called missing pins). When the needle is missing, the bearings need to be sorted out and reloaded into the missing rollers and enter the next bearing processing and manufacturing process. In the needle-missing detection method, a common bearing manufacturer adopts a manual method for detection, the method is time-consuming and labor-consuming, the working strength is high, and the risk of missed detection exists because the energy of people is limited.

In order to solve the problem of needle missing of the bearing, the related patents propose that the bearing with a certain difference between the weight of the bearing and the weight of a normal needle bearing is a needle-missing bearing by weighing the finished bearing. The detection method requires higher weighing precision, and particularly for bearings with small size and light weight, the measuring precision of the weighing device must be capable of distinguishing single needle roller. The hardware cost of the corresponding device of the method is higher, and the method is sensitive to the change of the environment, so that the measurement precision is difficult to ensure, and the accuracy of the missing needle detection is influenced; in addition, there is a patent to propose a probe detection apparatus and method. The needle-missing detection mechanism of the method comprises a linear displacement sensor and a controller, wherein a probe is arranged at the sliding end of the linear displacement sensor and is used for penetrating into a needle roller gap of a needle roller bearing on a bearing transmission channel. The diameter of the probe is larger than one half of the diameter of the needle roller and smaller than the diameter of the needle roller, and whether the needle roller is short or not is distinguished by judging the penetration distance of the probe or the needle roller is assembled according to requirements. The judgment basis of the method is that the difference between the penetration distance under the condition of needle missing and the penetration distance when the needle roller is full is large, and the judgment can be carried out without high-precision detection. However, it has the following problems as a method of contact measurement: 1) The processing precision requirement of the probe is high, and particularly under the condition that the diameter of the roller pin is small, the processing precision is difficult to guarantee; 2) The probe can only detect the current needle bearing, and if bearings with different diameters need to be detected, the universality is poor. In order to avoid the problems, the invention provides a method based on machine vision to solve the problem of needle shortage of the double-row needle roller bearing. The method comprises the steps of utilizing a monocular camera to collect images of the needle roller bearing in a certain direction, and utilizing an image processing method to judge whether the number of the obtained needle roller imaging sub-regions is consistent with the number of the actual needle rollers so as to judge whether the needles are missing.

Disclosure of Invention

The invention aims to solve the technical problems of needle missing detection of a double-row needle roller bearing, and the existing manual detection and contact detection methods have the defects of high labor intensity, high processing difficulty, poor detection flexibility and the like. In order to solve the technical problem, the invention provides a machine vision-based method for solving the problem of needle missing detection of a double-row needle roller bearing.

In order to achieve the purpose, the invention provides the following technical scheme:

a double-row needle bearing needle missing detection method based on a monocular comprises the following steps:

step S1: initializing, setting the numerical value of a counter, setting the numerical value of the counter to be M1 when a row of needle rollers close to the monocular camera lens are subjected to needle missing detection, and setting the numerical value of the counter to be M2 when a row of needle rollers far away from the monocular camera lens are converted into a row of needle rollers close to the camera lens are subjected to needle missing detection;

step S2: acquiring an image of the needle bearing by adopting a monocular camera;

and step S3: extracting independent sub-regions of the rolling needles through threshold segmentation; meanwhile, a ring area of the bearing is obtained through threshold segmentation, the outer boundary and the inner boundary of the ring are obtained through edge extraction, and fitting of the inner circle and the outer circle is further carried out through Hough transform to obtain the contours of the inner circle and the outer circle;

and step S4: comparing the number N1 of the needle roller independent sub-areas in the area outside the excircle with the number N0 of the needle rollers;

if N1= N0, performing step S5; if N1 is not equal to N0, comparing the sum of the number N2 and N1 of the needle rollers with pixel intersection between the internal independent sub-region and the internal circle with the size of N0, if N1+ N2= N0, executing the step S5, if N1+ N2 is not equal to N0, judging that the bearing is lack of the needle, eliminating the bearing and reassembling the needle rollers, and executing the steps S1-S4 again;

step S5: judging the value of the counter count;

if count = M1, the current detection is the first detection of the double rows of needle rollers of the bearing, and no needle is missing, the value of the counter is set to M2, and the steps S2-S5 are executed again; if count = M2, the current detection is the second detection of the double rows of needles of the bearing, and the bearing is transmitted to the next process without needle missing.

Furthermore, in step S2, the monocular camera collects images of the needle bearing under the action of the active light source.

Furthermore, the active light source is a ring light source.

Furthermore, the value of M1 is 0, and the value of M2 is 1.

Furthermore, the method for changing the row of the needle rollers far away from the lens of the monocular camera into the row of the needle rollers near the lens of the monocular camera comprises the following steps:

the needle bearing is fixed, and the camera is mirrored to detect a row of needles below, or;

and the needle bearing is turned over by 180 degrees, so that the positions of the lower row of needle rollers and the upper row of needle rollers are interchanged.

Compared with the prior art, the invention has the beneficial effects that:

the machine vision is applied to the needle missing detection of the double-row needle roller bearing, so that the problems of high working strength, time consumption and labor consumption of a manual measurement mode are solved, and the problem of poor universality of contact measurement is also solved. The method can be used as a needle missing detection means for the double-row needle bearing, is suitable for the problems of needle missing detection, size, accurate positioning and the like of other needle bearings, and has good universality. As a flexible detection means, the double-row needle bearing needle missing detection device and method based on machine vision can be very conveniently transplanted to other related application fields such as detection, identification and classification.

Drawings

FIG. 1 is a pictorial view of a double row needle bearing;

FIG. 2 is a schematic structural diagram of a double-row needle bearing needle missing visual detection device;

FIG. 3 is a flow chart of needle missing detection of the double-row needle bearing;

FIG. 4 is an original image acquired by the double row needle bearing;

FIG. 5 is a needle bearing ring and outer and inner circle extraction;

FIG. 6 is a needle bearing outer sub-area detection;

FIG. 7 is a partial needle offset into an inner independent sub-area;

FIG. 8 is a needle bearing outer sub-area missing the needle;

FIG. 9 is outer and inner subregion extraction and missing needle detection.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.

As shown in fig. 2, the key components of the double-row needle bearing needle missing visual detection device adopted by the invention comprise a camera, a light source and a detection piece. The detection idea is to utilize a monocular camera to combine with an active light source (annular light source) to carry out image acquisition and pretreatment on the bearing with a specific pose so as to judge whether the rolling needle is lack of needle. As a new detection method, the main steps are as shown in FIG. 3, including the division of the inside and outside of the bearing, the division and detection of the independent needle roller sub-area outside the bearing, the division and detection of the needle roller sub-area inside the bearing, and the judgment of the needle missing detection.

The division of the inside and outside of the bearing is based on the outer and inner circles of the ring of the needle bearing shown in fig. 5 as the boundary between the outside and the inside of the bearing. Since the needles are not free of play in the fixed position of the bearing, the play is sufficient to allow the needles to move. Thus, as can be seen from fig. 4, the individual needles may be displaced due to the play of the needle mounting, which also may be an undefined condition in which the same needle belongs to either the outer or inner sub-area. In general, the situation that the needle roller belongs to the outer subarea is more common under natural conditions, and the situation that the needle roller belongs to the inner subarea occasionally occurs. The number of needles of the same needle bearing is generally fixed. Therefore, whether the needle is missing can be preliminarily judged by judging the number of the independent outer subregions. However, even if the number of independent outer subareas is less than the number of needles in a full needle, the needle missing cannot be said to be true, and the reason for this may be that the needles are shifted to become inner subareas. In addition, since the needle roller bearings are double row, it can be seen from fig. 4 that the inner sub-regions of the needle roller bearings can be divided into two types. One is an inner subarea formed by an upper row of needle rollers, and the subareas and the inner wall of the needle roller bearing have intersection of pixel points (subareas in small rectangles in the figure); the other is an inner subarea formed by a lower row of needle rollers, and the subareas and the inner wall of the needle roller bearing do not have intersection of pixel points (subareas in small triangles in the drawing). Therefore, it is necessary to distinguish between the first case and the second case when determining the inner subregion. Further, due to the pose imaging relationship between the camera and the bearing, the one row of roller pins close to the camera lens can block the other row of roller pins far away from the camera lens. This results in a possible missed inspection problem even if the needles in the lower row are missing due to the obstruction of the needles in the upper row. Thus, there are two primary solutions: 1, a needle bearing does not move, and a camera is mirrored to detect a row of needles below; and 2, the needle bearing is turned over by 180 degrees, so that the needle rollers at the lower row and the needle rollers at the upper row are interchanged. In comparison, the operability of the 2 nd method is higher. The upset can adopt simple and easy manipulator commonly used to go on, and the work flow of repetition first again after the upset carries out the judgement that lacks the needle, can accomplish the needle that lacks of double kingpin and detect.

The method specifically comprises the following steps:

step 1:

initialization is mainly parameter initialization of various hardware and software systems, so that the detection system is in a normal working state and starts to work. The apparatus of operation comprises: camera, light source, bearing transmission device, image collection and image processing system. The camera is installed as far as possible to ensure that the optical axis of the camera and the rotating shaft of the bearing are in the coaxial direction, so that all needle rollers in a row closest to the camera lens on the bearing can be imaged by a single acquired image. Only if the row of needle rollers arranged on the bearing can be imaged on an image, the real needle missing or misjudgment of the bearing can be accurately judged. Here, a counter is mentioned in particular, which is used to determine the detection of which needle row is assigned to the bearing. When a certain needle bearing is detected, when count =0, the detection of needle missing of a row of needle rollers close to a camera lens is described; when count =1, the detection of missing needle is described when the row of needle rollers far from the camera lens is changed to the row of needle rollers near the camera lens through the overturning of the bearing. The numerical description that the overturning detection is needed when the monocular camera is used for detecting the double-row needle roller bearing is also provided.

Step 2:

and collecting an image of the needle bearing. In this step, the monocular camera captures an image of the needle bearing under the action of the light source. The light source is added mainly to ensure the imaging quality and avoid introducing excessive noise under natural light conditions so as to utilize subsequent image preprocessing. An example of the image is shown in fig. 4, from which it can be seen that the needle portion is easily distinguished from the background and other portions of the bearing due to the higher brightness. However, since the needle roller is partially shielded by the bearing mount, its bright portion is imaged as a "partial circle" area. Preliminarily, each "partial circle" area substantially corresponds to 1 needle roller, and the number of needle rollers can be determined according to the number of the independent areas. The image acquisition speed is determined according to the transmission speed of the bearings, and the image acquisition can be timely and accurately acquired as long as each bearing is guaranteed to be conveyed to a specified position. The acquired images ensure high imaging quality, simple background and comprehensive foreground as far as possible.

And step 3:

and (5) performing image threshold segmentation. The step is mainly to carry out image preprocessing based on the image acquired in the previous step so as to obtain the needle roller information. The threshold segmentation is to segment the foreground and the background by using appropriate gray values to obtain the region of interest. The threshold value can be determined experimentally by dividing the region of interest as far as possible from the region of non-interest. After the needle rollers are imaged, the brightness value is high, namely the gray value is large, and the needle rollers of the needle roller bearing after threshold segmentation are extracted as independent sub-areas. Meanwhile, as can be seen from fig. 4, the ring region of the bearing is also obtained by threshold segmentation, the outer and inner boundaries of the ring are obtained by edge extraction, and fitting of the inner circle and the outer circle is further performed by hough transform. Through the processing, the circular ring area, the inner circle and the outer circle outline of the bearing are obtained by using the methods of area extraction, boundary extraction and circle fitting.

And 4, step 4:

this step is a main step of determining the number of needles. Because a certain gap exists in the installation position of the needle roller, the needle roller can move, so that the needle roller imaging sub-region can possibly appear in a region outside the outer circle (a white small rectangular region as shown in fig. 6) or a region inside the inner circle (a sub-region inside a white rectangular frame as shown in fig. 7), and the difficulty in judging whether the needle of the bearing is missing is increased. The invention divides the conditions of needle missing detection into 2 types: 1, judging the number of the needle roller independent subregions in the region except the excircle; and 2, judging by combining the number of the needle roller independent sub-regions with pixel intersection between the inner circle region and the inner circle on the basis of the judgment of the step 1.

Step 4.1:

on the basis of the above steps, the outer circle of the bearing ring area and the outer independent subareas of the needles can be determined. To determine if a needle is missing, the number of outer independent sub-areas of needle needles is first calculated, assuming N1, and then compared to the number of needle needles N0. If N1= N0, the detected number of the needle rollers is proved to be equal to the actual number of the needle rollers, and at this time, the phenomenon that the needle of the current row of the bearing is not missed can be judged. Further, it is possible to determine whether the detection of the single needle roller or the detection of the double needle roller is completed by determining the value of the counter count. If the count =0, it is described that the current detection is the first detection of the double-row needle rollers of the bearing, and no needle is missing, and whether the needle is missing at the end can be determined only by performing the second detection; if count =1, the case where the current detection is the second detection of the double-row needle rollers of the bearing and there is no needle missing will be described, and it can be seen from the judgment that there is no needle missing in the above 2 detections, that the bearing does not have needle missing. At this point, the bearing may be transferred to a next process for subsequent assembly or other work. And meanwhile, returning to the initialized state, resetting the counter, and re-entering the needle missing detection judgment of the next bearing.

Step 4.2:

on the basis of the steps, the number N2 of the inner circles of the bearing circular ring areas and the needle roller inner independent sub-areas with pixel intersection can be determined. In order to judge whether the needle is lack, the number N1 of the independent sub-areas outside the needle roller is calculated, and then the sizes of the N1+ N2 and the number N0 of the needle roller are compared. If N1+ N2= N0, it is proved that the detected number of the needles is equal to the actual number of the needles, and it can be determined that the needle missing phenomenon does not occur in the current row of the bearing. Further, whether the detection of the single-row needle roller or the detection of the double-row needle roller is completed can be determined by judging the value of the counter count. If count =0, it indicates that the current detection is the first detection of the double rows of needle rollers of the bearing, and no needle is missing, and it is only necessary to perform the second detection to determine whether the needle is missing at all; if count =1, the case where the current detection is the second detection of the double-row needle rollers of the bearing and there is no needle missing will be described, and it can be seen from the judgment that there is no needle missing in the above-described second detection that the bearing is indeed needle missing. At this point, the bearing may be transferred to a next process for subsequent assembly or other work of the bearing. And the running program returns to the initialized state, the counter count is reset, and the next missing needle detection judgment of the bearing is carried out again.

And 5:

in the above step, if the number N1 of the outer independent sub-regions determined by the outer circle boundary is not equal to the number N0 of the needle rollers (the independent sub-regions cannot be detected outside the needle rollers in the rectangular frame shown in fig. 8), and the number N2 of the intersections of the pixels between the inner independent sub-regions and the inner circle is not equal to the number N0 of the needle rollers (the independent sub-regions cannot be detected inside and outside the needle rollers in the rectangular frame shown in fig. 9), it indicates that the bearing has a needle missing condition, that is, there is a problem that the needle roller is not installed properly. At the moment, the needle-missing bearing is removed and the needle roller is reinstalled, so that the needle-missing position can be covered by the needle roller. Meanwhile, the running program returns to the initialized state, the counter count is reset, and the next needle missing detection of the bearing is carried out again.

The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention, and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (5)

1. A double-row needle bearing needle missing detection method based on a monocular is characterized by comprising the following steps:

step S1: initializing, setting the numerical value of a counter count, setting the numerical value of the counter count to be M1 when a row of needle rollers close to the monocular camera lens are subjected to needle missing detection, and setting the numerical value of the counter count to be M2 when a row of needle rollers far away from the monocular camera lens are converted into a row of needle rollers close to the camera lens are subjected to needle missing detection;

step S2: acquiring an image of the needle bearing by adopting a monocular camera;

and step S3: extracting independent sub-regions of the rolling needles through threshold segmentation; meanwhile, a circular ring area of the bearing is obtained through threshold segmentation, the outer boundary and the inner boundary of the circular ring are obtained through edge extraction, and the inner circle and the outer circle are fitted through Hough transform to obtain the contours of the inner circle and the outer circle;

and step S4: comparing the number N1 of the needle roller independent sub-areas in the area outside the excircle with the number N0 of the needle rollers;

if N1= N0, performing step S5; if N1 is not equal to N0, comparing the sum of the number N2 and N1 of the needle rollers with pixel intersection between the internal independent sub-region and the internal circle with the size of N0, if N1+ N2= N0, executing the step S5, if N1+ N2 is not equal to N0, judging that the bearing is lack of the needle, eliminating the bearing and reassembling the needle rollers, and executing the steps S1-S4 again;

step S5: judging the value of the counter count;

if count = M1, the current detection is the first detection of the double rows of needle rollers of the bearing, and no needle is missing, the value of the counter is set to M2, and the steps S2-S5 are executed again; if count = M2, the current detection is the second detection of the double rows of the rolling needles of the bearing, and the bearing is transmitted to the next procedure without the condition of needle missing; the bearing double-row roller pins are the roller pins which are close to the lens of the monocular camera when the value of the counter is set to be M1 in the step 1.

2. The method for detecting needle missing in double-row needle bearings based on the monocular camera as claimed in claim 1, wherein in the step S2, the monocular camera acquires the image of the needle bearing under the action of the active light source.

3. The monocular based double row needle bearing needle miss detection method of claim 2, wherein the active light source is an annular light source.

4. The monocular based double row needle bearing needle missing detection method as recited in claim 1, wherein the value of M1 is 0 and the value of M2 is 1.

5. The monocular based double row needle bearing needle missing detection method of claim 1, wherein the method of changing the row of needle rollers farther from the monocular camera lens to the row of needle rollers closer to the camera lens comprises:

the needle bearing is stationary, and the camera is mirrored to detect a lower row of needles, or;

and the needle bearing is turned over by 180 degrees, so that the positions of the lower row of needle rollers and the upper row of needle rollers are interchanged.

Images