CN111814709A - Test method for bolt tightening on aircraft surface - Google Patents

Abstract

The invention discloses a method for detecting the tightening of bolts on the surface of an aircraft, and relates to the field of detecting the installation state of aircraft parts; in order to improve the detection efficiency and reliability, and at the same time, it is convenient for later inspection and data traceability; specifically, the system takes the bolt group as a whole to establish its local characteristic samples, First, the circular contour of each bolt in the bolt group is extracted, and a feature sample is constructed based on the relative positional relationship between the center points of the circular contour; when the system is running, the system matches the sample features within the range of the captured image, and after successful matching, intercepts Its local area is saved for bolt tightening state detection, which specifically includes the following steps: system initialization, including feature recognition preprocessing on sample images. The invention adopts feature detection, extracts the bolt area to be detected, extracts the local area of the sub-bolt to be detected by detecting circular feature points, and identifies the bolt tightening state by comparing the pixel average value in four quadrants, which greatly reduces the detection burden of workers and improves the detection performance. precision.

Description

Test method for bolt tightening on aircraft surface

technical field

The invention relates to the technical field of the detection of the installation state of aircraft parts, in particular to a method for detecting the tightening of bolts on the surface of the aircraft.

Background technique

The quality control inspection before the aircraft is released is the last process for the pilots to confirm the appearance of the aircraft, the position of the cockpit switch, the amount of refueling, and the pressure indication value after the oxygen system is supplemented with oxygen. Due to the particularity, these inspections are still done manually at home and abroad; although there are corresponding sensors and detection equipment for the visual inspection of the appearance, visual inspection or hand-held related equipment is also required for inspection.

Therefore, the traditional inspection methods mainly have the following shortcomings:

(1) In the traditional inspectors using the lifting platform or climbing to the top of the aircraft, it takes several hours to conduct quality inspection, which is inefficient;

(2) Incomplete extraction of inspection methods, discrimination standards, inspection records and other elements for aircraft release inspection, lack of guidance for the operation information of each key node, and insufficient ability to supplement temporary information, resulting in unskilled or newcomers. operators with high operational risk;

(3) In the stage of pure manual recording by the operator, the phenomenon of data islanding occurs, which reduces the data traceability and brings a great lag to the later result inspection; it increases the operational risk caused by human instability;

(4) The long-term paper recording method has brought great trouble to the later verification, filing and recording work;

(5) With the development of technology and the improvement of indicators, the operation difficulty and requirements of maintenance, inspection and other links have also increased; this is for many enterprises in the fields of high precision, high accuracy and high intensity, To improve the knowledge penetration rate of personnel and expand the channels of learning and accumulation, it requires a lot of human and economic costs to maintain the necessary quality requirements.

SUMMARY OF THE INVENTION

The purpose of the present invention is to propose a method for detecting the tightening of bolts on the surface of an aircraft in order to solve the shortcomings existing in the prior art.

In order to achieve the above object, the present invention adopts the following technical solutions:

The bolt tightening detection method on the surface of the aircraft uses the system to establish the local feature sample of the bolt group as a whole. First, the circular contour of each bolt in the bolt group is extracted, and the feature sample is constructed based on the relative positional relationship between the center points of the circular contour; the system runs , the system matches the sample features within the range of the captured image, and after successful matching, intercepts its local area and saves it for bolt tightening state detection, which includes the following steps:

S1: System initialization, including the feature recognition preprocessing of the sample image, as a feature sample for detecting the position of the bolt group, and the initialization of each detection parameter;

S2: real-time image acquisition, using CCD to collect images, the image resolution is 1920X1080;

S3: Filter, use median filter to filter to clutter;

S4: binarize the image;

S5: extract contour;

S6: Identify the bolt group position;

S7: Calculate the center point of the bolt group; extract the center of the bolt;

S8: Calculate the position of each sub-bolt of the bolt group;

S9: intercept the local area of the bolt to be detected;

S10: Calculate whether the tightening state complies with the average value of the quadrants and calculate the average value of each quadrant; if the average value of the 1 and 3 quadrants is less than the average value of the 2 and 4 quadrants, it is judged to be tightened, otherwise, it is not tightened;

S11: Repeat steps S8 to S10 until all bolts are detected;

S12: The detection is completed, and the result is output.

Preferably, the specific execution conditions of S9 to S12 are: each bolt group includes 12 bolts evenly distributed around the center point, and for the detection of the tightening state, the tightening state of each bolt is detected in turn, and each bolt is first intercepted. The local area is divided into four quadrants. If the average value of the four-quadrant pixels in the second and fourth quadrants is greater than the first and second quadrants, the tightening state is qualified, otherwise it is unqualified; until all bolts are detected.

Preferably: in the step S4, image binarization processing: for the collected grayscale image, set the threshold T=200, p is the pixel value of the corresponding point, the formula:

。

.

Preferably: in the step S5, the contour extraction is specifically the soble operator to extract the edge; the Sobel operator is mainly used for edge detection, including horizontal and vertical templates; the template is convolved with the image to obtain a gradient approximation; Adapt the algorithm to enhance the anti-noise ability of the image.

Preferably: in the step S5, the specific calculation method is:

Horizontal Landscape Template:

；

;

Horizontal portrait template:

；

;

The size of the gradient for each pixel of the image:

The magnitude of the gradient direction can then be estimated by the following formula:

Using G and P as boundary selection conditions, set P to 1 and G to 1.

Preferably: in the step S6, a feature detection and matching method is used for the positioning bolt tightening detection part, and the method matches the corresponding position of the bolt to be detected by establishing the local characteristic value of the bolt; this method is used to solve the rotation, scaling, translation (RST), Image affine/projective transformations, lighting effects, object occlusion, cluttered scenes, noise.

Preferably: in the step S7, the coordinates of the center point of the bolt group are:

。

.

Preferably: the system operation steps:

S01: Power on the system, wait for the initialization to complete, and the interface shows that the self-check is successful;

S02: CCD collects real-time images, and the system feeds back the results after calculation;

S03: When the result fed back by S02 is qualified, the system is running normally and the inspection is performed.

The beneficial effects of the present invention are:

1. The present invention uses feature detection to extract the area of the bolt to be detected. For each sub-bolt in the area, the local area of the sub-bolt to be detected is extracted by detecting circular feature points, and the bolt tightening state is identified by comparing the average value of pixels in four quadrants, which greatly reduces the problem. The detection burden of workers is reduced, and the detection accuracy is improved.

2. The present invention eliminates the stage of pure manual recording by the operator, avoids the phenomenon of data islands, has strong data traceability, and avoids the hysteresis effect of manual recording on later result inspection; eliminates subjective factors, reliability and stability. powerful.

3. The present invention extracts the edge by using the soble operator; convolves the template with the image to obtain the approximate gradient value; and uses the adaptive algorithm of the weight, which greatly enhances the anti-noise ability of the image.

4. The positioning bolt tightening detection part of the present invention adopts the feature detection and matching method, which matches the corresponding position of the bolt to be detected by establishing the local characteristic value of the bolt; this method can effectively solve the target rotation, scaling, translation (RST), image simulation. Problems such as beam/projection transformation, lighting effects, target occlusion, clutter scenes, noise, etc.

Description of drawings

Fig. 1 is the flow chart of the method for detecting bolt tightening on the surface of the aircraft proposed by the present invention;

2 is a schematic diagram of an image after binarization in the method for detecting bolt tightening on an aircraft surface proposed by the present invention;

3 is a schematic diagram of an operator extracting a contour image in a method for detecting bolt tightening on an aircraft surface proposed by the present invention;

Fig. 4 is the schematic diagram of all bolt center point calculation in the aircraft surface bolt tightening detection method proposed by the present invention;

5 is a schematic diagram of the Hough transform detection circle identifying the image of the local area of the bolt in the method for detecting bolt tightening on the surface of the aircraft proposed by the present invention;

6 is a schematic diagram of four-quadrant segmentation of a local area of a bolt in a method for detecting bolt tightening on an aircraft surface proposed by the present invention;

7 is a schematic diagram of the overall structure of the bolt group of the method for detecting the tightening of bolts on the surface of the aircraft proposed by the present invention;

FIG. 8 is a flow chart of feature matching of a method for detecting bolt tightening on an aircraft surface proposed by the present invention.

In the picture: 1-camera, 2-bolt to be detected.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments.

As shown in Figure 1-8, the bolt tightening detection method on the surface of the aircraft is used to establish its local feature sample by using the bolt group as a whole. The position relationship constructs a feature sample; when the system is running, the system matches the sample feature within the range of the captured image. After the matching is successful, it intercepts its local area and saves it for bolt tightening state detection. Specifically, it includes the following steps:

S1: System initialization, including the feature recognition preprocessing of the sample image, as a feature sample for detecting the position of the bolt group, and the initialization of each detection parameter;

S2: real-time image acquisition, using CCD to collect images, the image resolution is 1920X1080;

S3: Filter, use median filter to filter to clutter;

S4: binarize the image;

S5: extract contour;

S6: Identify the bolt group position;

S7: Calculate the center point of the bolt group; extract the center of the bolt;

S8: Calculate the position of each sub-bolt of the bolt group;

S9: intercept the local area of the bolt to be detected;

S10: Calculate whether the tightening state complies with the average value of the quadrants and calculate the average value of each quadrant; if the average value of the 1 and 3 quadrants is less than the average value of the 2 and 4 quadrants, it is judged to be tightened, otherwise, it is not tightened;

S11: Repeat steps S8 to S10 until all bolts are detected;

S12: The detection is completed, and the result is output.

As shown in FIG. 6 and FIG. 7 , the specific execution conditions of S9 to S12 are: each bolt group includes 12 bolts evenly distributed around the center point, and for the detection of the tightening state, the tightening state of each bolt is detected in turn , First intercept the local area of each bolt and divide it into four quadrants. If the average value of four quadrant pixels in two and four quadrants is greater than one or two quadrants, the tightening state is qualified, otherwise it is unqualified; until all bolts are tested.

As shown in Figure 2, in the step S4, image binarization processing: for the collected grayscale image, set the threshold T=200, p is the pixel value of the corresponding point, the formula:

。

.

As shown in Figure 3, in the step S5, the extraction contour is specifically the soble operator to extract the edge; the Sobel operator is mainly used for edge detection, including horizontal and vertical templates; the template is convolved with the image to obtain an approximate gradient value; The adaptive algorithm of the value is used to enhance the anti-noise ability of the image.

As shown in Figure 3, in the step S5, the specific calculation method is:

Horizontal Landscape Template:

；

;

Horizontal portrait template:

；

;

The size of the gradient for each pixel of the image:

The magnitude of the gradient direction can then be estimated by the following formula:

With G and P as boundary selection conditions, P is set to 1, and G is preferably 1.

As shown in Figure 8, in the step S6, the feature detection and matching method is adopted for the positioning bolt tightening detection part. This method matches the corresponding position of the bolt to be detected by establishing the local feature value of the bolt; this method is used to solve the rotation, scaling and translation of the target ( RST), image affine/projective transformation, lighting effects, object occlusion, clutter scenes, noise.

As shown in Figure 4, in the step S7, the coordinates of the center point of the bolt group are:

The system operation steps:

S01: Power on the system, wait for the initialization to complete, and the interface shows that the self-check is successful;

S02: CCD collects real-time images, and the system feeds back the results after calculation;

S03: When the result fed back by S02 is qualified, the system is running normally and the inspection is performed.

The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited to this. The equivalent replacement or change of the inventive concept thereof shall be included within the protection scope of the present invention.

Claims (8)

1. The method for detecting the bolt tightening on the surface of the airplane is characterized in that a local characteristic sample of the airplane is established by taking a bolt group as a whole through a system, firstly, the circular contour of each bolt in the bolt group is extracted, and the characteristic sample is established according to the relative position relation between the central points of the circular contours; when the system runs, the system matches the sample characteristics in the shot image range, and after the matching is successful, the local area of the sample characteristics is intercepted and stored for bolt tightening state detection, and the method specifically comprises the following steps:

s1: the system initialization comprises the steps of carrying out feature identification preprocessing on a sample image, taking the sample image as a feature sample for detecting the positions of the bolt groups, and initializing all detection parameters;

s2: collecting images in real time, and collecting the images by adopting a CCD (charge coupled device), wherein the resolution of the images is 1920X 1080;

s3: filtering, namely filtering clutter by adopting median filtering;

s4: carrying out binarization processing on the image;

s5: extracting a contour;

s6: identifying bolt group positions;

s7: calculating the center point of the bolt group; extracting the center of the bolt;

s8: calculating the position of each sub-bolt of the bolt group;

s9: intercepting a local area of a bolt to be detected;

s10: calculating whether the tightening state accords with the quadrant average value and calculating the average value of each quadrant; if the average values of the quadrants 1 and 3 are smaller than the average value of the quadrants 2 and 4, the screwing is judged, otherwise, the screwing is not carried out;

s11: repeating the steps S8 to S10 until all the bolts are detected;

s12: and (5) after the detection is finished, outputting a result.

2. The method for detecting the tightening of the bolt on the surface of the aircraft according to claim 1, wherein the specific execution conditions of S9 to S12 are as follows: each bolt group comprises 12 bolts which are uniformly distributed around a central point, the tightening state of each bolt is detected in sequence for the tightening state detection, firstly, the local area of each bolt is intercepted and divided into four quadrants, and if the average value of four quadrants is two quadrants and the four quadrants are two quadrants, the tightening state is qualified, otherwise, the tightening state is unqualified; until all bolt detections are completed.

3. The aircraft surface bolt tightening detection method according to claim 2, wherein in the step S4, the image binarization process: for the acquired gray-scale image, setting a threshold value T =200, wherein p is a pixel value of a corresponding point, and the formula is as follows:

。

4. the aircraft surface bolt tightening detection method according to claim 3, wherein in the step S5, the extraction contour is specifically a cable operator extraction edge; the Sobel operator is mainly used for edge detection and comprises a horizontal template and a vertical template; convolving the template with the image to obtain a gradient approximation; and adopting a weight adaptive algorithm to enhance the anti-noise capability of the image.

5. The aircraft surface bolt tightening detection method according to claim 4, wherein in the step S5, the specific calculation method is as follows:

horizontal transverse template:

；

horizontal longitudinal formwork:

；

magnitude of gradient of each pixel of image:

the gradient direction magnitude can then be estimated using the following equation:

and setting P as 1 and G as 1 by taking G and P as boundary selection conditions.

6. The aircraft surface bolt tightening detection method according to claim 5, wherein in the step S6, the bolt tightening detection part is positioned by adopting a characteristic detection matching method, and the method matches the corresponding position of the bolt to be detected by establishing a local characteristic value of the bolt; the method is adopted to solve the problems of rotation, scaling, translation (RST), image affine/projection transformation, illumination influence, target shielding, sundry scene and noise of the target.

7. The aircraft surface bolt tightening detection method according to claim 6, wherein in the step S7, the coordinates of the bolt group center point are as follows:

。

8. the aircraft surface bolt tightening detection method of claim 7, wherein the system operating steps:

s01: the system is powered on, the interface displays that the self-checking is successful after the initialization is completed;

s02: the CCD collects real-time images, and the system feeds back the result after calculating;

s03: when the result fed back from S02 is acceptable, the system operates normally, and the detection is performed.

Images