Connecting rod character and oil groove coplanar visual detection device

Abstract

The invention discloses a connecting rod character and oil groove coplanar visual detection device, which relates to the technical field of workpiece defect detection and comprises a light source unit arranged above a workpiece detection path and an image acquisition unit arranged on the light source unit. Light is projected through the light source unit, an optical contrast signal of a bright / dark reflection area is generated on the surface of a workpiece, and characters or oil groove edge features can be clearly distinguished through images collected by the image collection unit. And then the processing unit executes cross-azimuth correlation analysis in a single to-be-detected image through a pre-stored standard template, and judges whether dark reflection region characteristics matched with the two-azimuth reference template exist or not, so as to quickly judge whether a coplanar defect exists in the workpiece or not. All the units of the whole device cooperate with one another, complex manual operation is not needed, the detection efficiency and accuracy are greatly improved, subjectivity and errors of manual detection are effectively avoided, the device is suitable for industrial batch detection scenes, and stable and reliable workpiece quality is ensured.

Description

Technical Field

[0001] This invention relates to the field of workpiece defect detection technology, and in particular to a visual inspection device for coplanar detection of connecting rod characters and oil grooves. Background Technology

[0002] Connecting rods are crucial engine components, and their specifications and shapes adhere to strict standards. Some connecting rod components require characters and numbers on both sides of the front and oil grooves on the back. During the connecting rod molding process, due to operational errors, oil grooves are sometimes forged and machined on the front of the connecting rod, resulting in incorrect connecting rod components where the characters and oil grooves are coplanar.

[0003] Currently, manual visual inspection is used to check for coplanar defects. However, in continuous inspection scenarios, workers are prone to errors in logical recognition due to visual fatigue. Although some companies have tried to use visual inspection to detect surface defects or unique structures of connecting rods, such as grooves, these are mostly targeted at single features, such as only detecting oil grooves or only detecting characters.

[0004] Existing technologies lack the ability to determine the correlation between different machined surface features of a workpiece, such as the spatial relationship between the surface containing the connecting rod characters and the actual machined surface of the oil groove. Therefore, existing visual inspection solutions cannot identify coplanar defects in connecting rods. Summary of the Invention

[0005] The purpose of this invention is to solve the problem that existing technologies only detect single features of connecting rods and lack the ability to determine the correlation between features of different machined surfaces of the workpiece. When the correlation between the spatial position of the character on the surface of the connecting rod workpiece and the actual machined surface of the oil groove cannot be correctly determined, the coplanar defects of the connecting rod workpiece cannot be identified.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a visual inspection device for coplanar detection of connecting rod characters and oil grooves, comprising a light source unit disposed above the workpiece inspection path and an image acquisition unit disposed on the light source unit.

[0007] The light source unit projects light onto the surface of the workpiece, causing the flat area to form a bright reflection area and the feature edge to form a dark reflection area, thereby generating a feature-distinguishable optical signal. The image acquisition unit then acquires images of the first and second orientation surfaces of the workpiece based on this optical signal in a time-division manner.

[0008] The processing unit, which is communicatively connected to the image acquisition unit, pre-stores a first orientation feature reference template and a second orientation feature reference template of the standard workpiece.

[0009] The processing unit receives the image collected by the image collection unit, and performs cross-orientation correlation analysis on the image to detect whether the dark reflection area features matching the first orientation reference feature template and the second orientation reference feature template exist in the single image simultaneously, and if yes, it is determined that the workpiece under test has a coplanar defect, and if not, it is determined to be qualified.

[0010] In the above technical solution, in use, the light source unit projects light onto the surface of the connecting rod workpiece, according to the Fresnel reflection principle, the smooth surface of the workpiece flat area (no characters / oil grooves) forms a mirror reflection, the light returns along the original light path, and in the image collected by the camera, a bright reflection area is formed; and the feature edges (character protrusions or oil groove depressions) cause the light to be diffused or scattered due to geometric mutations, only a small amount of light enters the image collected by the camera, forming a dark reflection area in the image. This process converts physical features into identifiable optical contrast signals.

[0011] Then, based on the optical signals, the image collection unit of the camera collects images of the two orientation surfaces of the workpiece at different times, that is, the connecting rod workpiece image containing the dark reflection area of the characters is shot, and the connecting rod workpiece image containing the dark reflection area of the oil groove is shot.

[0012] Then, the processing unit pre-stores the first and second orientation reference feature templates (such as character edge templates and oil groove edge templates), and receives the images, and analyzes the connecting rod workpiece image containing the dark reflection area of the characters and / or the connecting rod workpiece image containing the dark reflection area of the oil groove.

[0013] If only the dark reflection area (characters) matching the first orientation reference feature template exists in the image, and no dark reflection area of the second orientation reference feature template (oil groove) exists, it is determined to be qualified (the characters and the oil groove belong to the two surfaces of the connecting rod workpiece).

[0014] If the dark reflection areas matching the first and second orientation reference feature templates exist in the image (such as the characters and the oil groove being on the same surface of the connecting rod workpiece), it is determined to have a coplanar defect.

[0015] The beneficial effects of the present application are:

[0016] By projecting light through the light source unit, the optical contrast signal of the bright / dark reflection area is generated on the surface of the workpiece, so that the image collected by the image collection unit can clearly distinguish the character or oil groove edge features. Then, the processing unit performs cross-orientation correlation analysis in the single image to be detected by using the pre-stored standard template, to determine whether the dark reflection area features matching the double orientation reference templates exist, so as to quickly determine whether the workpiece has a coplanar defect. The units of the whole device cooperate with each other, without complex manual operation, greatly improving the detection efficiency and accuracy, effectively avoiding the subjectivity and errors of manual detection, and being suitable for industrial batch detection scenes to ensure stable and reliable workpiece quality.

[0017] Further, in the embodiment of the present application, the optical axis of the light source unit is perpendicular to the surface of the workpiece, so that the light on the flat area of the workpiece is totally reflected to form a bright reflection area, and the light on the feature edge of the workpiece is obliquely reflected to form a dark reflection area.

[0018] Further, in the embodiment of the present application, the light source unit is integrated with alternatingly arranged LED visible light, ultraviolet light source and infrared light source, and the optical axes of the three groups of light sources converge at the same detection point through the reflecting prism to form coaxial multi-spectral illumination on the measured workpiece.

[0019] Further, in the embodiment of the present application, the processing unit pre-stores the first orientation feature reference template and the second orientation feature reference template, which contain the spectral feature library of the character edge, oil groove edge, oil stain and oxide layer of the standard workpiece, for excluding the interfering dark reflection area features.

[0020] Further, in the embodiment of the present application, a diaphragm is arranged in the converging light path of the light source unit, and the diaphragm is uniformly provided with light transmission lenses corresponding to the wavelengths of visible light, ultraviolet light and infrared light, and the diaphragm is driven to rotate by a first servo motor to switch the illumination of different light sources.

[0021] Further, in the embodiment of the present application, a light filtering module is arranged in front of the lens of the image acquisition unit, and the light filtering channels of the light filtering module are provided with light filters allowing the corresponding light source wavelengths to pass through, and the light filtering module is driven to rotate by a second servo motor to synchronously switch the light filtering channels with the diaphragm.

[0022] Further, in the embodiment of the present application, the light source unit further comprises a light source controller connected with the first servo motor, the second servo motor and the light source unit.

[0023] The light source controller is a PLC controller or a multi-channel light source driver based on an ARM chip, and the light source controller activates the LED visible light, ultraviolet light source and infrared light source in a preset time sequence.

[0024] Further, in the embodiment of the present application, the diaphragm and the light filtering module are both provided with driven gears, the drive gear of the first servo motor is engaged with the driven gear of the diaphragm to drive the diaphragm to rotate and switch the light transmission lenses.

[0025] The drive gear of the second servo motor is engaged with the driven gear of the light filtering module to drive the light filtering module to rotate and switch the light filtering channels.

[0026] Further, in the embodiment of the present application, the images of the first orientation surface and the second orientation surface of the workpiece at least include visible light images, ultraviolet images and infrared images.

[0027] The processing unit performs spatio-temporal registration on the images to correct the connecting rod motion offset before performing cross-orientation correlation analysis on the images, and then extracts the contour and texture features of the visible light image, the oxidation layer distribution features of the ultraviolet image, and the depth mutation region features of the infrared image, and integrates the three types of features into a feature vector, and uses the processing unit of the support vector machine algorithm to perform fusion analysis to distinguish the real oil groove / character edge and the interference dark area.

[0028] Further, in the embodiment of the present application, the processing unit is configured to calculate the defect probability value using the support vector machine algorithm, and determine whether the workpiece is qualified based on a preset probability threshold. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 It is a schematic diagram of the connecting rod character and oil groove coplanar visual detection device in the embodiment of the present application.

[0030] Figure 2 It is a schematic diagram of the first orientation reference feature template and the second orientation reference feature template in the embodiment of the present application.

[0031] Figure 3 It is a schematic diagram of the connecting rod character and oil groove coplanar visual detection device with an added diaphragm and filter module in the embodiment of the present application.

[0032] Figure 4 It is a cooperation schematic diagram of the diaphragm and filter module in the embodiment of the present application.

[0033] Figure 5 It is a thought guiding schematic diagram of the connecting rod character and oil groove coplanar visual detection in the embodiment of the present application.

[0034] 1, light source unit, 2, image acquisition unit, 3, measured workpiece, 4, reflecting prism;

[0035] 10, diaphragm, 11, light transmission lens, 12, first servo motor;

[0036] 20, filter module, 21, filter, 22, second servo motor. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical scheme of the present application clear, complete and the advantages more clear and obvious, the embodiments of the present application are further described in detail below in combination with the drawings. It should be understood that the specific embodiments described here are part of the embodiments of the present application, not all the embodiments, and are used to explain the embodiments of the present application, and do not limit the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0038] In the description of the present application, it should be noted that the terms "center", "middle", "upper", "lower", "left", "right", "inner", "outer", "top", "bottom", "side", "vertical", "horizontal" and the like indicate the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "one", "first", "second", "third", "fourth", "fifth", "sixth" are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0039] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium; it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0040] For the purpose of simplicity and illustration, the principles of the embodiments are mainly described by reference to examples. In the following description, many specific details are presented to provide a thorough understanding of the embodiments. However, it is obvious that these embodiments can not be limited to these specific details in practice. In some examples, the structures of well-known applications applied to the connecting rod character and oil groove coplanar visual detection device are not described in detail to avoid unnecessarily making these embodiments difficult to understand. In addition, all embodiments can be used in combination with each other.

[0041] Embodiment 1:

[0042] It should be noted that the drawings in the specification are part of the content of the specification, and the structure shape, connection relationship, fitting relationship, positional relationship in the drawings can be understood as the content of the specification.

[0043] A connecting rod character and oil groove coplanar visual detection device, as shown in Figure 1 The device includes a light source unit 1 arranged above the workpiece detection path and an image acquisition unit 2 arranged on the light source unit 1, and the image acquisition unit 2 is a camera.

[0044] The light source unit 1 projects light onto the workpiece surface, so that the flat area forms a bright reflection area and the feature edge forms a dark reflection area, thereby generating a feature distinguishable optical signal, and the image acquisition unit 2 acquires images of the first and second positional surfaces of the workpiece based on the optical signal.

[0045] The processing unit pre-stores the first orientation feature reference template and the second orientation feature reference template of the standard workpiece.

[0046] The processing unit should be understood as the image data processing chip of a computer.

[0047] The processing unit receives the image collected by the image collection unit 2, and performs cross-orientation correlation analysis on the image to detect whether the dark reflection area features matching the first orientation reference feature template and the second orientation reference feature template exist in the single image at the same time. If yes, it is determined that the measured workpiece 3 has a coplanar defect; otherwise, it is determined to be qualified. DETAILED DESCRIPTION

[0049] As shown in Figure 1 , Figure 2 , the light source unit 1 projects light onto the surface of the measured workpiece 3. According to the principle of Fresnel reflection, the smooth surface of the workpiece (no characters / oil grooves) forms a mirror reflection, and the light returns along the original light path, appearing as a bright reflection area in the image collected by the camera. The feature edges (character protrusions or oil groove depressions) cause the light to be diffused or scattered due to geometric mutations, and only a small amount of light enters the image collected by the camera, forming a dark reflection area in the image. This process converts physical features into identifiable optical contrast signals.

[0050] Next, the image collection unit 2 of the camera collects images of the two orientation surfaces of the workpiece based on the optical signals, i.e., it takes images of the dark reflection area of the connecting rod workpiece containing characters, and images of the dark reflection area of the connecting rod workpiece containing oil grooves.

[0051] Then, the processing unit pre-stores the first and second orientation reference feature templates (such as character edge templates and oil groove edge templates) receives the images and analyzes the connecting rod workpiece image containing the dark reflection area of the characters and / or the connecting rod workpiece image containing the dark reflection area of the oil grooves.

[0052] If only the dark reflection area (characters) matching the first orientation reference feature template exists in the image, and no dark reflection area of the second orientation reference feature template (oil grooves) exists, it is determined to be qualified (the characters and the oil grooves belong to the two surfaces of the connecting rod workpiece).

[0053] If the dark reflection areas matching the first and second orientation reference feature templates exist in the image at the same time (such as the characters and the oil grooves being on the same surface of the connecting rod workpiece), it is determined to have a coplanar defect.

[0054] It should be noted that the first orientation feature reference template is the a sample plate attached Figure 2 and the character features contained therein, and the second orientation feature reference template is the b sample plate attached Figure 2 and the oil groove features contained therein.

[0055] It should be noted that after the image of the one side of the connecting rod containing characters is shot, the connecting rod is turned over by manual or existing turning device to expose the one side containing oil groove, and the image acquisition unit 2 of the camera is generally used for shooting.

[0056] The advantage of the present application is that the light source unit 1 projects light to generate the optical contrast signal of bright / dark reflection area on the surface of the workpiece, so that the image acquired by the image acquisition unit 2 can clearly distinguish the characters or the edge features of the oil groove. Then the processing unit performs cross-orientation correlation analysis in the single image to be detected by using the pre-stored standard template, and judges whether there is dark reflection area feature matching the double-orientation reference template, so as to quickly determine whether the workpiece has coplanar defects. The units of the whole device cooperate with each other, without complex manual operation, which greatly improves the detection efficiency and accuracy, effectively avoids the subjectivity and error of manual detection, and is suitable for industrial batch detection scene to ensure stable and reliable product quality.

[0057] Specifically, the optical axis of the light source unit 1 is perpendicular to the surface of the workpiece, so that the light of the flat area of the workpiece is completely reflected to form a bright reflection area, and the light of the feature edge of the workpiece is obliquely reflected to form a dark reflection area. Its role is to strengthen the feature contrast through coaxial light illumination geometric relationship: when the light source is vertically incident, the flat area is mirror reflected (following the Fresnel reflection law) to make the reflected light completely enter the camera imaging, showing a high-brightness area; while the irregular surface such as characters / oil groove causes light scattering or deviation from the original light path due to geometric mutation, only a small part of the reflected light enters the camera, forming a dark area. This principle realizes high-contrast imaging of feature edges and backgrounds based on the difference of reflected light paths, so that the image acquisition unit 2 can clearly capture the position and shape of the dark reflection area, laying a foundation for subsequent template matching. The technical effect is to significantly improve the contour recognition of characters and oil grooves and reduce the risk of misidentification caused by uneven surface reflection.

[0058] Specifically, as shown in Figure 1 The light source unit 1 is integrated with alternating LED visible light, ultraviolet light source and infrared light source, and the optical axes of the three groups of light sources are converged on the same detection point through the reflecting prism 4 to form coaxial multi-spectral illumination of the workpiece 3 to be detected.

[0059] Because in actual production, there may be oxidation layer, oil stain and other interference on the surface of the connecting rod workpiece, the characters and oil grooves cannot be identified only by the reflection difference of the LED visible light, otherwise the reflection characteristics of the flat area and the uneven area of the connecting rod will be covered, especially when the oil groove depth is shallow or the character edge is worn, it is easy to appear "dark area misjudgment" (such as oil stain is mistaken as oil groove edge), which affects the accuracy of the coplanar defect identification of the connecting rod.

[0060] Therefore, the light source unit 1 is expanded to LED visible light, ultraviolet (UV), infrared (IR) multi-spectral coaxial irradiation, and the light is converged to the same detection point by the reflecting prism 4 for workpiece detection. The core function is to cope with complex working conditions (such as oxide layer coverage, oil stain interference): LED visible light reflects the surface texture, ultraviolet light excites the fluorescence characteristics of the metal oxide layer to highlight the processing stress area (forging oil groove edge), and infrared light penetrates the oil stain to directly image the metal matrix structure. The wavelength difference of the three groups of light sources (365nm UV, 850nm IR, 400-700nm visible light) makes the reflection / transmission characteristics complementary, realizing multi-dimensional feature acquisition.

[0061] More specifically, the first and second orientation feature reference templates pre-stored by the processing unit contain the spectral feature library of the character edge, oil groove edge, oil stain, and oxide layer of the standard workpiece, which is used to exclude the features of the interfering dark reflection area.

[0062] The spectral features of the oil stain and the oxide layer are included in the library of the first and second orientation feature reference templates, and the spectral response data of the pre-stored interference features (such as the low fluorescence intensity of the oil stain in the ultraviolet band and the absorption attenuation of the oxide layer in the infrared) are used to make the processing unit exclude the interfering dark area during cross-orientation analysis. The principle is to establish a spectral fingerprint library of the interference, and use a comparison mechanism (such as Euclidean distance or cosine similarity) to distinguish between real features (characters, oil grooves) and interference (oil stains, oxidation). The technical effect is to reduce the misjudgment rate, especially in the case of oil stain residue or uniform coverage of the oxide layer, the coplanar defects can still be accurately identified.

[0063] More specifically, as shown in Figure 3 , Figure 4 , a diaphragm 10 is provided in the converging light path of the light source unit 1, and the diaphragm 10 is uniformly provided with light transmission lenses 11 corresponding to the wavelengths of visible light, ultraviolet light and infrared light. The diaphragm 10 is driven to rotate by a first servo motor 12 to switch the irradiation of different light sources.

[0064] By switching the light transmission lenses 11 of different light sources, time-sharing multi-spectral irradiation is realized. The function is that the diaphragm 10 physically separates: the three groups of light transmission lenses 11 only allow the corresponding wavelengths to pass, such as the 365nm UV lens which transmits ultraviolet light and blocks other wavelengths, avoiding multi-source crosstalk.

[0065] More specifically, as shown in Figure 3 , Figure 4 , a filter module 20 is provided in front of the lens of the image acquisition unit 2, and the filter channels of the filter module 20 are provided with filter plates 21 allowing the corresponding light source wavelengths to pass. The filter module 20 is driven to rotate by a second servo motor 22 to synchronously switch the filter channels with the diaphragm 10.

[0066] The filter module 20 synchronously switched with the diaphragm 10 matches the filter 21 in front of the camera lens. For example, when detecting ultraviolet, the filter 21 blocks visible light and infrared light, so that the fluorescence signal (420-450 nm) of the oxidation layer is highlighted, so as to reduce image noise, enhance the intensity of the characteristic signal, and avoid the detection result drift caused by the fluctuation of the ambient light.

[0067] More specifically, as shown in Figure 5 The light source unit 1 further comprises a light source controller connected with the first servo motor 12, the second servo motor 22 and the light source unit 1 respectively.

[0068] The light source controller is a PLC controller or a multi-channel light source driver based on an ARM chip (such as a control board based on an ARM chip). The light source controller activates the LED visible light, ultraviolet light source and infrared light source in a preset time sequence.

[0069] The preset time sequence is to trigger and start the LED visible light, ultraviolet light source and infrared light source at an interval of 1 ms or 2 ms, and trigger and start the first servo motor 12 and the second servo motor 22 at the same time, so as to ensure strict synchronization.

[0070] More specifically, the diaphragm 10 and the filter module 20 are provided with driven teeth. The drive gear of the first servo motor 12 is engaged with the driven teeth of the diaphragm 10 to drive the diaphragm 10 to rotate and switch the light transmission lens 11.

[0071] The drive gear of the second servo motor 22 is engaged with the driven teeth of the filter module 20 to drive the filter module 20 to rotate and switch the light transmission channel.

[0072] More specifically, the images of the first and second orientation surfaces of the workpiece at least include visible light images, ultraviolet images and infrared images.

[0073] Before the processing unit performs cross-orientation correlation analysis on the images, the images are first subjected to space-time registration to correct the motion offset of the connecting rod, and then the contour and texture features of the visible light image, the oxidation layer distribution features of the ultraviolet image and the depth mutation area features of the infrared image are extracted respectively. The three types of features are integrated into a feature vector, and a support vector machine algorithm processing unit is used for fusion analysis to distinguish the real oil groove / character edge from the interference dark area.

[0074] More specifically, the processing unit is configured to calculate a defect probability value using a support vector machine algorithm, and determine whether the workpiece is qualified based on a preset probability threshold.

[0075] Specifically, as shown in Figures 1-5As shown, the final workflow of the embodiment is: when the measured workpiece 3, i.e. the connecting rod, enters the detection area, the visible light, ultraviolet light source and infrared light source are activated in turn by the light source controller according to the preset timing sequence to perform multispectral irradiation. During this process, the rotating diaphragm 10 is synchronously switched to the light transmission hole corresponding to the light source, and the camera is synchronously used to collect images by matching the corresponding filter 21. The collected images include the images of the first and second orientation surfaces of the workpiece. The first and second orientation surface images respectively include visible light images, ultraviolet images and infrared images (three images of the front surface of the connecting rod workpiece with characters, and three images of the back surface with oil grooves): the visible light image records the surface texture of the connecting rod, and presents the basic outline of the character edge and the oil groove; the ultraviolet light excites the metal oxide layer to produce fluorescence, and the ultraviolet image highlights the processing stress area such as the edge of the forging oil groove; the infrared light penetrates the oil stain covering layer, and the infrared image reflects the bottom structure of the metal matrix, so that the recessed features of the real oil groove are clearly presented.

[0076] Then, after the images are transmitted to the processing unit of the computer, the space-time registration (existing mature image preprocessing technology) is performed first to correct the motion offset of the connecting rod, and then the features are extracted respectively: the contour and texture features are obtained by edge detection of the visible light image, the oxide layer distribution is marked according to the fluorescence intensity of the ultraviolet image, and the depth mutation area is determined by threshold segmentation of the infrared image.

[0077] Then, the processing unit of the computer integrates the three types of features into a feature vector. Support vector machine algorithm is used for fusion analysis to distinguish the real oil groove / character edge and the interfering dark area.

[0078] Finally, the processing unit performs cross-orientation correlation analysis on the images to detect whether the dark reflection area features matching the first and second orientation reference feature templates exist in a single image. If they exist, it is determined that the measured workpiece has a coplanar defect.

[0079] Generally, before detection, the computer will first start the vector machine algorithm model and load the trained feature classifier (including the spectral feature library of normal characters, oil grooves, oil stains and oxide layers).

[0080] It should be noted that the support vector machine (SVM) algorithm is an existing classification method based on statistical learning theory, and its core is to construct a hyperplane to separate different categories of data in the feature space as accurately as possible. In the present application, in order to distinguish the real oil groove / character edge and the interfering dark area, the processing unit of the computer first integrates the contour closure, texture gradient and fluorescence area into a 12-dimensional feature vector. These features describe the characteristics of the regions in the image from different dimensions.

[0081] The feature vectors are then inputted into a SVM classifier using a Radial Basis Function (RBF) as the kernel function. The RBF kernel function is capable of mapping the original feature vectors into a higher dimensional space, such that data which is originally linearly inseparable in the low dimensional space becomes linearly separable in the high dimensional space. In this high dimensional feature space, the SVM classifier separates the feature vectors representing the true sump / character edge from the feature vectors representing the interfering dark regions by finding the optimal hyperplane. Specifically, the SVM determines the position and orientation of the hyperplane based on the training data (i.e. a set of feature vectors whose classes are known to be either true sump / character edge or interfering dark regions), such that the margin between the two classes of data points to the hyperplane is maximized, thereby improving the accuracy and generalization ability of the classification.

[0082] The trained SVM classifier is then capable of classifying new input feature vectors. When the processing unit inputs a 12-dimensional feature vector of a region of interest in an image to the classifier, the classifier computes the probability value of the feature vector belonging to either "true sump / character edge" or "interfering dark regions". These probability values reflect the similarity of the feature vector to the two classes of data distribution. The processing unit then performs the cross-orientation correlation analysis on the image to detect whether the dark reflection region features matching the first and second orientation reference feature templates are present in the same image, and combines the probability values outputted by the SVM classifier to compute the probability value of belonging to "coplanar defects". If the probability value is greater than 80% or 90%, the workpiece under test is determined to be acceptable; otherwise, it is determined to be unacceptable. In this way, the support vector machine algorithm achieves effective discrimination of different regions in the image and accurate determination of workpiece defects.

[0083] Although the above describes the illustrative embodiments of the present application in order to enable a person skilled in the art to understand the present application, the present application is not limited to the scope of the embodiments, and all the inventions using the concept of the present application within the spirit and scope of the present application as defined and determined by the appended claims are within the protection of the present application.

Claims

1. A connecting rod character and oil groove coplanar visual inspection device, comprising a light source unit arranged above a workpiece inspection path and an image acquisition unit arranged on the light source unit; characterized in that The light source unit projects light onto the surface of the workpiece, so that the flat area forms a bright reflection area and the feature edge forms a dark reflection area, thereby generating a feature distinguishable optical signal, and the image acquisition unit acquires images of the first orientation surface and the second orientation surface of the workpiece based on the optical signal; A processing unit in communication with the image acquisition unit pre-stores first orientation feature reference templates and second orientation feature reference templates of standard workpieces; The processing unit receives the images acquired by the image acquisition unit and performs cross-orientation correlation analysis on the images to detect whether the dark reflection area features matching the first orientation reference feature template and the second orientation reference feature template exist in a single image simultaneously, and if so, it is determined that the workpiece under test has a coplanar defect, otherwise it is determined to be qualified.

2. The linkage character and oil sump coplanar visual inspection apparatus of claim 1, wherein, The optical axis of the light source unit is perpendicular to the surface of the workpiece, so that the light of the flat area of the workpiece is completely reflected to form a bright reflection area, and the light of the feature edge of the workpiece is obliquely reflected to form a dark reflection area.

3. The linkage character and oil sump coplanar visual inspection apparatus of claim 1, wherein, The light source unit is integrated with alternating LED visible light, ultraviolet light source and infrared light source, and the optical axes of the three groups of light sources converge at the same detection point through a reflecting prism to form coaxial multi-spectral illumination of the workpiece under test.

4. The linkage character and sump coplanar visual inspection apparatus of claim 3, wherein, The first orientation feature reference templates and the second orientation feature reference templates pre-stored by the processing unit include a spectral feature library of the character edge, the oil groove edge, the oil stain and the oxide layer of the standard workpiece, which is used to exclude interfering dark reflection area features.

5. The linkage character and sump coplanar vision inspection apparatus of claim 3, wherein, A diaphragm is arranged in the converging light path of the light source unit, and the diaphragm is uniformly provided with light transmission lenses corresponding to the wavelengths of visible light, ultraviolet light and infrared light, and the diaphragm is driven to rotate by a first servo motor to switch the illumination of different light sources.

6. The linkage character and sump coplanar visual inspection apparatus of claim 5, wherein, An optical filter module is arranged in front of the lens of the image acquisition unit, and the filter channels of the optical filter module are provided with optical filters allowing the passage of corresponding light source wavelengths, and the optical filter module is driven to rotate by a second servo motor to switch the filter channels synchronously with the diaphragm.

7. The linkage character and sump coplanar vision inspection apparatus of claim 6, wherein, The light source unit further comprises a light source controller connected with the first servo motor, the second servo motor and the light source unit; The light source controller is a PLC controller or a multi-channel light source driver based on an ARM chip, and the light source controller activates the LED visible light, ultraviolet light source and infrared light source at a pre-set time sequence.

8. The linkage character and sump coplanar visual inspection apparatus of claim 6, wherein, The diaphragm and the optical filter module are provided with driven teeth, the drive gear of the first servo motor engages with the driven teeth of the diaphragm to drive the diaphragm to rotate and switch the light transmission lenses; The drive gear of the second servo motor engages with the driven teeth of the optical filter module to drive the optical filter module to rotate and switch the filter channels.

9. The linkage character and oil sump coplanar visual inspection apparatus of claim 3, wherein, The images of the first orientation surface and the second orientation surface of the workpiece at least include visible light images, ultraviolet images and infrared images. The processing unit performs space-time registration on the images before performing cross-orientation correlation analysis on the images, so as to correct the connecting rod motion offset, and then extracts contour and texture features of the visible light image, oxidation layer distribution features of the ultraviolet image and depth mutation region features of the infrared image, and integrates the three types of features into a feature vector, and uses the processing unit of the support vector machine algorithm to perform fusion analysis, so as to distinguish the real oil groove / character edge from the interference dark region.

10. The linkage character and sump coplanar vision inspection apparatus of claim 9, wherein, The processing unit is configured to calculate a defect probability value by using the support vector machine algorithm, and determine whether the workpiece is qualified based on a preset probability threshold.

Images