A three-dimensional fluorescence detection device and method for part surface defects

Abstract

The present invention relates to a three-dimensional fluorescence detection device and method for part surface defects, wherein an objective lens is mounted on an objective lens translation stage, and the objective lens translation stage can be slidably connected to an integrated base; a white light illumination source is used for illumination, thereby obtaining part surface texture information, and a fluorescence excitation light source emits intrinsic light and irradiates the surface of the part to be tested, thereby exciting a fluorescence signal; the imaging position is changed layer by layer by moving the objective lens translation stage, and the white light texture image and the fluorescence image of the part to be tested are respectively extracted, and the obtained white light texture image is three-dimensionally fused to obtain the surface texture information and the white light depth map of the fluorescent part to be tested, and the sparse fluorescence image is reconstructed and three-dimensionally fused based on the coordinate position information of the data points in the white light depth map, to obtain a three-dimensional fluorescence detection image, and to achieve defect detection. Compared with the prior art, the present invention has the advantages of high imaging resolution, compact size, and the ability to ensure the true restoration of defect morphology.

Description

Technical Field

[0001] The present invention relates to the field of high-precision optical detection equipment, and in particular to a three-dimensional fluorescence detection device and method for surface defects of parts. Background Art

[0002] The rapid and high-precision detection of scratches, damages and cracks on parts can ensure the safety and reliability of parts during use, and has important application value in industries such as aerospace, automobile manufacturing, nuclear industry and rail transportation. Traditional detection is usually carried out by manual visual methods based on experience under different lighting conditions. However, with the gradual increase in production tasks, the shortcomings of low efficiency of manual inspection have become increasingly prominent, and manual inspection is susceptible to manual visual fatigue, resulting in deviations in the inspection results, which poses a serious risk to product quality. With the continuous improvement of computer image processing capabilities, machine vision inspection methods have been gradually applied to the detection of scratches, damages and cracks on parts. The machine vision method illuminates the parts to be tested through a light source, and then uses an imaging lens and camera to collect image information of the object and transmit it to the computer. The scratches, damages and cracks on the surface of the parts are processed and identified through algorithms and programs. However, this method has the following two important problems: (1) Due to the limited depth of field of the high-resolution imaging optical system, it is impossible to achieve high-resolution imaging of scratches, damages and cracks at all heights on the three-dimensional surface of objects with three-dimensional undulations; (2) Some damages are very minor, and conventional lighting and imaging technologies cannot obtain information on scratches, damages and cracks on the surface of parts with high sensitivity, resulting in frequent missed judgments.

[0003] Therefore, how to quickly and accurately detect scratches, damage and cracks on the surface of three-dimensional parts, realize efficient digital characterization and analysis of geometric quantities such as width, length and depth of these defects, and ensure the true restoration of defect morphology and inspection traceability, is still a difficult problem that plagues the field of high-precision parts processing and inspection and needs to be solved urgently. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a three-dimensional fluorescent detection device and method for detecting scratches, damage and cracks on the surface of three-dimensional parts quickly and with high precision, to achieve efficient digital characterization and analysis of geometric quantities such as the width, length and depth of these defects, to ensure the true restoration of the defect morphology and to inspect and trace the surface defects of parts.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A three-dimensional fluorescence detection device for surface defects of parts, comprising an integrated base, an image sensor, a tube lens, an intrinsic light filtering structure, a white light illumination source, a fluorescence excitation light source, an objective lens and an objective lens translation stage, wherein the image sensor, the tube lens, the intrinsic light filtering structure and the objective lens translation stage are all mounted on the integrated base;

[0007] The image sensor, the tube lens, the intrinsic light filtering structure and the objective lens are located in the same optical axis direction;

[0008] The objective lens is mounted on an objective lens translation stage and faces the surface of the fluorescent part to be measured. The objective lens translation stage can be slidably connected to the integrated base, and the sliding direction is the direction of the optical axis in the objective lens;

[0009] The fluorescence excitation light source is used to excite the fluorescence signal and irradiate the surface of the fluorescent part to be tested;

[0010] The white light illumination source is used to output a white light signal and illuminate the surface of the fluorescent part to be tested.

[0011] Furthermore, the intrinsic light filtering structure is a dichroic mirror, which is located between the tube lens and the objective lens; the fluorescence excitation light source is a coaxial fluorescence excitation light source, which is coaxially arranged with the input end on one side of the dichroic mirror; the fluorescence signal excited by the coaxial fluorescence excitation light source is reflected by the dichroic mirror and then irradiated to the surface of the fluorescent part to be measured through the objective lens.

[0012] Furthermore, the three-dimensional fluorescence detection device for part surface defects also includes a spectrometer, the white light illumination light source is a coaxial white light illumination light source, the coaxial white light illumination light source is coaxially arranged with one side input end of the spectrometer, the spectrometer is located between the dichroic mirror and the coaxial fluorescence excitation light source, and the white light signal emitted by the coaxial white light illumination light source is reflected by the spectrometer, passes through the dichroic mirror, and is irradiated to the surface of the fluorescent part to be measured through the objective lens.

[0013] Furthermore, the fluorescence excitation light source is an annular fluorescence excitation light source, which has an annular structure and is installed on the outer edge of the objective lens.

[0014] Furthermore, the intrinsic light filtering structure is a dichroic mirror or a filter, the dichroic mirror is located between the tube lens and the objective lens, and the filter is located between the image sensor and the tube lens.

[0015] Furthermore, the device for three-dimensional fluorescence detection of surface defects of parts also includes an annular white light illumination source, which is an annular structure and is installed on the outer edge of the objective lens.

[0016] The present invention also provides a detection method using the above-mentioned three-dimensional fluorescence detection device for surface defects of parts, comprising the following steps:

[0017] The objective lens is moved by the objective lens translation stage. When the objective lens moves to a certain position, the white light illumination source is turned on and irradiated to the surface of the fluorescent part to be measured. The texture information of the surface of the fluorescent part to be measured is reflected into the objective lens, and then passes through the tube lens and is captured by the image sensor; the objective lens position remains unchanged, the white light illumination source is turned off, the fluorescence excitation light source is turned on and irradiated to the surface of the fluorescent part to be measured, and the fluorescent material on the surface of the fluorescent part to be measured is excited to emit fluorescence. After the fluorescence enters the objective lens, the intrinsic light is filtered out by the intrinsic light filtering structure, and then passes through the tube lens and is captured by the image sensor;

[0018] The objective lens is moved along the optical axis by the objective lens translation stage, and each layer of height steps of the fluorescent material attached to the surface of the fluorescent part to be measured is scanned at a preset scanning distance, thereby changing the object plane corresponding to the photosensitive surface of the image sensor, and recording the white light texture image and fluorescent image corresponding to the object plane of each layer of height steps;

[0019] The obtained white light texture images of each layer of height steps are fused in three dimensions to obtain the surface texture information and white light depth map of the fluorescent part to be tested. The fluorescent images at the corresponding coordinate positions are traversed according to the coordinates of each data point in the white light depth map. After three-dimensional fusion processing of the traversed fluorescent information, a fluorescent depth map is obtained, thereby detecting defects on the surface of the fluorescent part to be tested.

[0020] Furthermore, the method further comprises:

[0021] Perform image feature processing on each layer of white light texture image, and the process of image feature processing includes:

[0022] The fluorescence image is feature decomposed by a feature space transformation function to obtain multiple sub-images containing different frequency components;

[0023] Each sub-image obtained is evaluated by a saliency evaluation function, and a fusion weight map is constructed according to the evaluation results;

[0024] Through the filtering function, the weight values ​​are cross-optimized according to the correlation degree and position proximity of each weight value in the fusion weight map to improve the overall signal-to-noise ratio of the fusion weight map and the signal-to-noise ratio of each sub-region;

[0025] The optimized fusion weight map is used to perform weighted fusion on the sub-images after decomposition of the white light texture image to obtain the optimized feature image.

[0026] Furthermore, the process of performing three-dimensional fusion processing on the white light texture images corresponding to the object planes of each layer of height steps is specifically as follows:

[0027] Using the saliency evaluation function, the image cube composed of the collected white light texture image sequence is evaluated for saliency, and an initial saliency evaluation cube is obtained;

[0028] The feature image pairs are used to provide the initial saliency evaluation cube with prior knowledge of color, shape and distance, and the edge-preserving guided filtering method is applied to optimize the initial saliency evaluation cube to obtain the modified saliency evaluation cube.

[0029] For each pixel in the modified saliency evaluation cube, extract the saliency evaluation from the pixel in the fluorescence image sequence to form a set, and construct a saliency evaluation curve; fit the saliency evaluation curve, and mark the position corresponding to the maximum saliency value; use the marked position as the depth of the image cube along the optical axis, so as to form an initial depth map;

[0030] According to the saliency evaluation value of the modified saliency evaluation cube and its correlation with the initial depth map, the confidence information of the initial depth map is constructed and encoded into a binary mask; the feature image is used as the feature map, the area to be processed is specified according to the binary mask, and the initial depth map is subjected to weighted median filtering, so as to distinguish and process the strong feature area and the weak feature area, and obtain the final modified depth map;

[0031] Constructing a fluorescence Gaussian pyramid and a white light Gaussian pyramid for the fluorescence depth map data and the white light depth map data respectively;

[0032] Based on the fluorescence Gaussian pyramid and the white light Gaussian pyramid, the fluorescence Laplacian pyramid and the white light Laplacian pyramid are constructed respectively;

[0033] Define the mask image and construct the corresponding Gaussian pyramid for the mask image;

[0034] The corresponding layer of the Gaussian pyramid of the mask image is used to perform weighted fusion on the fluorescence Laplacian pyramid and the white light Laplacian pyramid to obtain a fused depth map, which is then sampled and filtered to obtain the final fused depth map. The calculation expression of the final fused depth map is:

[0035]

[0036]

[0037] Where D c is the final fused depth map, n is the number of layers, is the i-th layer of the fused depth map, is the i-th layer of the Gaussian pyramid of the mask image, is the i-th layer of the fluorescence Laplacian pyramid, is the i-th layer of the white light Laplace pyramid.

[0038] Compared with the prior art, the present invention has the following advantages:

[0039] (1) The three-dimensional fluorescence detection device and method for part surface defects proposed in the present invention combine the technical advantages of fluorescence detection and three-dimensional detection. The imaging position is changed layer by layer by moving the objective lens translation stage, and the white light texture image and fluorescence image of the tested part are extracted respectively. The surface texture information and white light depth map of the tested fluorescent part are obtained through three-dimensional fusion processing. The sparse fluorescence image is reconstructed and three-dimensionally fused based on the coordinate position information of the data points in the white light depth map to obtain a three-dimensional fluorescence detection image, thereby realizing part surface defect detection. It can effectively realize high-resolution imaging of scratches, damages and cracks at various heights of three-dimensional undulating objects, and can effectively solve the problems of low efficiency and easy deviation in traditional manual visual inspection.

[0040] (2) The three-dimensional fluorescence detection device for part surface defects proposed by the present invention adopts an integrated base connection, has a simple structure and a compact size, so that it can be used for detection in some space-constrained environments and has strong applicability.

[0041] (3) The present invention constructs an optical system whose imaging position can be changed layer by layer, collects the fluorescence image information at different imaging positions and then performs image fusion on them, thereby solving the problem of limited depth of field in high-resolution imaging optical systems.

[0042] (4) The synchronous detection of white light texture images and fluorescent images and the three-dimensional fusion technology proposed in the present invention can achieve comprehensive detection of defects in the object to be tested and accurately obtain the relative position of the defect area. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic structural diagram of a three-dimensional fluorescence detection device for surface defects of parts provided in an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of the optical path of a three-dimensional fluorescence detection device for surface defects of a part provided in an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of a light path of a ring light source added to a three-dimensional fluorescence detection device for surface defects of a part provided in an embodiment of the present invention;

[0046] Figure 4 A schematic diagram of a light path in which a coaxial light source is eliminated and an annular light source is added in a three-dimensional fluorescence detection device for surface defects of a part provided in an embodiment of the present invention;

[0047] Figure 5A schematic diagram of a light path in which a coaxial light source and a dichroic mirror are eliminated and an annular light source and a filter are added in a three-dimensional fluorescence detection device for surface defects of a part provided in an embodiment of the present invention;

[0048] Figure 6 A schematic diagram of a white light texture image provided in an embodiment of the present invention;

[0049] Figure 7 A schematic diagram of a fluorescent image provided in an embodiment of the present invention;

[0050] Figure 8 A schematic diagram of a three-dimensional fusion result provided in an embodiment of the present invention;

[0051] In the figure, 1. integrated base, 2. image sensor, 3. tube lens, 4. coaxial fluorescence excitation light source, 5. dichroic mirror, 6. objective lens translation stage, 7. objective lens, 8. beam splitter, 9. coaxial white light illumination light source, 10. annular fluorescence excitation light source, 11. filter, 12. annular white light illumination light source. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0055] In the description of the present invention, it should be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings, or are the directions or positional relationships in which the inventive product is usually placed when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as a limitation on the present invention.

[0056] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0057] In addition, the terms "horizontal", "vertical" and the like do not mean that the components are required to be absolutely horizontal or suspended, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0058] Example 1

[0059] like Figure 1 As shown, this embodiment provides a three-dimensional fluorescence detection device for surface defects of parts, including an integrated base 1, an image sensor 2, a tube lens 3, an intrinsic light filtering structure, a white light illumination source, a fluorescence excitation light source, an objective lens 7 and an objective lens translation stage 6, wherein the image sensor 2, the tube lens 3, the intrinsic light filtering structure and the objective lens translation stage 6 are all mounted on the integrated base 1;

[0060] The image sensor 2, the tube lens 3, the intrinsic light filtering structure and the objective lens 7 are located in the same optical axis direction;

[0061] The objective lens 7 is mounted on the objective lens translation stage 6 and faces the surface of the fluorescent part to be measured. The objective lens translation stage 6 is slidably connected to the integrated base 1, and the sliding direction is the optical axis direction of the objective lens 7;

[0062] The white light illumination source emits white light and illuminates the surface of the three-dimensional part to obtain the texture information on the surface of the fluorescent part being tested.

[0063] The fluorescence excitation light source is used to emit intrinsic light to the surface of the fluorescent part to be tested, thereby exciting the fluorescence signal;

[0064] The intrinsic light filtering structure is used to filter out the intrinsic light emitted by the fluorescence excitation light source, and transmit the fluorescence with a wavelength greater than the intrinsic light. The intrinsic light filtering structure can be a dichroic mirror 5 or a filter 11, the dichroic mirror 5 is located between the tube lens 3 and the objective lens 7, and the filter 11 is located between the image sensor 2 and the tube lens 3.

[0065] The fluorescence excitation light source may be a coaxial fluorescence excitation light source 4 or an annular fluorescence excitation light source 10 . The annular fluorescence excitation light source 10 is an annular structure and is installed on the outer edge of the objective lens 7 .

[0066] Preferably, the above-mentioned three-dimensional fluorescence detection device for part surface defects can also be provided with an annular white light illumination source 12, which is a ring-shaped structure and is installed on the outer edge of the objective lens 7 for the surface texture image of the fluorescent part to be tested.

[0067] It should be noted that the fluorescent parts to be tested mentioned in the above method are pre-processed, that is, the parts to be tested are immersed in fluorescent liquid and then taken out, and the fluorescent liquid will remain in the defective area on the surface to achieve defect marking, thereby obtaining the fluorescent parts to be tested.

[0068] like Figure 1 and Figure 2 As shown, in this embodiment, the intrinsic light filtering structure is a dichroic mirror 5, which is located between the tube lens 3 and the objective lens 7; the fluorescence excitation light source is a coaxial fluorescence excitation light source 4, which is coaxially arranged with the input end on one side of the dichroic mirror 5; the fluorescence signal excited by the coaxial fluorescence excitation light source 4 is reflected by the dichroic mirror 5, and then irradiated to the surface of the fluorescent part to be measured through the objective lens 7; the fluorescent substance on the surface of the fluorescent part to be measured is excited to emit fluorescence, and the fluorescence enters the objective lens, passes through the dichroic mirror, and then passes through the tube lens, and is finally captured by the image sensor.

[0069] The detection method using the above-mentioned three-dimensional fluorescence detection device for part surface defects includes the following steps:

[0070] S1: The objective lens 7 is moved by the objective lens translation stage 6. When the objective lens 7 moves to a certain position, the white light illumination source is turned on to illuminate the surface of the fluorescent part to be measured. The texture information of the surface of the fluorescent part to be measured is reflected into the objective lens 7, and then passes through the tube lens 3 and is captured by the image sensor 2; the position of the objective lens 7 remains unchanged, the white light illumination source is turned off, the fluorescence excitation light source is turned on, the fluorescence signal is excited, and illuminated to the surface of the fluorescent part to be measured. The fluorescent substance on the surface of the fluorescent part to be measured is excited to emit fluorescence. After the fluorescence enters the objective lens 7, the intrinsic light is filtered out by the intrinsic light filtering structure, and then passes through the tube lens 3 and is captured by the image sensor 2;

[0071] S2: The objective lens 7 is moved along the optical axis by the objective lens translation stage 6, and each layer of height steps of the fluorescent material attached to the surface of the fluorescent part to be measured is scanned at a preset scanning distance, thereby changing the object plane corresponding to the photosensitive surface of the image sensor 2, and recording the white light texture image and the fluorescent image corresponding to the object plane of each layer of height steps, such as Figure 6 and Figure 7 As shown;

[0072] S3: Perform three-dimensional fusion on the obtained white light texture images of each layer of height steps to obtain the surface texture information and white light depth map of the fluorescent part to be tested, traverse the fluorescent images at the corresponding coordinate positions according to the coordinates of each data point in the white light depth map, perform three-dimensional fusion processing on the traversed fluorescent information, and obtain the fluorescent depth map, so as to detect defects on the surface of the fluorescent part to be tested.

[0073] Specifically, in step S2, image feature processing is required for the image data recorded by the three-dimensional fluorescence detection device for surface defects of parts at each height step, and the processing method is as follows:

[0074] S21: Decompose the white light texture image by using the feature space transformation function to obtain multiple sub-images containing different frequency components, specifically:

[0075] Assume that the image size is H×W, and use the feature space transformation function f trans (·), decompose the image features. For each image I in the original image sequence obtained by scanning the objective lens at equal intervals n ,in Perform multi-scale decomposition into sub-images containing different frequency components Where n=1,2,...,N is the image number, c∈[2,C] is the component number, and C is the decomposition level.

[0076] S22: Evaluate each obtained sub-image using a saliency evaluation function, and construct a fusion weight map based on the evaluation results, specifically:

[0077] Through the significance evaluation function f sal (·), for each sub-image The evaluation results are And construct the fusion weight map based on the significance level evaluation results in and

[0078] S23: Through the filtering function, the weight values ​​are cross-optimized according to the correlation degree and position proximity of each weight value in the fusion weight map to improve the overall signal-to-noise ratio of the fusion weight map and the signal-to-noise ratio of each sub-region, specifically:

[0079] Through the filtering function, the fusion weight map The weight values ​​with high correlation and adjacent positions are cross-optimized to improve the overall signal-to-noise ratio (SNR) of the fused weight map. overall And the signal-to-noise ratio (SNR) of each sub-area local .

[0080] S24: using the optimized fusion weight map to perform weighted fusion on each sub-image after decomposition of the white light texture image to obtain an optimized feature image, specifically:

[0081] Using weights The multi-scale decomposed image sequence is weighted fused. Specifically, for each sub-image And its corresponding weight graph Fusion Component It is obtained by weighted averaging the pixel values ​​at the same position:

[0082]

[0083] Where (x, y) is the pixel position, W i,j (x,y) is the fusion weight of image component i,j at position (x,y), I i,j (x,y) is the pixel value of image component i,j at position (x,y). Thus, the fused pixel value at each position (x,y) is It is the weighted average of all image sequences and frequency components.

[0084] By step feature space transformation function f trans The inverse transformation of (·) These fused components can be recombined to obtain a globally clear feature image

[0085] In step S22, the saliency evaluation function is implemented in the following manner:

[0086] 1. Saliency evaluation functions based on information statistics, such as grayscale variance function, image information entropy function, etc.;

[0087] For an image f(x,y) of size M×N, the expression of the grayscale variance function Variance is:

[0088]

[0089] Among them, μ is the mean value of the image grayscale value:

[0090]

[0091] 2. Saliency evaluation function based on frequency domain information, such as spatial frequency function, discrete cosine transform function, etc.;

[0092] For an image f(x,y) of size M×N, the expression of the spatial frequency function SF is:

[0093]

[0094] Where RF and CF are the row and column frequencies respectively:

[0095]

[0096]

[0097] 3. Spatial domain-based saliency evaluation functions, such as modified Laplace sum function, grayscale gradient function, gradient energy function, FSWM filter function, local maximum gradient, etc.;

[0098] For a window size of N×N, the expression of the modified Laplace and function SML is:

[0099]

[0100] Among them, T is the discrimination threshold, N is the window size, The discrete approximate expression for the modified Laplace operator ML is:

[0101]

[0102] Specifically, in step S3, the white light texture images corresponding to the object planes of each layer of height steps are subjected to three-dimensional fusion processing to obtain a white light depth map and a fluorescence depth map, thereby detecting defects on the surface of the fluorescent part being tested. Specifically, the process is as follows:

[0103] 31) Using the saliency evaluation function, the image cube composed of the collected white light texture image sequence is evaluated for saliency, and an initial saliency evaluation cube is obtained, which is specifically:

[0104] Suppose the image cube formed by the original image sequence is Where H and W represent the height and width of the image respectively, and N represents the number of image sequences. sal (·), for C orig Perform pixel-level or window-level saliency evaluation to obtain the initial saliency evaluation cube in

[0105] 32) The feature image pairs are used to provide the initial saliency evaluation cube with prior knowledge on color, shape and distance, and the edge-preserving guided filtering method is applied to optimize the initial saliency evaluation cube to obtain a modified saliency evaluation cube, which is specifically:

[0106] Using feature image for Provides prior knowledge of color, shape, and distance, and applies edge-preserving guided filtering methods to Optimize to obtain the modified saliency evaluation cube

[0107] 33) For each pixel in the modified saliency evaluation cube, extract the saliency evaluation of the pixel from the white light texture image sequence to form a set, and construct a saliency evaluation curve; fit the saliency evaluation curve, and mark the position corresponding to the maximum saliency value; use the marked position as the depth of the image cube along the optical axis direction, thereby forming an initial depth map, specifically:

[0108] consider For each pixel (i,j), extract the saliency evaluation from the pixel in the N image sequence to form a set {v1,v2,...,v N}. Process the set and establish the significance evaluation curve Curve ij (k) = v k , where k∈[1,N]. Then, use the high-order polynomial p(k) to fit the Curve ij , and mark the positions corresponding to the maximum saliency values ​​on the curve. These marked positions correspond to the cube C orig The depth along the optical axis, thus forming the initial depth map D init .

[0109] 34) According to the saliency evaluation value of the modified saliency evaluation cube and its correlation with the initial depth map, the confidence information of the initial depth map is constructed and encoded into a binary mask; the feature image is used as the feature map, the area to be processed is specified according to the binary mask, and the initial depth map is subjected to weighted median filtering, so as to distinguish and process the strong feature area and the weak feature area, and obtain the final modified depth map;

[0110] The significance evaluation value and its relationship with D init The correlation between and constructs the initial depth map confidence information, which is encoded as a binary mask M. As the feature map, and according to M to specify the area to be processed, D init Perform weighted median filtering to distinguish and process strong feature areas and weak feature areas to obtain the final corrected depth map D final .

[0111] Based on the final corrected white light depth map, the fluorescence depth map is obtained, and the defects on the surface of the fluorescent part being tested are displayed, specifically:

[0112] According to the coordinates of each data point in the white light depth map, the fluorescence image at the corresponding coordinate position is traversed, and the fluorescence information obtained by the traversal is 3D fused to obtain a fluorescence depth map, thereby detecting defects on the surface of the fluorescent part being tested.

[0113] Example 2

[0114] This embodiment is substantially the same as the first embodiment, except that the detection method of the three-dimensional fluorescence detection device for surface defects of parts further includes:

[0115] S4: To highlight potential defect areas and accurately obtain the relative position of the defect areas, the white light depth map and the fluorescence depth map can be fused. The specific process is as follows:

[0116] 41) Extract fluorescence depth data D f and white light depth map data D p ;

[0117] 42) D f and D p Construct Gaussian pyramids separately and Where i is the level.

[0118] 43) Based on Gaussian pyramid and Constructing a Laplacian Pyramid and Among them, each layer of the Laplacian pyramid is obtained by calculating the difference between the Gaussian pyramid of the current layer and the Gaussian pyramid of the next layer after upsampling and filtering, that is, The same is true for D p Perform similar processing.

[0119] 44) Define the mask image M and construct the corresponding Gaussian pyramid for the mask image

[0120] 45) Use the corresponding layer of the Gaussian pyramid of the mask image to perform weighted fusion on the fluorescence Laplacian pyramid and the white light Laplacian pyramid to obtain a fused depth map, and perform sampling and filtering to obtain the final fused depth map, specifically:

[0121] At each scale i, use the corresponding layer of the mask map Come to and Specifically, the fused depth map

[0122] Starting from the coarsest Laplacian pyramid layer, each layer is added to the upsampled and filtered result of the layer below it until the finest layer. Specifically, Where n is the number of layers; finally, the final fused depth map is obtained.

[0123] like Figure 8As shown, the final fused depth map is combined with the surface texture information obtained under white light to generate a three-dimensional fused image, thereby achieving comprehensive detection of defects in the object to be tested and accurately obtaining the relative position of the defect area.

[0124] Example 3

[0125] This embodiment is substantially the same as Embodiment 1, except that Figure 3 As shown, an annular white light illumination source 12 is placed at the outer edge of the objective lens. At each level step, the annular white light source and the coaxial fluorescent light source are turned on alternately. The fluorescent image is reconstructed by detecting and calculating the coordinates of the data points of the generated white light depth map, and a three-dimensional fluorescent detection image is obtained by calculation.

[0126] Example 4

[0127] This embodiment is substantially the same as Embodiment 1, except that Figure 4 As shown, the fluorescence excitation light source is an annular fluorescence excitation light source 10, which is an annular structure and is installed at the outer edge of the objective lens 7. The annular fluorescence excitation light source 10 illuminates the surface of the fluorescent part to be measured, excites the fluorescence signal, and the intrinsic light cannot pass through the tube lens 3 to enter the image sensor 2 after being reflected by the dichroic mirror 5. The fluorescence signal can be recorded by the image sensor 2 after passing through the tube lens 3.

[0128] Example 5

[0129] This embodiment is substantially the same as Embodiment 1, except that Figure 5 As shown, the fluorescence excitation light source is an annular fluorescence excitation light source 10, which is an annular structure and is installed on the outer edge of the objective lens 7. The three-dimensional fluorescence detection device for surface defects of parts also includes an annular white light illumination light source 12, which is an annular structure and is installed on the outer edge of the objective lens 7. The intrinsic light filtering structure is a filter 11, and the filter 11 is located between the image sensor 2 and the tube lens 3. The annular fluorescence excitation light source 10 illuminates the surface of the fluorescent part to be tested, excites the fluorescence signal, the intrinsic light is filtered by the filter 11, and the fluorescence signal can be recorded by the image sensor 2 after passing through the tube lens 3.

[0130] Example 6

[0131] The fluorescence excitation light source is an annular fluorescence excitation light source 10, which is an annular structure and is installed on the outer edge of the objective lens 7. The three-dimensional fluorescence detection device for surface defects of parts also includes an annular white light illumination light source 12, which is an annular structure and is installed on the outer edge of the objective lens 7.

[0132] The annular fluorescent excitation light source 10 and the annular white light illumination light source 12 are combined into an annular light source. The annular white light illumination light source 12 in the annular light source is used to shoot the surface texture of the object to be measured at the next moment. The annular fluorescent excitation light source 10 in the annular light source is used to excite and shoot the fluorescent image. By recording the image data of steps at different heights, the synchronous detection of the white light texture image and the fluorescent image is realized. The fluorescent image is reconstructed by detecting and calculating the coordinates of the data points of the generated white light depth map, and a three-dimensional fluorescent detection image is obtained by calculation.

[0133] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A detection method of a three-dimensional fluorescence detection device for part surface defects, characterized in that: The device comprises an integrated base (1), an image sensor (2), a tube lens (3), an intrinsic light filtering structure, a white light illumination light source, a fluorescence excitation light source, an objective lens (7) and an objective lens translation stage (6); the image sensor (2), the tube lens (3), the intrinsic light filtering structure and the objective lens translation stage (6) are all mounted on the integrated base (1); The image sensor (2), the tube lens (3), the intrinsic light filtering structure and the objective lens (7) are located in the same optical axis direction; The objective lens (7) is mounted on the objective lens translation stage (6) and faces the surface of the fluorescent part to be measured. The objective lens translation stage (6) is slidably connected to the integrated base (1), and the sliding direction is the direction of the optical axis in the objective lens (7); The fluorescence excitation light source is used to excite the fluorescence signal and irradiate the surface of the fluorescent part to be tested; The white light illumination source is used to output a white light signal and illuminate the surface of the fluorescent part to be tested; The detection method comprises the following steps: The objective lens (7) is moved by the objective lens translation stage (6); when the objective lens (7) moves to a certain position, the white light illumination source is turned on to illuminate the surface of the fluorescent part to be measured, and the texture information of the surface of the fluorescent part to be measured is reflected into the objective lens (7), and then passes through the tube lens (3) and is captured by the image sensor (2); the position of the objective lens (7) is kept stationary, the white light illumination source is turned off, and the fluorescence excitation light source is turned on to illuminate the surface of the fluorescent part to be measured, and the fluorescent material on the surface of the fluorescent part to be measured is excited to emit fluorescence, and the fluorescence enters the objective lens (7), passes through the intrinsic light filtering structure to filter out the intrinsic light, and then passes through the tube lens (3) and is captured by the image sensor (2); The objective lens (7) is moved along the optical axis by the objective lens translation stage (6), and each layer of height steps of the fluorescent material attached to the surface of the fluorescent part to be measured is scanned at a preset scanning distance, thereby changing the object plane corresponding to the photosensitive surface of the image sensor (2), and recording the white light texture image and the fluorescent image corresponding to the object plane of each layer of height steps; The obtained white light texture images of each layer of height steps are fused to obtain the surface texture information and white light depth map of the fluorescent part to be tested. The fluorescent image at the corresponding coordinate position is traversed according to the coordinates of each data point in the white light depth map. After three-dimensional fusion processing of the traversed fluorescent information, a fluorescent depth map is obtained to detect defects on the surface of the fluorescent part to be tested.

2. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 1, characterized in that: The intrinsic light filtering structure is a dichroic mirror (5), and the dichroic mirror (5) is located between the tube mirror (3) and the objective lens (7); the fluorescence excitation light source is a coaxial fluorescence excitation light source (4), which is coaxially arranged with an input end on one side of the dichroic mirror (5); the fluorescence signal excited by the coaxial fluorescence excitation light source (4) is reflected by the dichroic mirror (5) and then irradiated onto the surface of the fluorescent part to be measured through the objective lens (7).

3. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 2, characterized in that: The component surface defect three-dimensional fluorescence detection device further comprises a beam splitter (8); the white light illumination source is a coaxial white light illumination source (9); the coaxial white light illumination source (9) is coaxially arranged with an input end on one side of the beam splitter (8); the beam splitter (8) is located between the dichroic mirror (5) and the coaxial fluorescence excitation light source (4); the white light signal emitted by the coaxial white light illumination source (9) is reflected by the beam splitter (8), passes through the dichroic mirror (5), and is irradiated onto the surface of the fluorescent component to be detected through the objective lens (7).

4. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 1, characterized in that: The fluorescence excitation light source is an annular fluorescence excitation light source (10), which is an annular structure and is installed on the outer edge of the objective lens (7).

5. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 1, characterized in that: The intrinsic light filtering structure is a dichroic mirror (5) or a filter (11); the dichroic mirror (5) is located between the tube lens (3) and the objective lens (7); and the filter (11) is located between the image sensor (2) and the tube lens (3).

6. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 1, characterized in that: The device for three-dimensional fluorescence detection of surface defects of parts further comprises an annular white light illumination source (12). The annular white light illumination source (12) is an annular structure. The annular white light illumination source (12) is mounted on the outer edge of the objective lens (7).

7. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 1, characterized in that: The method further comprises: The image feature processing is performed on the white light texture image of each layer of height steps, and the process of the image feature processing includes: The white light texture image is feature decomposed by a feature space transformation function to obtain multiple sub-images containing different frequency components; Each sub-image obtained is evaluated by a saliency evaluation function, and a fusion weight map is constructed according to the evaluation results; Through the filtering function, the weight values ​​are cross-optimized according to the correlation degree and position proximity of each weight value in the fusion weight map to improve the overall signal-to-noise ratio of the fusion weight map and the signal-to-noise ratio of each sub-region; The optimized fusion weight map is used to perform weighted fusion on the sub-images after decomposition of the white light texture image to obtain the optimized feature image.

8. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 7, characterized in that: The specific process of three-dimensional fusion processing of the white light texture images of each layer of height steps is as follows: Using the saliency evaluation function, the image cube composed of the collected white light texture image sequence is evaluated for saliency, and an initial saliency evaluation cube is obtained; The feature image pairs are used to provide the initial saliency evaluation cube with prior knowledge of color, shape and distance, and the edge-preserving guided filtering method is applied to optimize the initial saliency evaluation cube to obtain the modified saliency evaluation cube. For each pixel in the modified saliency evaluation cube, extract the saliency evaluation of the pixel from the white light texture image sequence to form a set, and construct a saliency evaluation curve; fit the saliency evaluation curve, and mark the position corresponding to the maximum saliency value; use the marked position as the depth of the image cube along the optical axis, so as to form an initial depth map; According to the saliency evaluation value of the modified saliency evaluation cube and its correlation with the initial depth map, the confidence information of the initial depth map is constructed and encoded into a binary mask; The feature image is used as the feature map, the area to be processed is specified according to the binary mask, and the initial depth map is processed by weighted median filtering to distinguish and process the strong feature area and the weak feature area to obtain the final corrected depth map.

9. The detection method of a three-dimensional fluorescence detection device for part surface defects according to claim 7, characterized in that: The method further comprises: Constructing fluorescence Gaussian pyramid and white light Gaussian pyramid for the fluorescence depth map and the white light depth map respectively; Based on the fluorescence Gaussian pyramid and the white light Gaussian pyramid, the fluorescence Laplacian pyramid and the white light Laplacian pyramid are constructed respectively; Define the mask image and construct the corresponding Gaussian pyramid for the mask image; The corresponding layer of the Gaussian pyramid of the mask image is used to perform weighted fusion on the fluorescence Laplacian pyramid and the white light Laplacian pyramid to obtain a fused depth map, which is then sampled and filtered to obtain the final fused depth map. The calculation expression of the final fused depth map is: Where D c = is the final fused depth map, n is the number of layers, is the i-th layer of the fused depth map, is the i-th layer of the Gaussian pyramid of the mask image, is the i-th layer of the fluorescence Laplacian pyramid, is the i-th layer of the white light Laplace pyramid.

Images