A method for processing dynamic target scanning image data based on geometric features

Abstract

The present invention discloses a method for processing image data of a dynamic target object scanned based on geometric features. The main design concept of the present invention is that, based on the imaging characteristics of an image scanner, the scanned image of the target object is geometrically decomposed in a panoramic and dynamic scene to generate multiple image sections, and feature quantities are queried in these image sections. These feature quantities are then reorganized to correspond to the target object and converted into a complete and linear image. Finally, the optical flow of the reorganized image is estimated to find the image vanishing point so as to balance the image color difference. The present invention can effectively identify waste generated by lubrication in motion, and has high adaptability to working scenes, and can greatly reduce the image distortion caused by the halo generated by the image scanner, thereby significantly improving the image discrimination or recognition performance of the target object.

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology, and in particular to a method for processing dynamic target scanning image data based on geometric features. Background Art

[0002] Cigarette making machines are one of the main production equipment of tobacco enterprises. They roll tobacco, filters and cigarette paper into finished cigarettes and send them to downstream packaging equipment. Each production module of the cigarette making machine will generate a lot of heat and mechanical wear during operation, so lubrication of the equipment is very important. Lubricating oil can keep the heat exchange of the machine running normally and avoid hard contact of the contact surface during the circulation process, thereby increasing the service life of the machine.

[0003] During the circulation process of lubricating oil, some production waste (target objects) will be generated, and will pass through the reflux channel and settle at the bottom of the oil tank. In addition, external dust will slowly accumulate and settle at the bottom of the oil tank due to poor sealing of the oil tank. These will accelerate the deterioration of oil and clog the fuel injection nozzle or pipe joint, resulting in insufficient lubrication or lack of oil for the mechanical parts of the equipment, causing irreversible damage to the mechanical parts.

[0004] For the detection of waste in lubricating oil, a method of using an image scanner to identify the characteristics of the waste can be adopted. This method has a compact structure, a wide scanning range, and can scan the target object in real time. However, this method also has obvious disadvantages. For example, the feature extraction of the target object is not sufficient. This is because the image scanned by the image scanner is limited by the effective range of the scanner. Insufficient effective range will lead to a decrease in the accuracy of image recognition. Moreover, the faster the speed of the target object, the lower the accuracy of image recognition. Although the recognition degree can be improved by replacing a higher-precision image scanner, the investment cost is bound to increase. For example, when the target in the scanning channel of the detection system is moving, that is, the imaging process is strip imaging. When the image scanner scans the moving targets one by one, the image in the center area of ​​the scan will be the clearest due to the focal length of the lens, because the focal length is the shortest at this time, and the most features are extracted. However, for the image in the area far from the center, due to the strip depth of field, the image scanner cannot guarantee that all feature points are "fully sampled". At this time, the scanning of the target far from the center will produce "stretching" distortion, making the center area of ​​the final target image clear, while the image in the farther area is "squeezed", which ultimately causes the recognition of dynamic targets in the panorama to decrease. For example, during the scanning process, the image scanner will have a "halo" in some scenes. In bright or flickering scenes or scenes with strong reflections around, the "halo" will cause the target's feature quantity to be extracted incorrectly, and will also cause the accuracy of image recognition to decrease. Summary of the invention

[0005] In view of the above, the present invention aims to provide a method for processing dynamic target scanning image data based on geometric features, so as to improve the recognition accuracy of the dynamic target scanning image.

[0006] The technical solution adopted by the present invention is as follows:

[0007] A method for processing dynamic target scanning image data based on geometric features, comprising:

[0008] Obtaining dynamic scanning images of lubrication waste;

[0009] Based on a preset virtual rectangular frame sampling strategy, the scanned image is geometrically split to obtain a plurality of image sections;

[0010] Based on a preset panoramic image reconstruction strategy, query the characteristic quantity of the image section, and perform target reconstruction according to the characteristic quantity to obtain a complete and linear lubrication waste image;

[0011] Based on the preset optical flow estimation strategy, the optical flow vanishing point that causes chromatic aberration in the reconstructed lubricating waste image is located.

[0012] In at least one possible implementation, the virtual rectangular frame sampling strategy includes:

[0013] Preset the length and width of the image frame in the sampling channel and introduce a geometric correction factor;

[0014] A virtual rectangular sampling frame is established, and the image intercepted by the rectangular sampling frame during the scanning process is divided into a plurality of sampling sections.

[0015] In at least one possible implementation manner, the correction factor is used to ensure that the image always maintains a relatively correct position in the rectangular sampling frame.

[0016] In at least one possible implementation, the panoramic image reconstruction strategy includes:

[0017] Counting the image frames of lubricating waste in each of the image sections in time sequence;

[0018] Extracting features of the image cross section based on a preset boundary sampling function;

[0019] Linear fitting is performed based on the features, and a preset back-projection mechanism is used to reconstruct and generate a panoramic superposition image.

[0020] In at least one possible implementation, the back-projection mechanism includes:

[0021] Obtaining the projection point source of the lubrication waste in the three-dimensional coordinate system relative to the current frame;

[0022] According to the linear fitting result, the projected linear superposition value is determined;

[0023] Based on the projected point source and in combination with a preset convolution function, the projected linear superposition value is reconstructed to obtain the convolved linear superposition value and the reconstructed point source of the lubricating waste in the three-dimensional coordinate system.

[0024] In at least one possible implementation, the optical flow estimation strategy includes:

[0025] Based on the image scanning order, determining different sampling areas and the splicing areas between adjacent sampling areas;

[0026] Acquiring position information between the splicing areas through feature matching;

[0027] Performing a similarity transformation according to the change of the position information, and estimating the optical flow value of the lubricating waste during movement according to the change of the image frames in the splicing area;

[0028] The coordinate value of the corresponding vanishing point is obtained by projecting the optical flow value and the expansion center point of the lubricating waste into the panoramic three-dimensional coordinate system; wherein the expansion center point is the intersection point formed by the convergence of the sampled optical flows.

[0029] In at least one possible implementation manner, the motion state of the lubricating waste is a uniform motion with a stable posture maintained within a preset time.

[0030] In at least one possible implementation manner, acquiring a dynamic scanning image of the lubricating waste includes: acquiring a scanning image of the lubricating waste in a panoramic and dynamic scene based on a scanning light of a set shape of an image scanning device.

[0031] The main design concept of the present invention is that, based on the imaging characteristics of the image scanner, the scanned image of the target object is geometrically split in a panoramic and dynamic scene to generate multiple image sections, and the feature quantities are queried in these image sections, and then these feature quantities are reorganized corresponding to the target object to convert them into a complete and linear image, and finally the optical flow of the reorganized image is estimated to find the image vanishing point to balance the image color difference. The present invention can effectively identify the waste generated by lubrication in motion, and has a high adaptability to the working scene, and can greatly reduce the image distortion caused by the halo generated by the image scanner, thereby significantly improving the image discrimination or recognition performance of the target object. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described below with reference to the accompanying drawings, in which:

[0033] Figure 1A flowchart of a method for processing dynamic target scanning image data based on geometric features provided by an embodiment of the present invention;

[0034] Figure 2 A schematic diagram of a virtual rectangular frame sampling strategy provided by an embodiment of the present invention;

[0035] Figure 3 A schematic diagram of a panoramic image reconstruction strategy provided by an embodiment of the present invention;

[0036] Figure 4 A schematic diagram of an expansion center point provided by an embodiment of the present invention;

[0037] Figure 5 A schematic diagram of a vanishing point provided by an embodiment of the present invention;

[0038] Figure 6 A schematic diagram of an optical flow estimation strategy provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0039] Embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0040] The present invention proposes an embodiment of a method for processing dynamic target scanning image data based on geometric features. Specifically, Figure 1 shown, including:

[0041] Step S1, obtaining a scanning image of dynamic lubrication waste;

[0042] In actual operation, based on the imaging characteristics of the image scanning device, a scanned image of the lubricating waste (target object) can be obtained in a panoramic and dynamic scene. Specifically, the scanning light emitted by the image scanner is irradiated to the target object in the form of a pencil beam or a fan-shaped surface. After the scanning light is reflected, the image frame (can be referred to as a frame) of the target object is fed back to the computer through the image scanner in the form of an electrical signal for subsequent processing.

[0043] Step S2, based on a preset virtual rectangular frame sampling strategy, performing geometric structure splitting on the scanned image to obtain a plurality of image sections;

[0044] Step S3, based on a preset panoramic image reconstruction strategy, querying the characteristic quantity of the image section, and performing target reconstruction according to the characteristic quantity to obtain a complete and linear lubricating waste image;

[0045] Step S4: based on a preset optical flow estimation strategy, locate the optical flow vanishing point causing the chromatic aberration in the reorganized lubricating waste image.

[0046] The present invention provides the following specific descriptions of the three preset strategies mentioned in steps S2 to S4 of the above embodiment:

[0047] (1) Virtual rectangular frame sampling strategy

[0048] The target object is scanned and imaged by an image scanner. During the imaging process of the target object, a virtual scanning channel can be established in advance. In order to maximize the scanning range, the scanning channel can be defined as a rectangular sampling frame (hereinafter referred to as the sampling frame).

[0049] The main processing idea of ​​the virtual rectangular frame sampling strategy is: first, define the length and width of the image frame in the sampling channel, and introduce a geometric correction factor (used to keep the sampled image always in the correct position in the sampling frame) to avoid the inability to effectively collect the target when it is tilted. Then, establish a virtual rectangular sampling frame (i.e., the sampling frame), and divide the image in the sampling frame into several sampling sections, so as to achieve full sampling of the maximum limit.

[0050] Specific as Figure 2 As shown, we can first define L t is the length of the t-th frame image of the target object, W t is the width of the t-th frame image of the target object, and U t is the panoramic collection frame of the collected t-th frame image, S t is the set of the tth valid frames collected, and S t =L t *W t and

[0051] In actual operation, the acquired frame S t In the panorama collection frame U t When intercepting, it is inevitable that not all frames collected in the sampling frame are valid due to the tilt of the target object, so a geometric correction factor J can be introduced t , which is used to select and intercept the valid frame of the target object in the sampling frame again, so J t = mid(0,L t ,W t ,0)∧left(-5,L t )∧right(W t ,12), where the constants -5 and 12 represent the maximum limit values ​​of the sampling length and width of the sampling frame (which can be set), and then there is S t =J t (L t *Wt ), introduce J t Then, the sampled image is always kept in the correct position in the sampling frame to ensure that the sampling of each frame is valid.

[0052] During the imaging process of the target object, a virtual sampling frame can be established, in which the image is divided into several sections for sampling, and each section is scanned and frame collected one by one. Because the target object moves, the effective frames on each section scanned one by one are S tn A subset of S tn ={S t1 ,S t2 ,...S tm}, n∈ natural number and m<n, S t After the full sampling that meets the maximum limit of the sampling frame, the frames obtained from each section of the full sampling are geometrically corrected and merged together to regenerate a new panoramic image as described below.

[0053] (2) Panoramic Image Reconstruction Strategy

[0054] Based on the above embodiment, after the image of the target object scanned during the sampling process is divided into several image sections for complete sampling, it is known that each divided section S of the sampled image tn All are L t *W t When the target object passes through the sampling frame, the frame of the target object is effectively intercepted (or called sampling) by each section in the sampling frame. At this time, the frames on each section are messy and disordered, which requires reorganization of these frames to find the corresponding feature information.

[0055] In actual operation, the ordinate and abscissa coordinates of each section in the sampling frame can be defined as the polar axis of the plane rectangular coordinate system, that is, the ordinate of the coordinate system is L t , the horizontal axis is W t ,like Figure 2 As shown, the polar axes of the vertical and horizontal coordinates are represented as (0,L t ) and (W t ,0).

[0056] When the target object passes through the sampling frame, since the sampling frame is divided into several sections, the scanning light scans and feeds back the image frames (referred to as frames) on each section layer by layer and in turn, and then the computer counts the image frames (referred to as frames) of the target object in each section, performs linear fitting and then integral calculation, and finally uses back projection reconstruction to generate a new superimposed image. It should be noted that each section in the sampling frame is scanned by the image scanner in turn, and when the image frame (referred to as frame) of the target object in the current section is sampled, it will jump to the next section after the set time ends regardless of whether the sampling is complete.

[0057] The aforementioned back-projection reconstruction algorithm does not involve the motion image of the target under nonlinear speed, but is only used as the motion image of the target under linear speed and the sampling section with the same time interval. It can be explained that projection reconstruction, as an imaging system, is a process that converts the scattered image frames (can be referred to as frames) of the target after sampling into a complete and linear image. In this system, the quality (resolution, clarity, blur, etc.) of the image frames (can be referred to as frames) extracted from each sampling section in all sampling frames plays a vital role in the reconstruction of the image. Back-projection reconstruction can be regarded as a linear process. Figure 3 As shown:

[0058] Let f t is the linear superposition value. In the two-dimensional coordinate system, we have:

[0059]

[0060] Among them B t (L t ,W t ) represents the boundary sampling function of each image.

[0061] In a panoramic image, the image boundary sampling function is regarded as the boundary sampling function of each section in the sampling frame. It represents the maximum feature extraction surface of the current image section and can also be called the kernel of the image. If the image imaging process has a linear invariant property, the kernel can be expressed as a linear fitting equation, that is, B t =αL t +βW t +ε, where α and β represent the fitting parameters corresponding to the length and width of the t-th frame image, respectively, and ε represents the random quantity reflecting the distribution of scattered frames in the sampling interval, and the ε described in the present invention obeys the normal distribution, then the linear superposition value f t It can be calculated as a three-dimensional coordinate system, namely:

[0062]

[0063] For simplicity, in the above expression, let x t =αL t S t ,y t =βW t S t ,z t =εS t , then f t =(x t +y t +z t )dt, where x t ,y t , zt They respectively represent the frames of the target object in the three-dimensional coordinate system, also called point sources. In the present invention, when compensating for the image degradation problem caused by factors such as defects in the image scanner and illumination angle, a key step in the two-dimensional coordinate system is to determine the correct kernel (feature). Similarly, for the three-dimensional coordinate system, the focus should be on analyzing the point source.

[0064] For example, the scanning mode of the image scanner is translational sequential scanning, and the installation angle of the image scanner is 25° with the horizontal plane. Assuming that the angle is θ, the sampling of the target object conforms to the geometric relationship of the three-dimensional coordinate system. x'(t), y'(t) and z'(t) respectively represent the projection frame of the target object in the three-dimensional coordinate system relative to the current frame or the projection point source, and f' t is the projected linear superposition value, such as Figure 3 As shown, there is an angle θ between the current coordinate system and the projection coordinate system, that is, x' t =x t cosθ,y't=y t cosθ, z't = ztcosθ. Although there is an angle θ between the current coordinate system and the projected coordinate system, the origins of the two coordinate systems are consistent and coincident. The origin is O(t) and O' t =O t , then the angle on the horizontal plane can be understood as the angle at the origin, then the projected linear superposition value of the angle θ is:

[0065]

[0066] Next, f'(t) is back-projected and reconstructed, and the convolution function K can be introduced, as follows:

[0067] but

[0068] Assume that the back projection is O t Reconstructing the origin, we have:

[0069]

[0070] Obviously, for back-projection reconstruction, we use the point source x' t , y' t and z' t Perform a back-projection operation and introduce the convolution function K to linearly superimpose the projection value f' t Reconstruct the linear superposition value f after convolution K , so x K ,y K and z K It can be regarded as a filtered frame of the target object in the three-dimensional coordinate system or called a reconstructed point source. It can be understood that the role of the convolution operation here is to filter the frame.

[0071] (3) Optical flow estimation strategy

[0072] As mentioned above, when reconstructing a panoramic image, the illumination angle of the image scanner will cause the sampling light to change, thereby reducing the image sampling quality. Although this can be remedied by the aforementioned method of reconstructing the point source, the process is relatively cumbersome, and the image reconstruction itself consumes more system resources. To this end, another method can be combined to achieve rapid chromatic aberration balance of the sampled image to facilitate effective frame extraction, that is, a fast optical flow estimation method can be used. It should be noted that the fast optical flow estimation method is only applicable to sampling analysis of sampling sections, such as deriving and estimating the optical flow of the current target and performing chromatic aberration balance. If the image scanner continues to sample the next sampling section, the current optical flow derivation will no longer be applicable and must be re-derived.

[0073] The main processing idea of ​​the optical flow estimation strategy is: first, the image is defined as different sampling areas according to the scanning order of the image scanner, then the positions of the stitching areas are obtained through feature matching, and the change of the position is approximated as a similarity transformation, and then the optical flow value of the target object when it moves is estimated by the change of the image frame in the stitching area of ​​the sampled image, and finally the value (coordinate value) of the corresponding vanishing point is estimated by the optical flow value and the expansion center point, so as to achieve the color difference balance of the sampled image and facilitate the effective extraction of the image frame.

[0074] Specifically, the panoramic scanning process of the sampling frame is used as a reference to quickly estimate the optical flow of the image frame (can be referred to as the frame). When the target enters the sampling frame, the image scanner performs a panoramic scan of the target. As long as the target can be reconstructed by back projection, the image of the target can be determined. However, the illumination angle of the image scanner will have a certain impact on the image imaging effect due to the movement of the target, causing some optical flows in the sampled image to disappear and form two areas of different colors, that is, color difference. Therefore, according to the movement speed of the target, a panoramic sampling image is spliced ​​by these two areas, and the color difference is generally concentrated in the splicing area. Therefore, in actual operation, the area scanned first by the image scanner can be defined as area A, the area scanned later can be defined as area B, and the stitching area between the two areas can be defined as area C. They constitute the spatial sampling area R (A, B, C). In order to ensure that the sampling between two adjacent frames in the image is panoramic sampling, the width D of the stitching area C should be equal to the distance moved by the image scanner between two adjacent frames, that is, D = V / R, where V represents the movement speed of the target object, and R represents the set of sampling frames of the image scanner within the sampling frame.

[0075] Furthermore, it is assumed that the position of the internal sampling line in the t-1th frame must ensure that the rectangular sampling area R in the two adjacent frames t-1 (A t-1,B t-1 ,C t-1 ) and R t (A t ,B t ,C t ) have no overlap and no gap (R t Represents the set of all sampling frames in the back area, R t-1 Represents the set of all sampling frames in the front area, A t represents the set of all frames in region A, B t represents the set of all frames in region B, C t represents the set of all frames in region C), then R t-1 With R t are two adjacent regions (front region and back region) in the same sampled image, and the splicing region C t With C t-1 Included in R t With R t-1 Inside, make C t With C t-1 are two overlapping regions, namely Therefore, after determining C t or C t-1 Any one of them can be used to obtain the homography transformation matrix H between adjacent frames through feature matching. H is also called the position parameter of the frame, from which the splicing area C can be obtained. t With C t-1 The position between adjacent frames, i.e., C t =H -1 C t-1 When the target object moves at a steady and uniform speed and there is no obvious posture change in a short period of time, the adjacent frames in the splicing area can be approximated as similar transformations. Therefore, the optical flow value of the target object when it moves can be estimated by the changes in the frames in the splicing area of ​​the sampled image, that is:

[0076]

[0077] Where G t Indicates the optical flow value, area A t-1 , B t-1 With C t-1 After panoramic scanning, the linear superposition value f' is obtained with the projection t After extraction, the target object itself is projected into the panoramic three-dimensional coordinate system, and the intersection point where the sampled optical flows converge is called the expansion center point P t ,like Figure 4 As shown, the image scanner illuminates the target object and projects it into the panoramic three-dimensional coordinate system. The intersection point where the scanning light flow converges is called the vanishing point Q t ,like Figure 5 As shown. Vanishing point Qt The existence of is due to the inconsistency between the sampling light and the scanning optical flow caused by the angle θ between the image scanner and the horizontal plane. The geometric estimation based on the optical flow consistency assumption can improve the optical flow estimation speed.

[0078] When the optical flow consistency assumption is established, it can be judged that there must be a set of optical flows parallel to the target object's motion direction and the angle θ formed by the scanning optical flow emitted by the image scanner in the sampling three-dimensional coordinate system, corresponding to the vanishing point Q t With the expansion center point P t There must be an estimation algorithm for the geometric relationship, namely:

[0079]

[0080] For any scanning light in the sampling space that has an angle with the target object's motion direction, its projection in the three-dimensional coordinate system must form an optical flow expansion point after projection superposition. Therefore, if the position of the optical flow expansion center point is determined, the position of the optical flow vanishing point can be quickly determined and is relatively reliable. The aforementioned fast optical flow estimation method can be referenced. Figure 6 shown.

[0081] In summary, the main design concept of the present invention is to geometrically decompose the scanned image of the target object in a panoramic and dynamic scene based on the imaging characteristics of the image scanner, generate multiple image sections, query feature quantities in these image sections, and then reorganize these feature quantities corresponding to the target object to convert them into a complete and linear image, and finally estimate the optical flow of the reorganized image to find the image vanishing point to balance the image color difference. The present invention can effectively identify the waste generated by lubrication in motion, and has a high adaptability to work scenes, and can greatly reduce the image distortion caused by the halo generated by the image scanner, thereby significantly improving the image discrimination or recognition performance of the target object.

[0082] In the embodiments of the present invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. A and B may be singular or plural. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c may represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c may be single or multiple.

[0083] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings, but the above is only a preferred embodiment of the present invention. It should be noted that the technical features involved in the above embodiments and their preferred methods can be reasonably combined and matched into a variety of equivalent schemes by those skilled in the art without departing from or changing the design ideas and technical effects of the present invention; therefore, the present invention is not limited to the scope of implementation shown in the drawings, and all changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the specification and drawings, should be within the protection scope of the present invention.

Claims

1. A method for processing dynamic target scanning image data based on geometric features, characterized in that: include: Obtaining dynamic scanning images of lubrication waste; Based on a preset virtual rectangular frame sampling strategy, the scanned image is geometrically split to obtain multiple image sections, specifically including: presetting the length and width of the image frame in the sampling channel and introducing a geometric correction factor; establishing a virtual rectangular sampling frame, and dividing the image intercepted by the rectangular sampling frame during the scanning process into a plurality of sampling sections; Based on a preset panoramic image reconstruction strategy, the characteristic quantity of the image section is queried, and the target is reorganized according to the characteristic quantity to obtain a complete and linear lubricant waste image, specifically including: counting the image frames of the lubricant waste in each image section in time series; extracting the characteristics of the image section based on a preset boundary sampling function; performing linear fitting based on the characteristics, and reconstructing and generating a panoramic superposition image using a preset back-projection mechanism; Based on a preset optical flow estimation strategy, the optical flow vanishing point causing chromatic aberration in the reorganized lubricating waste image is located, specifically including: determining different sampling areas and the splicing areas between adjacent sampling areas based on the image scanning order; obtaining the position information between the splicing areas through feature matching; performing a similarity transformation according to the change of the position information, and estimating the optical flow value of the lubricating waste during movement through the change of the image frame in the splicing area; obtaining the coordinate value of the corresponding vanishing point through the optical flow value and the expansion center point of the lubricating waste projected to the panoramic three-dimensional coordinate system; wherein the expansion center point is the intersection point formed by the convergence of the sampled optical flows.

2. The method for processing dynamic target scanning image data based on geometric features according to claim 1, characterized in that: The correction factor is used to ensure that the image always maintains a relatively correct position in the rectangular sampling frame.

3. The method for processing dynamic target scanning image data based on geometric features according to claim 1, characterized in that: The back-projection mechanism includes: Obtaining the projection point source of the lubrication waste in the three-dimensional coordinate system relative to the current frame; According to the linear fitting result, the projected linear superposition value is determined; Based on the projected point source and in combination with a preset convolution function, the projected linear superposition value is reconstructed to obtain the convolved linear superposition value and the reconstructed point source of the lubricating waste in the three-dimensional coordinate system.

4. The method for processing dynamic target scanning image data based on geometric features according to claim 1, characterized in that: The motion state of the lubricating waste is a uniform motion with a stable posture within a preset time.

5. The method for processing dynamic target scanning image data based on geometric features according to any one of claims 1 to 4, characterized in that: The acquiring of the dynamic scanning image of the lubricating waste includes: acquiring the scanning image of the lubricating waste in a panoramic and dynamic scene based on a scanning light of a set shape of an image scanning device.

Images