WO2020103075A1 - 一种图像处理方法及装置 - Google Patents

一种图像处理方法及装置

Info

Publication number
WO2020103075A1
WO2020103075A1 PCT/CN2018/116904 CN2018116904W WO2020103075A1 WO 2020103075 A1 WO2020103075 A1 WO 2020103075A1 CN 2018116904 W CN2018116904 W CN 2018116904W WO 2020103075 A1 WO2020103075 A1 WO 2020103075A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
target
porous
small hole
spot
Prior art date
Application number
PCT/CN2018/116904
Other languages
English (en)
French (fr)
Inventor
王曙光
王远靖
Original Assignee
深圳印象认知技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳印象认知技术有限公司 filed Critical 深圳印象认知技术有限公司
Priority to PCT/CN2018/116904 priority Critical patent/WO2020103075A1/zh
Publication of WO2020103075A1 publication Critical patent/WO2020103075A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformation in the plane of the image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformation in the plane of the image
    • G06T3/40—Scaling the whole image or part thereof
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof

Definitions

  • the invention relates to the field of image processing, in particular to a method and device for processing images acquired by a matrix-type small-hole imaging system.
  • the Matrix Pinhole Imaging System is used to acquire images of the surface of objects at close range, such as fingerprint images. It includes an image collector, a light blocking layer is arranged above the image collector, and a plurality of imaging holes are arranged in a matrix on the light blocking layer.
  • the object to be detected is placed above the light blocking layer, and the built-in light source or the external light source in the image collector illuminate the surface of the object to be detected.
  • the principle of small hole imaging as shown in FIG. 19, the light from the surface of the object passes through the small hole, and a matrix-shaped image of small holes reflecting the surface of the object to be detected is formed on the image collector accordingly.
  • the image collected by the matrix-type small-hole imaging system is shown in FIG. 3, and a plurality of small-circular small-hole image spots distributed in a matrix manner on the image.
  • the small hole image spots on the image require further processing to obtain a uniform, distortion-free, continuous and complete target image, because the image collected by the matrix small hole imaging system includes multiple small hole image spots, Each small hole image spot only contains part of the information of the target object; each small hole image spot is an inverted image; because the light passes through a transparent medium layer of different refractive index during the imaging process, the object image in the small hole image spot exists geometrically Distortion: According to the optical principle, there is parallax between multiple images due to multi-eye vision, and there is brightness distortion that gradually decreases in brightness from the center to the outer periphery of the small hole image spot.
  • the present application proposes an image processing method and device to further process the porous images obtained by the matrix small hole imaging system to obtain continuous and complete target images.
  • the purpose of the present application is to provide a method for online processing such as inversion correction and stitching of a porous image obtained by a MAPIS image collector.
  • This application uses the stitching parameters of the matrix small hole imaging system to first perform inversion correction on the target small hole image spot on the target image obtained by the system, and then from the target small hole image spot after the inverted image correction Select an image segment with a size greater than or equal to the maximum non-overlapping image field size, and then stitch the image segments to obtain the target image.
  • correct Distortion correction, brightness correction and other corrections are performed on the target small hole image spot.
  • the method of the present application performs reverse image correction, distortion correction, brightness correction and other processing on the small hole image spot before stitching, so that the stitched result is a complete and intact target image with less distortion and uniform brightness.
  • the present application provides an image processing method, which includes: obtaining a target porous image corresponding to a target by using a matrix porous imaging system, the target porous image containing a plurality of target small hole image spots; based on a matrix porous imaging system Stitching parameters, performing inversion correction on the target small aperture image spot to obtain multiple target imagery small aperture image spots after correction; based on the stitching parameters of the matrix porous imaging system, correcting the inverted image target The small image spots are stitched together to generate the target image.
  • the acquiring the splicing parameters includes: acquiring the calculation parameters; and calculating the splicing parameters using the calculation parameters.
  • the calculation parameters are preset parameters when designing the matrix porous imaging system.
  • the acquiring calculation parameters include: acquiring a standard porous image, the standard porous image including a preset pattern porous image or a surface light source porous image; and calculating the calculation parameter according to the standard porous image.
  • obtaining the preset mode porous image includes: using a matrix-type porous imaging system to obtain a preset mode porous image corresponding to the preset mode, the preset mode porous image containing a plurality of preset mode pores Like spots.
  • acquiring the porous image of the surface light source includes: using a matrix-type porous imaging system to form a porous image of the surface light source under a uniform surface light source luminous condition, the porous image of the surface light source includes a plurality of bright-field pore image .
  • the preset pattern is composed of a single pattern containing lines in two directions or multiple directions, which are regularly and repeatedly arranged according to the cycle length.
  • the period length is a positive integer multiple of the aperture period.
  • using the matrix-type porous imaging system to obtain the preset mode porous image corresponding to the preset mode includes: placing a standard object having the preset mode on the image collector of the matrix-type porous imaging system On the surface, illuminate the standard object with the preset mode with an external light source to make a transmission imaging to obtain the preset mode porous image; or, place no object on the surface of the image collector of the matrix porous imaging system, use An external structure light source with a preset mode illuminates the upper surface of the image collector of the matrix porous imaging system, so that the external structure light source is imaged by the matrix porous imaging system to obtain the preset mode porous image;
  • the standard object with the preset mode is placed on the upper surface of the image collector of the matrix porous imaging system, and the standard object with the preset mode is irradiated with a built-in light source to reflect the image to obtain the porous image with the preset mode
  • no object is placed on the upper surface of the image collector of the matrix porous imaging system
  • a display screen is used as the external light source, the external structure light source with a preset mode, the internal light source, or the internal light source with display function.
  • using the calculation parameters to calculate the stitching parameters includes: using the calculation parameters to calculate the aperture position and the maximum non-overlapping image field of view size; using the maximum non-overlapping image field of view The size calculates the imaging resolution.
  • the calculation of the position of the small hole using the calculation parameters includes: acquiring a porous image of a surface light source using a matrix porous imaging system; and using the porous surface light source on the porous image of the surface light source An arbitrary point on the image is used as the origin to establish a rectangular coordinate system; the position of the small hole is determined, and the position of the small hole is the corresponding coordinate of the geometric center of the bright field small hole image spot in the rectangular coordinate system.
  • the calculating the maximum non-overlapping image field of view size using the calculation parameters includes: measuring the maximum non-overlapping image field size from a preset mode porous image; or, using the The object distance and the image distance in the calculation parameters are calculated to obtain the maximum non-overlapping image field size; or, the stitching effect of the small hole image spot in the preset mode is optimally estimated to obtain the maximum non-overlapping image side field size.
  • inversion correction of the target small hole image spot includes: dividing the target porous image into several sub-pictures according to the position of the small hole, each sub-picture has only and only A complete target small hole image spot; obtaining the position coordinates of the pixel on the target small hole image spot in a rectangular coordinate system with the center of the target small hole image spot as the origin; The position coordinates are symmetrically flipped about the center of the origin to obtain the target small hole image spot after the inverted image correction.
  • the stitching of the image hole of the target object after image correction includes: removing an image segment from the image hole of the target object, the center of the image segment and all The center of the image hole of the target object after image correction is the same, the size of the image segment is the largest non-overlapping image field of view size; according to the relative position of each target image hole in the target porous image Describe the image clip.
  • the stitching of the image hole of the target object after image correction includes: removing an image segment from the image hole of the target object, the center of the image segment and all The center of the target small hole image spot after the image correction is the same, and the size of the image segment is larger than the maximum non-overlapping image field of view size; according to the relative position of each target small hole image spot in the target porous image Describe the image clip.
  • combining the image segments according to the relative positions of the target small hole image spots in the target porous image includes: for information that does not overlap in adjacent image segments, retain the original information; for adjacent The weighted average of the overlapping information in the image fragments or the optimal preservation according to the stitching effect.
  • the calculation of the stitching parameters using the calculation parameters of the matrix porous imaging system further includes: using the calculation parameters to calculate the brightness correction parameters and the distortion correction parameters.
  • calculating the brightness correction parameters using the calculation parameters includes: calculating the brightness correction parameters based on the image distance and the pixel size of the image sensor in the calculation parameters; or, using matrix porous imaging
  • the system acquires a standard porous image; obtains a preset pattern small hole image spot or bright field small hole image spot from the standard porous image; obtains a preset from the preset pattern small hole image spot or bright field small hole image spot Mode pixels; obtaining target pixels from the target small aperture image; measuring the preset mode pixels and the target pixels and performing comparative analysis to obtain brightness correction parameters.
  • the method before merging the image segment according to the relative positions of each target small hole image spot in the target porous image, the method further includes: segmenting the target porous image according to the position of the small hole There are several sub-pictures, and there is only one complete target small hole image spot on each sub-picture; use the brightness correction parameter to adjust the gray value of pixels in the target small hole image spot to eliminate the target in the sub-picture
  • the brightness of the object small hole image spot decreases from the center to the outer periphery, so that the brightness of the target object small hole image spot from the center to the outer periphery in the sub-image is uniform.
  • the calculating the distortion correction parameters using the calculation parameters includes: calculating the distortion correction parameters using the object distance, image distance, and transparent medium refractive index in the calculation parameters; or, using a matrix Standard multi-hole imaging system to obtain standard multi-hole images; from the standard multi-hole images to obtain preset mode small hole image spots or bright field small hole image spots; from the preset mode small hole image spots or bright field small hole image spots Obtaining a preset pattern pixel; acquiring a target pixel from the target object aperture; performing geometric feature matching on the preset pattern pixel and the target pixel to obtain a matching point pair; obtaining a position of the matching point pair ; Analyze and measure the position difference of the pair of matching points to obtain distortion correction parameters for each pixel.
  • the method before merging the image segment according to the relative position of each target small hole image spot in the target porous image, the method further includes: segmenting the target porous image according to the position of the small hole into Several sub-pictures, each of which has only one complete target small hole image spot; use distortion correction parameters to adjust the position of pixels in the target object small hole image spot in the sub-picture to eliminate the target in the sub-picture The geometrical distortion of the image hole of the object from the center to the periphery.
  • a normalization process before stitching the image segments according to the relative positions of the target small hole image spots in the target porous image, a normalization process is further included, and the normalization process includes:
  • the target image of is normalized by normalization of brightness, contrast and imaging resolution, so that the average brightness and contrast variance of the target image are within a preset range, so that the imaging resolution is a standard value.
  • the method before stitching the target small hole image spot after the inverted image correction based on the stitching parameters of the matrix-type porous imaging system, the method further includes: a three-dimensional reconstruction process based on a multi-eye vision method.
  • the 3D reconstruction process includes recovering the depth information of each point on the target according to the arrangement information of the small holes and the parallax.
  • the present application also provides an image processing device including: a target porous image acquisition unit for acquiring a target porous image corresponding to a target using a matrix porous imaging system, the target porous image containing a plurality of target pores Image spot; the target image correction unit, which is used to perform inverted image correction on the target object small hole image spot in the porous image of the target based on the stitching parameters of the matrix porous imaging system, to obtain multiple inverted image corrected targets Small hole image spot; target image stitching unit, used to stitch the small hole image spot of the target object after the inverted image correction based on the stitching parameters of the matrix porous imaging system to generate the target image.
  • a target porous image acquisition unit for acquiring a target porous image corresponding to a target using a matrix porous imaging system, the target porous image containing a plurality of target pores Image spot
  • the target image correction unit which is used to perform inverted image correction on the target object small hole image spot in the porous image of the target based on the stitching parameters of the
  • Figure 1 is a schematic diagram of the imaging principle of a matrix-type small-hole imaging system
  • Figure 2 is the complete target image after stitching
  • Figure 3 is the porous image of the target obtained by MAPIS
  • FIG. 4 Schematic diagram of imaging principle using external light source and preset mode
  • FIG. 1 Schematic diagram of imaging principle using external structure light source
  • Figure 6 Schematic diagram of imaging principle using built-in light source and preset mode
  • Figure 7 Schematic diagram of imaging principle using built-in light source with display function
  • Figure 8 Example of display mode of built-in light source
  • Figure 9 An example of a bright spot image for scaling the center position of a hole
  • Figure 13 uses Figure 3 as the input image and the image obtained by stitching with incorrect Si parameters
  • Figure 14 Schematic diagram of brightness attenuation caused by small hole imaging
  • 15 is a schematic diagram of geometric distortion caused by imaging through a dielectric layer with different refractive indexes
  • Figure 16 Schematic diagram of the principle of geometric distortion in the MAPIS optical path
  • 18b is a flowchart of inversion correction of the target small hole image spot in the target porous image
  • Fig. 18c is a flow chart of the small hole image spot of the target after the stitching inverted image is corrected
  • FIG. 18d is a flowchart of another target porous image processing process provided in this embodiment (where the process in the dotted frame is an optional step);
  • FIG. 19 Schematic diagram of the generation of inverted images in small-hole imaging
  • Figure 20 Schematic diagram of the inverted image correction method of the small hole image
  • Figure 21 Schematic diagram of the field of view and overlapping area on the object surface when imaging in multiple holes
  • Figure 22 Porous image of the target after brightness correction (the original image is Figure 3);
  • FIG 23 The porous image of the target after the geometric distortion correction (the original image is Figure 3);
  • Figure 24 The porous image of the target after the inverted image correction (the original image is Figure 3);
  • FIG 25 The stitched image after image enhancement (the original image is Figure 3);
  • 26 is a schematic structural diagram of a porous image processing device provided in this embodiment.
  • FIG. 27 is a schematic structural diagram of a reverse image correction module 200 according to this embodiment.
  • FIG. 28 is a schematic structural diagram of a mosaic parameter acquisition module 400 according to this embodiment.
  • FIG. 1 is an imaging principle diagram of the MAPIS porous image in this embodiment.
  • the matrix porous imaging system includes a light blocking layer 3 in which a plurality of imaging holes 2 are arranged in a matrix.
  • Figure 1 shows two adjacent imaging holes as an example.
  • the light of the target object on the object plane 1 irradiates the light blocking layer 3, part of the light is blocked outside the light blocking layer 3, and the other part of the light irradiates the image plane 4 (Image Plane) through the imaging hole 2
  • Image Plane image plane 4
  • the matrix-type porous imaging system is used for image sampling, and a porous image with a plurality of small-hole image spots arranged in a matrix is obtained.
  • This embodiment uses the image obtained by the matrix porous imaging system as an example to illustrate the technical solution of the present application.
  • FIG. 3 is a porous image of a target acquired by the matrix porous imaging system shown in FIG. 1, which is a porous image of a fingerprint.
  • a plurality of circular small-hole image spots are arranged in a matrix.
  • Each small hole image spot can reflect a part of the object field of view, but not the entire object side field of view.
  • the method before processing the porous image of the target object, the method further includes the step of acquiring the stitching parameters of the matrix porous imaging system.
  • the method of acquiring the stitching parameters includes: acquiring the calculation parameters; using The calculation parameter calculates the splicing parameter.
  • the calculation parameter is a preset parameter when designing the matrix porous imaging system, or the calculation parameter is acquired by a method including the following steps, and acquiring the calculation parameter includes: acquiring a standard porous image, the standard porous image includes Preset mode porous image or surface light source porous image; calculate the calculation parameters according to the standard porous image.
  • the calculation parameters include the position of the small hole, the object distance, the image distance, the pixel size of the image sensor, the distance between adjacent small holes, and the refractive index of the transparent medium.
  • obtaining a preset mode porous image includes: obtaining a preset mode porous image corresponding to the preset mode by using a matrix-type porous imaging system, the preset mode porous image containing a plurality of preset mode pore images.
  • the preset pattern is composed of a single pattern including lines in two directions or multiple directions, which are regularly and repeatedly arranged according to a period length, and the period length is a positive integer multiple of the aperture period.
  • using the matrix-type porous imaging system to obtain the preset mode porous image corresponding to the preset mode includes: placing a standard object having the preset mode on the image collector of the matrix-type porous imaging system On the surface, illuminate the standard object with the preset mode with an external light source to make a transmission imaging to obtain the preset mode porous image; or, place no object on the surface of the image collector of the matrix porous imaging system, use The external structure light source with a preset mode illuminates the upper surface of the image collector of the matrix porous imaging system, and the external structure light source with a preset mode is imaged by the matrix porous imaging system to obtain the pre Set the mode porous image; or, place the standard object with the preset mode on the upper surface of the image collector of the matrix porous imaging system, and use the built-in light source to illuminate the standard object with the preset mode for reflection imaging to obtain the Preset mode porous image; or, no object is placed on the upper surface of the image collector of the matrix porous imaging system, so that the external structure light source with
  • a display screen is used as the external light source, the external structure light source with a preset mode, the internal light source, or the internal light source with display function.
  • the use of a matrix porous imaging system to obtain a preset mode porous image corresponding to a preset mode includes: using an external light source and a preset mode imaging, using an external structure light source with a preset mode for imaging, using a built-in light source and a preset mode Imaging, or using the built-in light source with display function.
  • the imaging using the external light source and the preset mode includes:
  • the external standard light source 13 is used to irradiate the first standard object 12 having a preset pattern to make a transmission image to obtain an image including a transmission image spot, and the image including the transmission image spot is used as the preset pattern porous image.
  • the preset mode refers to a specific figure.
  • the standard object with the preset mode is a standard object that forms a specific pattern on the surface through a certain process, such as evaporation, etching, printing, etc., such as a film with a specific geometric pattern printed, and a specific geometric figure etched Glass sheet.
  • the external light source is a general point light source or a surface light source.
  • the imaging using an external structure light source with a preset mode includes:
  • the external structure light source 14 with a preset mode is a point light source or a surface light source that can generate a preset mode.
  • the external structure light source with a preset mode is imaged by the matrix-type porous imaging system to output the preset mode porous image.
  • FIG. 6 is a schematic diagram of imaging using a built-in light source and a preset mode.
  • the imaging using the built-in light source and the preset mode includes:
  • the second standard 16 with a preset pattern is placed on the surface of the image collector 15 with a built-in light source in the matrix porous imaging system.
  • the standard object with a preset mode is irradiated with a built-in light source to reflect and image, and a porous image of the preset mode is output.
  • the built-in light source is a point light source or a surface light source provided inside an image collector of a matrix porous imaging system .
  • the imaging using the built-in light source with display function includes:
  • No objects are placed on the surface of the image collector 17 with a built-in light source of the matrix porous imaging system, so that the built-in light source with a display function emits light reflecting a preset pattern through the surface of the image collector of the matrix porous imaging system , Output the porous image in the preset mode.
  • the built-in light source with display function refers to a light source that can be displayed in a preset mode and placed inside the MAPIS collector.
  • the built-in light source with display function may be a mobile phone display screen.
  • FIG. 8 is an example of a display mode of a built-in light source with a display function.
  • the preset mode displayed by the built-in light source with display function is set to the regular "Tian" character.
  • the true pinhole position of all imaging pinholes may have an overall translation or rotation error, so the pinhole position in the stitching parameters can be used to design the matrix porous imaging
  • the parameters preset in the system can also be determined by calculation using calculation parameters.
  • using the calculation parameters to calculate the stitching parameters includes: using the calculation parameters to calculate the aperture position and the maximum non-overlapping image field size; calculating the imaging resolution using the maximum non-overlapping image field size rate.
  • FIG. 9 is an example of a multi-hole image of a surface light source containing a plurality of bright-field small-hole image spots.
  • the calculation of the position of the small hole using the calculation parameters includes: obtaining a porous image of a surface light source using a matrix porous imaging system; and using the porous image of the surface light source on the porous image of the surface light source An arbitrary point is used to establish a rectangular coordinate system for the origin; the position of the small hole is determined, and the position of the small hole is the corresponding coordinate of the geometric center of the bright field small hole image spot in the rectangular coordinate system.
  • a rectangular coordinate system is established by using the vertex at the upper left corner of the porous image of the surface light source as an origin, so as to visually represent the position of each imaging hole.
  • a solid circle can be used to fit the image spot of each imaging aperture, and the center of the solid circle is used as the geometric center of the image aperture of the aperture to obtain the position of the aperture.
  • acquiring the porous image of the surface light source includes: forming a porous image of the surface light source under a uniform surface light source lighting condition using a matrix-type porous imaging system, the porous image of the surface light source containing a plurality of bright field small hole image spots.
  • each small hole image spot 10 on the image plane 4 is defined as a field of view on image plan (FOVI).
  • FOVI field of view on image plan
  • Each small hole image spot 10 corresponds to an area on the object plane, and is referred to as a field of view on object plane 6 (FOVO).
  • FOVO field of view on object plane 6
  • the object-side field of view 6 of adjacent imaging apertures overlaps, that is, the information on the image spots of adjacent apertures is redundant, as shown in FIG. 21.
  • the entire object-side field of view is evenly divided according to the imaging aperture, so that each point in the object-side field of view is assigned to the closest imaging aperture, and the object-side field of view area of the imaging aperture obtained in this way is divided defined as the maximum object side non-overlapping fields of view 7 (maximum non-overlapping FOV on Object Plane, MNFOVO), which is referred to as a size S o.
  • the image field of view corresponding to each maximum non-overlapping object field of view 7 is defined as the maximum non-overlapping image field of view 9 (Maximum Non-overlapping FOV on Image Plane, MNFOVI), and its size is denoted as S i .
  • Figure 1 shows the relationship between them.
  • MAPIS products calibrate the S i value when they leave the factory.
  • the image distance ID does not change, but some user operations will change the object distance OD.
  • the value of S i needs to be recalibrated.
  • MAPIS is combined with a mobile phone screen to form an image acquisition device, or in the case of mounting a MAPIS image acquisition device under the mobile phone screen, if the user sticks a film on the surface of the mobile phone screen, the object distance will be changed, which will change the S i value of the system .
  • the calculation of the maximum non-overlapping image field size using the calculation parameters includes: measuring the maximum non-overlapping image field size from a preset mode porous image; or, using the calculation parameters The object distance and the image distance are calculated to obtain the maximum non-overlapping image field size; or, the stitching effect of the small hole image spot in the preset mode is optimally estimated to obtain the maximum non-overlapping image field size.
  • the maximum non-overlapping view fields image size S i includes a preset mode in a porous measurement image from:
  • the preset pattern may be composed of a single pattern containing lines in two directions or multiple directions, which are regularly and repeatedly arranged according to the cycle length, as shown in FIG. 10, FIG. 11, and FIG. 12.
  • the period length is a positive integer multiple of the aperture period 8, for example, in FIG. 10, the period length is the distance between the geometric centers of two adjacent squares, and the period length is positive of the aperture period 8 Integer multiple.
  • the object distance OD and image distance ID are expressed by the following formula (2) and formula (3),
  • ID ID 0 + ⁇ ID formula (2)
  • ID 0 represents the preset image distance value
  • OD 0 represents the preset object distance value
  • ⁇ ID is the processing error of the image distance in the production process
  • ⁇ OD is the processing error of the object distance in the production process
  • ⁇ OD is the object The amount of change after leaving the factory.
  • the object distance OD and image distance ID of MAPIS can be accurately known, or the design parameters of ID 0 and OD 0 and the change of object distance after delivery ⁇ OD are known , and the processing errors ⁇ ID and ⁇ OD can be ignored, then the calculating a first preset rule S i.
  • the maximum non-overlapping image field size is calculated using the object distance and the image distance in the calculation parameters, specifically calculated according to the following formula (4):
  • the value range of the object distance change ⁇ OD and processing error ⁇ ID and ⁇ OD after leaving the factory is can be more accurately known, it can be estimated by S i optimal splicing.
  • S i optimal splicing I.e., the actual object (such as a fingerprint) imaging, the contrast will be obvious to try every possible value S i splicing, splicing evaluate the effects, the selection of the optimum value, i.e., the preset pattern image patch apertures
  • the optimal stitching effect is estimated to obtain the largest non-overlapping image field size.
  • the possible value of S i is determined according to the value range of the object distance change ⁇ OD and the processing error ⁇ ID and ⁇ OD .
  • the imaging resolution of the MAPIS is obtained according to the following formula (5) using the maximum non-overlapping image-side field size.
  • R is the imaging resolution of the MAPIS
  • S p is the pixel size of the image sensor
  • S p is a parameter preset during the design of the matrix porous imaging system.
  • the calculation of the splicing parameters using the calculation parameters of the matrix-type porous imaging system further includes: calculating the brightness correction parameters and distortion correction parameters using the calculation parameters.
  • the small hole imaging itself will cause the brightness of each small hole image spot to decay from the center to the periphery
  • the brightness of each small hole image spot from the center to the periphery of the porous image collected by the MAPIS image acquisition device is inconsistent, as shown in the figure 14 shown. If the brightness of the porous image is not corrected, the target image obtained by stitching will have the problem of uneven brightness.
  • the method before merging the image segment according to the relative position of each target small hole image spot in the target porous image, the method further includes: dividing the target porous image into several sub-pictures according to the position of the small hole, There is only one complete target small hole image spot on each sub-picture; use the brightness correction parameters to adjust the gray value of pixels in the target small hole image spot to eliminate the target small hole image spot in the sub-picture
  • the brightness from the center to the periphery is attenuated, so that the brightness of the target small hole image spot in the sub-picture is uniform from the center to the periphery.
  • the calculation of the brightness correction parameters using the calculation parameters includes:
  • the brightness correction parameter is calculated according to the image distance in the calculation parameters and the pixel size of the image sensor, specifically,
  • E x represents the illuminance at a point X on the small hole image spot with a distance of x from the center of the small hole
  • E o represents the illuminance at the center of the small hole image spot
  • ⁇ is the line connecting the X point to the center of the small hole and the small hole light
  • the included angle of the axis that is, ⁇ 2 in FIG. 16, the ⁇ of each point on the small hole image can be different from the object point OD, the image distance ID, the refractive index n 1 , n 2 of the transparent medium, and the current point on the image plane.
  • the distance y i of the hole center is calculated.
  • 1 / cos 4 ⁇ be the brightness correction parameter. Therefore, the brightness after X point correction is calculated according to formula (7):
  • I ′ x is the brightness after correction
  • I x is the brightness before correction
  • the calculation of the brightness correction parameters using the calculation parameters includes: acquiring a standard porous image using a matrix porous imaging system; acquiring a small hole image spot of a preset pattern or a bright field from the standard porous image; Obtaining the preset mode pixels from the preset mode small hole image spot or bright field small hole image spot; obtaining the target object pixel from the target object small hole image spot; measuring the preset mode pixel and the target Object pixels are compared and analyzed to obtain brightness correction parameters.
  • I ′ x is the brightness after correction
  • I x is the brightness before correction
  • T o is the brightness value at the center of the brightness attenuation template
  • T x is the brightness value at point X in the brightness attenuation template.
  • the luminance correction parameter becomes T o / T x.
  • FIG. 15 is a small hole image spot obtained by using a preset pattern without distortion correction.
  • the MAPIS image acquisition device When the MAPIS image acquisition device is working, the light reaches the image sensor from the surface of the object through the small hole. Due to the penetration of the dielectric layer with different refractive index, there is a certain geometric distortion of the image, as shown in Figure 15, if the distortion is not corrected, The stitched image will have a certain degree of distortion and aliasing.
  • the method before merging the image segment according to the relative position of each target small hole image spot in the target porous image, the method further includes: segmenting the target porous image into several sub-pictures according to the position of the small hole, each There is one and only one complete target hole image spot on each sub-picture; use distortion correction parameters to adjust the position of pixels in the target object hole image spot in the sub-picture to eliminate the target object hole image spot in the sub-picture Geometric distortion from center to periphery.
  • the calculation of the distortion correction parameters using the calculation parameters includes: obtaining a standard porous image with a matrix porous imaging system; obtaining a small hole image spot of a preset pattern from the standard porous image or Bright field small hole image spots; obtaining preset mode pixels from the preset mode small hole image spots or bright field small hole image spots; obtaining target object pixels from the target object small hole image spots; It is assumed that the pattern pixels and the target object pixels are geometrically matched to obtain matching point pairs; the positions of the matching point pairs are obtained; the position differences of the matching point pairs are analyzed and measured to obtain distortion correction parameters for each pixel.
  • FIG. 16 is a schematic diagram of small-hole image spot distortion.
  • the transparent medium I from the object plane 1 to the light blocking layer 3 has a refractive index of n 1
  • the transparent medium II and the refractive index is n 2
  • y o is the distance from the object side point to the center of the small hole
  • y i is the distance from the image side point to the center of the small hole
  • OD and ID are the object distance and image distance of MAPIS
  • ⁇ 1 represents the incident angle
  • ⁇ 2 represents Exit angle.
  • Formula (12) can be obtained from Formula (9), Formula (10) and Formula (11):
  • y i does not change linearly with y o . Therefore, when the object square point is imaged through the small hole at different positions from the center of the small hole, the magnification is different and there is distortion.
  • Fig. 17 shows the change curve of y i with y o . As shown in Fig. 17, when n 1 < n 2 , the correction parameters of each position on the small hole image can be estimated. which is
  • (x i , y i ) is the coordinate before correction of a certain point on the small hole image spot
  • (x ′ i , y ′ i ) is the corrected coordinate of that point
  • C x, y is the correction parameter, which is related to the distance from the point (x i , y i ) to the center of the small hole.
  • the calculating the distortion correction parameters using the calculation parameters includes: calculating the distortion correction parameters using the object distance, the image distance, and the refractive index of the transparent medium in the calculation parameters, specifically, calculating according to formula (14). If the structure of the MAPIS system does not conform to FIG. 16, the formula (14) needs to be adjusted according to the actual structure.
  • the matrix porous imaging system is used to obtain a preset pattern porous image, and the pattern image contained in each target small hole image spot in the target porous image is compared with the preset pattern image, and the difference is analyzed and measured to obtain the value of each point.
  • the distortion correction parameters it may be assumed that a target small hole image spot is obtained as shown in FIG. 15, and the preset mode small hole image spot is shown in FIG. 12.
  • a point-by-point correspondence can be obtained; for regions lacking geometric features, the correspondence can be obtained by interpolation.
  • FIG. 18a is a flowchart of a method for processing a porous image of a target obtained by MAPIS in this embodiment. The method of this embodiment will be described with reference to FIG. 18a.
  • the porous image of the target refers to a porous image of the target obtained by using MAPIS
  • the porous image of the preset mode refers to a porous image of a standard object having a preset mode obtained by using MAPIS.
  • S100 uses a matrix-type porous imaging system to acquire a target porous image corresponding to the target, and the target porous image contains a plurality of target small hole image spots.
  • the inverted image correction may be an inverted image correction of the target small hole image spot in the target porous image, or a part of the target small hole image spot in the target porous image Perform reverse image correction.
  • S300 stitches the target small hole image spot after the inverted image correction to generate the target image.
  • the stitching may be stitching of the target small hole image spots after the inverted image correction, or may be stitching on a part of the target small hole image spots after the inverted image correction.
  • the target object hole image spot may have any shape.
  • the small hole image spot is circular or square.
  • One is to facilitate the manufacture of matrix-type porous imaging systems, and the other is that regular shaped spots are easier to perform operations such as image segmentation, inverted image correction, and stitching.
  • FIG. 18b is a flowchart of image inversion correction on the target small hole image spot in the target porous image.
  • image inversion correction on the small hole image spot in the porous image includes:
  • S201 divides the porous image of the target into several sub-pictures according to the positions of the small holes, and each sub-picture has one and only one complete image spot of the small hole of the target.
  • S202 Acquire position coordinates of pixels on the target small hole image spot in the rectangular coordinate system with the center of the target small hole image spot as the origin.
  • a rectangular coordinate system is established with the center of the target small hole image spot as the origin, and the position of each pixel is represented by its corresponding position coordinate in the rectangular coordinate system.
  • the position coordinates of the pixel are flipped symmetrically about the center of the origin to obtain the target small hole image spot after the inverted image correction.
  • the pixels on the target small hole image spot are symmetrically reversed about the center of the origin to obtain the target small hole image spot after the inverted image correction.
  • FIG. 20 is a schematic diagram of center-symmetrical flipping in this embodiment.
  • a rectangular coordinate system is established with the center of the target small hole image spot as the origin, and each pixel corresponds to a position coordinate (x, y) and a gray value i;
  • the position coordinates corresponding to the pixels are flipped symmetrically with respect to the center of the origin, that is, (x, y, i) becomes (-x, -y, i), and the target small hole image spot after the inverted image correction is obtained.
  • the inverted porous image of the target obtained by using FIG. 3 as the original image is shown in FIG. 24.
  • FIG. 18c is a flow of stitching the target small hole image spots after the inverted image correction in a achievable manner.
  • stitching the target small hole image spots after the inverted image correction includes:
  • S301 Take an image segment from the target small hole image spot, the center of the image segment is the same as the center of the target small hole image spot after the inverted image correction, and the size of the image segment is the largest without overlap Image field size; specifically, taking the geometric center of the target small hole image spot corrected for each inverted image as the center, taking the size from the target small hole image spot exactly equal to the matrix porous imaging
  • the image segment of the system's largest non-overlapping image-side field size S i The image segment of the system's largest non-overlapping image-side field size S i .
  • the image segments are combined according to the relative positions of the target small hole image spots in the target porous image to generate a complete target image.
  • the stitching of the target small hole image spot corrected by the inverted image in S300 includes:
  • An image segment is taken from the target small hole image spot, the center of the image segment is the same as the center of the target small hole image spot after the inverted image correction, and the size of the image segment is larger than the largest non-overlapping image
  • the size of the square field of view specifically, taking the center of each small hole image spot corrected by the inverted image as the center, the size of the largest non-overlapping image square view of the matrix aperture imaging system is taken from the small hole image spot Field-sized image segments, so that there is information overlap in the peripheral part of adjacent image segments, the information refers to the image information on the image segments; according to the relative position of each target small hole image spot in the target porous image Image fragment.
  • the stitching the image segment according to the relative position of each target small hole image spot in the target porous image includes: for information that does not overlap in adjacent image segments, retain the original information;
  • the overlapping information in adjacent image segments is weighted average or optimally retained.
  • the optimal information is retained according to a preset rule.
  • the preset rule may be to preferentially select the point with the highest contrast with its neighborhood or to select the highest brightness. Point.
  • FIG. 18d is a flowchart of another porous image processing method provided in this embodiment.
  • the method includes: small-hole inverted image correction and image stitching, and the small-hole inverted image correction refers to the target small hole Invert image correction for speckles, so that the stitching of your double eyelid images refers to the image stitching of the target small hole image spots after the inverted image correction.
  • the correction of the inverted image of the small hole further includes one or more steps of image segmentation at the small hole level, correction of the brightness distortion of the small hole, correction of the geometric distortion of the small hole, three-dimensional reconstruction of the small hole and depth calculation.
  • stitching the target small hole image spot after the inverted image correction, after generating the target image may also include stitching image enhancement and segmentation and image normalization One or more steps.
  • the present inventors have found that, if taken as the only spot size of each object in the hole exactly image segment S i, the target image mosaic obtained easily generated blockiness where image spots of adjacent target was contact holes. Therefore, if the acquisition range of each target small hole image spot is extended to make the size of the obtained image segment larger than S i , it will make the adjacent image segments have partial overlap information on the content, and then overlap
  • the weighted average of information or the optimal retention according to preset rules can make the transition of the target image content more smooth and natural, and avoid the blockiness in the stitching process.
  • the image processing method further includes a brightness correction process according to formula (7) or (8), so that each target object pore image spot in the target porous image is from the center to the periphery The brightness attenuation is corrected to the uniform brightness of each individual small hole image spot.
  • the brightness correction is performed before dividing the target porous image into several sub-pictures.
  • the image processing method further includes a distortion correction process according to formula (13), formula (14), or (15), thereby eliminating each target pore in the target porous image The geometric distortion of the image spot from the center to the periphery.
  • the distortion correction is performed after brightness correction, before segmenting the porous image of the target into several sub-pictures.
  • FIG. 1 As the original target porous image, the brightness of the target porous image is corrected first, and the result is shown in FIG. 22, and then the distortion correction is performed on the target porous image after brightness correction, and the result is shown in FIG.
  • the target porous image processing method further includes image enhancement processing and image segmentation processing, so that the target image is easy to recognize.
  • the image enhancement processing may use a histogram equalization method.
  • FIG. 25 Exemplarily, the image obtained by image enhancement using FIG. 3 as the original image is shown in FIG. 25.
  • the image segmentation process can eliminate the interference of the environment on the imaging of the target object.
  • the image processing method further includes normalization processing.
  • the normalization process is to normalize the brightness, contrast and imaging resolution of the target image respectively, so that the average brightness and contrast variance of the target image are within a preset range, and the imaging resolution is a preset standard value.
  • the normalized target image can facilitate subsequent identification or other applications.
  • the gray scale and the variance of the gray scale are normalized according to the following formula (16):
  • I ′ (x, y) (I (x, y) - ⁇ ) * ⁇ ideal / ⁇ + ⁇ ideal formula (16)
  • ⁇ and ⁇ are the grayscale mean and grayscale variance of the target image before normalization, and ⁇ ideal and ⁇ ideal are preset grayscale mean and grayscale variance.
  • I (x, y) is the gray value of the (x, y) point on the target image before normalization
  • I ′ (x, y) is the gray of the (x, y) point on the target image after normalization Degree value. Since gray scale corresponds to brightness, the variance of gray scale corresponds to contrast, so the above formula normalizes brightness and contrast.
  • the imaging resolution is standardized according to the following formula (17):
  • R is the imaging resolution of the target image before normalization
  • R ideal is the standard value of the preset imaging resolution
  • the image processing method further includes a three-dimensional reconstruction process based on a multi-eye vision method before segmenting the target multi-hole image into several sub-pictures according to the positions of small holes.
  • a three-dimensional reconstruction process based on a multi-eye vision method includes recovering the depth information of each point on the target according to the arrangement information of the small holes and the parallax.
  • a point on the target will form multiple images with parallax through multiple imaging holes, using the information of these multiple images and the object distance, image distance, hole spacing of the matrix porous imaging system, etc.
  • the parameter recovers the depth information of the points on the target, and normalizes and outputs the depth information of all points on the target to obtain a complete target image.
  • FIG. 26 is a schematic structural diagram of the porous image processing device provided in this embodiment.
  • the device includes: a target porous image acquisition module 100 for acquiring a target porous image corresponding to a target using a matrix porous imaging system, the target porous image containing a plurality of target pore image spots
  • Inverted image correction module 200 which is used to perform inverted image correction on the target small aperture image spot based on the stitching parameters of the matrix-type porous imaging system, to obtain multiple inverted image corrected target small aperture image spots; the target image
  • the generating module 300 is used for stitching the target small hole image spot after the inverted image correction based on the stitching parameters of the matrix porous imaging system to generate the target image.
  • FIG. 27 is a schematic structural diagram of an inverted image correction module 200.
  • the inverted image correction module 200 includes: an image segmentation unit 201, which is used to divide the target object according to the position of the small hole The porous image is divided into several sub-pictures, and each sub-picture includes a complete small hole image spot; the pixel determining unit 202 is used to establish a rectangular coordinate system with the center of the small hole image spot as the origin on each sub-image, Obtain the position and gray value of each pixel, and the position of each pixel is represented by its corresponding position coordinate in the rectangular coordinate system; the inversion correction unit 203 is used to make the pixel on the small hole image spot The center of the origin is symmetrically reversed to obtain the target small hole image spot after the inverted image correction.
  • the target image generation module 300 includes: an image segment interception unit for extracting an image segment from the target object aperture image spot, the center of the image segment and the inverted The center of the small hole image spot of the target after image correction is the same, and the size of the image segment is greater than or equal to the maximum non-overlapping image field size; The relative position of the image fragments is combined to generate a complete target image. If there is information overlap between adjacent image fragments, the original information is retained for the information that does not overlap in the adjacent image fragments, and for the information that overlaps in the adjacent image fragments, Do the weighted average or keep the best. In particular, keep the best according to a preset rule.
  • the preset rule may be to preferentially select the point with the highest contrast with its neighborhood or preferentially select the point with the highest brightness.
  • the device further includes a splicing parameter acquisition module 400 for acquiring splicing parameters.
  • a splicing parameter acquisition module 400 for acquiring splicing parameters.
  • FIG. 28 is a schematic structural diagram of the stitching parameter acquisition module 400.
  • the stitching parameter acquisition module 400 includes a calculation parameter acquisition unit 401 for acquiring calculation parameters; stitching parameter measurement Unit 402 is configured to calculate the splicing parameters using the calculation parameters.
  • the calculation parameter acquisition unit 401 includes a standard porous image acquisition subunit for acquiring a standard porous image, and the standard porous image includes a preset pattern porous image or a surface light source porous image; a calculation parameter measurement subunit is used for calculating Standard multi-hole images calculate the calculation parameters.
  • the standard porous image acquisition sub-unit includes a preset mode porous image acquisition slave unit for acquiring a preset mode porous image corresponding to a preset mode using a matrix porous imaging system, the preset mode porous image contains multiple presets Mode small hole image spot; surface light source porous image acquisition slave unit for forming a surface light source porous image using a matrix type porous imaging system under uniform surface light source lighting conditions, the surface light source porous image containing a plurality of bright field small hole image spots .
  • the preset mode porous image acquisition slave unit includes at least one of the following preset mode porous image acquirers:
  • the first preset mode porous image acquirer is used to place the standard object with the preset mode on the upper surface of the image collector of the matrix porous imaging system, and irradiate the standard object with the preset mode with an external light source to make it Transmission imaging to obtain the porous image in the preset mode;
  • the second preset mode porous image acquirer is used to irradiate the image acquisition of the matrix porous imaging system with an external structure light source having a preset mode without placing any object on the upper surface of the image collector of the matrix porous imaging system The upper surface of the device, so that the external structure light source with a preset mode is imaged by the matrix porous imaging system to obtain the preset mode porous image;
  • the third preset mode porous image acquirer is used to place the standard object with the preset mode on the upper surface of the image collector of the matrix porous imaging system, and irradiate the standard object with the preset mode with a built-in light source to make it reflect Imaging to obtain the porous image in the preset mode;
  • the fourth preset mode porous image acquirer is used to place no object on the upper surface of the image collector of the matrix porous imaging system, so that the built-in light source with a display function emits light with a preset mode to illuminate the matrix porous imaging
  • the upper surface of the image collector of the system is reflected and imaged to obtain the porous image in the preset mode.
  • the stitching parameter measurement unit 402 includes: a small hole position acquisition subunit, a maximum non-overlapping image side field size acquisition subunit, and an imaging resolution acquisition subunit.
  • the small hole position acquisition subunit includes: a surface light source porous image acquisition slave unit for acquiring a surface light source porous image using a matrix-type porous imaging system; a rectangular coordinate system establishing a slave unit for use in the surface light source porous image, A rectangular coordinate system is established by taking any point on the porous image of the surface light source as the origin; the pinhole position determining slave unit is used to determine the pinhole position, and the pinhole position is the geometric center of the bright field pinhole image spot The corresponding coordinates in the rectangular coordinate system are described.
  • the maximum non-overlapping image side field size acquisition subunit includes at least one of the following slave units: a maximum non-overlapping image side field size measurement slave unit for measuring the maximum non-overlapping image from a preset mode porous image Square field of view size; maximum non-overlapping image field field size calculation slave unit for calculating the maximum non-overlapping image field field size using the object distance and image distance in the calculation parameters; maximum non-overlapping image field field size
  • the optimal estimation slave unit is used for optimally estimating the stitching effect of the small hole image spots in the preset mode to obtain the maximum non-overlapping image field size.
  • the splicing parameter measurement unit 402 further includes a brightness correction parameter acquisition sub-unit, which is used to obtain brightness correction parameters and distortion correction parameter acquisition sub-units, which are used to obtain distortion correction parameters.
  • the device further includes a normalization processing module 500, which is used to normalize the brightness, contrast and imaging resolution of the stitched target image, so that all The average brightness and contrast variance of the target image are within a preset range, so that the imaging resolution is a standard value.
  • a normalization processing module 500 which is used to normalize the brightness, contrast and imaging resolution of the stitched target image, so that all The average brightness and contrast variance of the target image are within a preset range, so that the imaging resolution is a standard value.
  • the device further includes a three-dimensional reconstruction module 600 based on a multi-eye vision method, which is used to recover the depth of each point on the target according to the arrangement information of the small holes and the parallax information.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Processing (AREA)

Abstract

本申请提供一种图像处理方法及装置，所述图像由矩阵式多孔成像系统获得，包括多个目标物小孔像斑。该方法首先对目标物小孔像斑进行倒像纠正，再从倒像纠正后的目标物小孔像斑中取出图像片段，再对所述图像片段进行拼接，从而由目标物多孔图像获得能够真实反映目标物外貌的目标图像。本申请还提供了一种用于处理多孔图像的装置。所述装置能够应用所述图像处理方法将所述目标物多孔图像处理为能够真实反映目标物表面形貌的目标图像。

Description

一种图像处理方法及装置 技术领域
本发明涉及图像处理领域，特别涉及一种处理矩阵式小孔成像系统所获取图像的方法及装置。
背景技术
矩阵式小孔成像系统(Matrix Pinhole Imaging System，MAPIS)用于近距离采集物体表面图像，如指纹图像等。其包括图像采集器，在图像采集器的上方设置有阻光层，在所述阻光层上矩阵式排布有多个成像小孔。待检测物体被放置于阻光层上方，图像采集器中的内置光源或者外置光源对待检测物体的表面进行照射。根据小孔成像原理，如图19所示，来自物体表面的光线经过小孔，在图像采集器上相应地形成矩阵式排布的反映待检测物体表面的小孔像斑图像。利用矩阵式小孔成像系统采集到的图像如图3所示，在所述图像上矩阵式分布有近似圆形的多个小孔像斑。
然而，所述图像上的小孔像斑需要进一步的处理才能获得亮度均匀、无畸变、连续完整的目标图像，原因在于：矩阵式小孔成像系统采集到的图像包括多个小孔像斑，每个小孔像斑仅包含目标物的部分信息；每个小孔像斑为倒立的像；由于成像过程中光线经过不同折射率的透明介质层，使得小孔像斑中的物像存在几何畸变；根据光学原理可知，由于多目视觉造成多个像之间存在视差，小孔像斑存在从中心向外周亮度逐渐降低的亮度畸变。
发明内容
本申请提出一种图像处理方法和装置，以进一步处理由矩阵式小孔成像系统获得的多孔图像，获得连续完整的目标图像。
本申请的目的是，提供一种对MAPIS图像采集器获得的多孔图像进行倒像纠正和拼接等在线处理的方法。本申请利用矩阵式小孔成像系统的拼接参数，首先对由该系统获得的目标物图像上的目标物小孔像斑进行倒像纠正，再从倒像纠正后的目标物小孔像斑中选取尺寸为大于或者等于最大无重叠像方视场尺寸的图像片段，再对所述图像片段进行拼接，获得目标图像，可选地，在对目标物小孔像斑进行倒像纠正前，对所述目标物小孔像斑进行畸变校正、亮度校正等校正。本申请的方法通过在拼接前对小孔像斑进行倒像纠正、畸变校正、亮度校正等处理，使得拼接所得为畸变较小、亮度均匀的完整无缺的目标图像。
本申请提供一种图像处理方法，包括：利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑；基于矩阵式多孔 成像系统的拼接参数，对所述目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑；基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。
在一种可能的实现方式中，在对所述目标物小孔像斑进行倒像纠正之前还包括获取拼接参数。
在一种可能的实现方式中，所述获取拼接参数包括：获取计算参数；使用所述计算参数计算得到拼接参数。
在一种可能的实现方式中，所述计算参数为在设计所述矩阵式多孔成像系统时预设的参数。
在一种可能的实现方式中，所述获取计算参数包括：获取标准多孔图像，所述标准多孔图像包括预设模式多孔图像或者面光源多孔图像；根据所述标准多孔图像测算所述计算参数。
在一种可能的实现方式中，获取预设模式多孔图像包括：利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像，所述预设模式多孔图像含有多个预设模式小孔像斑。
在一种可能的实现方式中，获取面光源多孔图像包括：利用矩阵式多孔成像系统在均匀面光源发光条件下形成面光源多孔图像，所述面光源多孔图像含有多个亮场小孔像斑。
在一种可能的实现方式中，所述预设模式由包含两个方向或者多个方向线条的单个模式按照周期长度有规律重复排布构成。
在一种可能的实现方式中，所述周期长度是小孔周期的正整数倍。
在一种可能的实现方式中，所述利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像包括：将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用外置光源照射所述具有预设模式的标准物使之透射成像，获得所述预设模式多孔图像；或者，在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，利用具有预设模式的外置结构光源照射所述矩阵式多孔成像系统的图像采集器上表面，使所述外置结构光源通过所述矩阵式多孔成像系统成像，获得所述预设模式多孔图像；或者，将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用内置光源照射所述具有预设模式的标准物使之反射成像，获得所述预设模式多孔图像；或者，在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，使具有显示功能的内置光源发出具有预设模式的光线照射所述矩阵式多孔成像系统的图像采集器上表面反射成像，获得所述预设模式多孔图像。
在一种可能的实现方式中，使用显示屏作为所述外置光源、所述具有预设模式的外置结构光源、所述内置光源或所述具有显示功能的内置光源。
在一种可能的实现方式中，使用所述计算参数计算得到拼接参数包括：使用所述计算参数计算得到小孔位置和最大无重叠像方视场尺寸；使用所述最大无重叠像方视场尺寸计算成像分辨率。
在一种可能的实现方式中，所述使用所述计算参数计算得到小孔位置包括：利用矩阵式多孔成像系统获取面光源多孔图像；在所述面光源多孔图像上，以所述面光源多孔图像上任意一点为原点建立直角坐标系；确定小孔位置，所述小孔位置为所述亮场小孔像斑的几何中心在所述直角坐标系中对应的坐标。
在一种可能的实现方式中，所述使用所述计算参数的计算最大无重叠像方视场尺寸包括：从预设模式多孔图像中测量得到最大无重叠像方视场尺寸；或者，利用所述计算参数中的物距和像距计算得到最大无重叠像方视场尺寸；或者，对所述预设模式小孔像斑的拼接效果进行最优估计得到最大无重叠像方视场尺寸。
在一种可能的实现方式中，对所述目标物小孔像斑进行倒像纠正包括：根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；获取所述目标物小孔像斑上的像素在以所述目标物小孔像斑的中心为原点的直角坐标系中的位置坐标；将所述像素的位置坐标做关于所述原点的中心对称翻转，得到倒像纠正后的目标物小孔像斑。
在一种可能的实现方式中，所述对倒像纠正后的目标物小孔像斑进行拼接包括：从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸为最大无重叠像方视场尺寸；根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段。
在一种可能的实现方式中，所述对倒像纠正后的目标物小孔像斑进行拼接包括：从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸大于最大无重叠像方视场尺寸；根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段。
在一种可能的实现方式中，根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段包括：对于相邻图像片段中不重叠的信息，保留原信息；对于相邻图像片段中重叠的信息，做加权平均或根据拼接效果保留最优。
在一种可能的实现方式中，所述使用矩阵式多孔成像系统的计算参数计算得到拼接参数还包括：使用所述计算参数计算得到亮度矫正参数和畸变矫正参数。
在一种可能的实现方式中，使用所述计算参数计算得到亮度矫正参数包括：根据所述计算参数中的像距和图像传感器的像元尺寸计算得到亮度矫正参数；或者，用矩阵式多孔成像系统获取标准多孔图像；从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；从所述目标物小孔像斑中获取目标物像素；测量所述预设模式像素和所述目标物像素并对比分析，获得亮度矫正参数。
在一种可能的实现方式中，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前，还包括：根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；使用亮度矫正参数调整所述目标物小孔像斑中像素的灰度值，以消除所述子图中目标物小孔像斑由中心向外周的亮度衰减，使得所述子图中目标物小孔像斑由中心向外周的亮度均匀。
在一种可能的实现方式中，所述使用所述计算参数计算得到畸变矫正参数包括：利用所述计算参数中的物距、像距、透明介质折射率计算获得畸变矫正参数；或者，用矩阵式多孔成像系统获取标准多孔图像；从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；从所述目标物小孔像斑中获取目标物像素；对所述预设模式像素与所述目标物像素进行几何特征匹配，获得匹配点对；获得匹配点对的位置；对所述匹配点对的位置差异进行分析测量，获得每个像素的畸变矫正参数。
在一种可能的实现方式中，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前还包括：根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；使用畸变矫正参数调整所述子图中目标物小孔像斑中像素的位置，以消除所述子图中目标物小孔像斑由中心向外周的几何畸变。
在一种可能的实现方式中，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前还包括归一化处理，所述归一化处理包括：将拼接后的目标图像进行亮度、对比度及成像分辨率标准化归一化，使得所述目标图像的平均亮度以及对比度方差在预设范围内，使得成像分辨率为标准值。
在一种可能的实现方式中，在基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接之前还包括：基于多目视觉方法的三维重建过程，所述三维重建过程包括根据小孔排布信息及视差恢复出目标物上各点的深度信息。
本申请还提供一种图像处理装置，包括：目标物多孔图像获取单元，用于利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑；目标图像纠正单元，用于基于矩阵式多孔成像系统的拼接参数，对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑；目标图像拼接单元，用于基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。
附图说明
构成本申请的一部分的说明书附图用来提供对本发明的进一步理解，本发明的示意性实施例及其说明用于解释本发明，并不构成对本发明的不当限定。
图1为一种矩阵式小孔成像系统的成像原理示意图；
图2为拼接后的完整的目标图像；
图3为MAPIS获得的目标物多孔图像；
图4使用外置光源及预设模式成像原理示意图；
图5使用外置结构光源成像原理示意图；
图6使用内置光源及预设模式成像原理示意图；
图7使用具有显示功能的内置光源成像原理示意图；
图8内置光源显示模式示例；
图9定标多孔中心位置的亮斑图像示例；
图10预设模式的定标图案示例1；
图11预设模式的定标图案示例2；
图12预设模式的定标图案示例3；
图13以图3作为输入图像，利用不正确的Si参数拼接得到的图像；
图14小孔成像产生的亮度衰减示意图；
图15穿透不同折射率的介质层成像引起的几何畸变示意图；
图16 MAPIS光路中几何畸变的产生原理示意图；
图17 y i随y o的变化曲线；
图18a为本实施例提供的一种目标物多孔图像处理流程图；
图18b为对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正的流程图；
图18c为拼接倒像纠正后的目标物小孔像斑的流程图；
图18d为本实施例提供的另一种目标物多孔图像处理过程流程图(其中虚线框内的过程为可选步骤)；
图19小孔成像中倒像的产生原理示意图；
图20小孔图像的倒像纠正方法示意图；
图21多孔成像时，物面上的视场及重叠区示意图；
图22亮度矫正后的目标物多孔图像(原始图像为图3)；
图23几何畸变矫正后的目标物多孔图像(原始图像为图3)；
图24倒像纠正后的目标物多孔图像(原始图像为图3)；
图25图像增强后的拼接图像(原始图像为图3)；
图26为本实施例提供的多孔图像处理装置的结构示意图；
图27为本实施例所述倒像纠正模块200的结构示意图；
图28为本实施例所述拼接参数获取模块400的结构示意图。
附图标记说明：
1-物面，2-成像小孔，3-阻光层，4-像面，6-物方视场，7-最大无重叠物方视场，8-小孔周期，9-最大无重叠像方视场，10-小孔像斑，11-矩阵式多孔成像系统不具有内置光源的图像采集器，12-具有预设模式的第一标准物，13-外置光源，14-具有预设模式的外置结构光源，15-矩阵式多孔成像系统具有内置光源的图像采集器，16-具有预设模式的第二标准物，17-矩阵式多孔成像系统具有显示功能的内置光源的图像采集器。
具体实施方式
图1是本实施例中MAPIS多孔图像的成像原理图。矩阵式多孔成像系统包括矩阵式排布有多个成像小孔2的阻光层3。图1中以两个相邻的成像小孔为例示出。物面1(Object Plane)上目标物体的光线照射至阻光层3上，一部分光线被阻挡于阻光层3外部，另一部分光线经过成像小孔2照射至像面4(Image Plane)上，形成小孔 像斑。因此，使用该矩阵式多孔成像系统进行图像采样，得到的是一幅矩阵式排布有多个小孔像斑的多孔图像。本实施例用该矩阵式多孔成像系统获得的图像为例说明本申请的技术方案。
图3为用图1所示的矩阵式多孔成像系统采集到的目标物多孔图像，是一枚指纹的多孔图像。如图3所示，该多孔图像中矩阵式排布有多个圆形的小孔像斑。具体地，在多孔图像中每行有6个小孔像斑，每列有6个小孔像斑，每个小孔像斑呈圆形。每个小孔像斑能够反映一部分物方视场，而不是全部的物方视场。相邻两个甚至相邻多个小孔像斑存在重叠信息。为了得到如图2所示完整无重叠无遗漏的目标图像，需要将各个小孔像斑进行拼接。
具体地，在一种可实现的方式中，在对目标物多孔图像进行处理前，还包括获取矩阵式多孔成像系统的拼接参数的步骤，获取所述拼接参数的方法包括：获取计算参数；使用所述计算参数计算得到拼接参数。所述计算参数为在设计所述矩阵式多孔成像系统时预设的参数，或者，所述计算参数通过包括以下步骤的方法获取，获取计算参数包括：获取标准多孔图像，所述标准多孔图像包括预设模式多孔图像或者面光源多孔图像；根据所述标准多孔图像测算所述计算参数。
所述计算参数包括小孔位置、物距、像距、图像传感器的像元尺寸、相邻小孔间距和透明介质折射率等。
在本申请中，获取预设模式多孔图像包括：利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像，所述预设模式多孔图像含有多个预设模式小孔像斑。
在一种可能的实现方式中，所述预设模式由包含两个方向或者多个方向线条的单个模式按照周期长度有规律重复排布构成，所述周期长度是小孔周期的正整数倍。
在一种可能的实现方式中，所述利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像包括：将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用外置光源照射所述具有预设模式的标准物使之透射成像，获得所述预设模式多孔图像；或者，在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，利用具有预设模式的外置结构光源照射所述矩阵式多孔成像系统的图像采集器上表面，使所述具有预设模式的外置结构光源通过所述矩阵式多孔成像系统成像，获得所述预设模式多孔图像；或者，将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用内置光源照射所述具有预设模式的标准物使之反射成像，获得所述预设模式多孔图像；或者，在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，使具有显示功能的内置光源发出具有预设模式的光线照射所述矩阵式多孔成像系统的图像采集器上表面反射成像，获得所述预设模式多孔图像。
在一种可能的实现方式中，使用显示屏作为所述外置光源、所述具有预设模式的外置结构光源、所述内置光源或所述具有显示功能的内置光源。
所述利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像包括：使用外置光源及预设模式成像，使用具有预设模式的外置结构光源成像，使用内置光源及预设模式成像，或者使用具有显示功能的内置光源成像。
图4为使用外置光源及预设模式成像的示意图。结合图4，所述使用外置光源及预设模式成像包括：
将具有预设模式的第一标准物12放置于矩阵式多孔成像系统不具有内置光源的图像采集器11上表面，
利用外置光源13照射所述具有预设模式的第一标准物12使之透射成像，获得包括透射像斑的图像，将包括透射像斑的图像作为所述预设模式多孔图像。
所述预设模式是指特定图形。所述具有预设模式的标准物为通过一定的工艺，如蒸镀、刻蚀、打印等，在表面形成特定图形的标准物，如打印有特定几何图形的胶片，刻蚀有特定几何图形的玻璃片。
所述外置光源为通用的点光源或面光源。
图5为使用具有预设模式的外置结构光源成像的示意图。结合图5，所述具有预设模式的使用外置结构光源成像包括：
在矩阵式孔成像系统不具有内置光源的图像采集器11表面不放置任何物体，利用具有预设模式的外置结构光源14照射矩阵式孔成像系统不具有内置光源的图像采集器11表面，所述具有预设模式的外置结构光源14为可产生预设模式的点光源或面光源。
使所述具有预设模式的外置结构光源通过所述矩阵式多孔成像系统成像，输出所述预设模式多孔图像。
图6为使用内置光源及预设模式成像的示意图。结合图6，所述使用内置光源及预设模式成像包括：
将具有预设模式的第二标准物16放置于矩阵式多孔成像系统具有内置光源的图像采集器15表面。利用内置光源照射所述具有预设模式的标准物使之反射成像，输出所述预设模式多孔图像，所述内置光源为设置于矩阵式多孔成像系统的图像采集器内部的点光源或面光源。
图7为使用具有显示功能的内置光源成像的示意图。结合图7，所述使用具有显示功能的内置光源成像包括：
在矩阵式多孔成像系统具有显示功能的内置光源的图像采集器17表面不放置任何物体，使具有显示功能的内置光源发出反映预设模式的光线通过矩阵式多孔成像系统的图像采集器表面反射成像，输出所述预设模式多孔图像。
具有显示功能的内置光源指置于MAPIS采集器内部的可显示预设模式的光源。
可选地，所述有显示功能的内置光源可以是手机显示屏幕。
图8为有显示功能的内置光源显示模式示例。图8中，有显示功能的内置光源所显示的预设模式设置为规则的“田”字。
受加工误差影响，与设计的小孔位置相比，所有成像小孔的真实小孔位置可能会有整体的平移或者旋转误差，因此拼接参数中小孔位置可以使用在设计所述矩阵式多孔成像系统时预设的参数，也可以利用计算参数通过计算来确定。
在本申请中，使用所述计算参数计算得到拼接参数包括：使用所述计算参数计算 得到小孔位置和最大无重叠像方视场尺寸；使用所述最大无重叠像方视场尺寸计算成像分辨率。
图9为含有多个亮场小孔像斑的面光源多孔图像的示例。具体地，结合图9，所述使用所述计算参数计算得到小孔位置包括：利用矩阵式多孔成像系统获取面光源多孔图像；在所述面光源多孔图像上，以所述面光源多孔图像上任意一点为原点建立直角坐标系；确定小孔位置，所述小孔位置为所述亮场小孔像斑的几何中心在所述直角坐标系中对应的坐标。
在一种可实现的方式中，以所述面光源多孔图像左上角顶点作为原点建立直角坐标系，以便直观表示各成像小孔的小孔位置。
可选的，可用实心圆来拟合各成像小孔的像斑，以实心圆的圆心作为小孔像斑的几何中心，从而获得小孔位置。
在本申请中，获取面光源多孔图像包括：利用矩阵式多孔成像系统在均匀面光源发光条件下形成面光源多孔图像，所述面光源多孔图像含有多个亮场小孔像斑。
结合图1，将水平或垂直方向上相邻两个成像小孔2的中心之间的距离定义为小孔周期8，记为P pinhole。将每个小孔像斑10在像面4上的区域定义为像方视场(Field of View on Image Plane,FOVI)。每个小孔像斑10对应了物面上的一块区域，记为物方视场6(Field of View on Object Plane,FOVO)。按照MAPIS的设计原则，相邻成像小孔的物方视场6是有重叠的，即相邻小孔像斑上的信息是有冗余的，如图21所示。将整个物方视场按成像小孔进行均匀划分，使得物方视场中的每个点被分配到距离最近的成像小孔，用这种方式划分得到的成像小孔的物方视场区域定义为最大无重叠物方视场7(Maximum Non-overlapping FOV on Object Plane,MNFOVO)，其尺寸记为S o。每个最大无重叠物方视场7对应的像方视场定义为最大无重叠像方视场9(Maximum Non-overlapping FOV on Image Plane,MNFOVI)，其尺寸记为S i。图1显示了它们之间的关系。
MAPIS产品在出厂时会对S i值进行定标。在实际使用过程中像距ID不发生变化，但用户的一些操作却会改变物距OD，在这种情况下需要对S i值重新定标。例如，当MAPIS与手机屏幕结合构成图像采集设备时，或者在手机屏幕下贴装MAPIS图像采集设备的情况下，如果用户在手机屏幕表面贴膜则会改变物距，这会改变系统的S i值。
在本申请中，所述使用所述计算参数的计算最大无重叠像方视场尺寸包括：从预设模式多孔图像中测量得到最大无重叠像方视场尺寸；或者，利用所述计算参数中的物距和像距计算得到最大无重叠像方视场尺寸；或者，对所述预设模式小孔像斑的拼接效果进行最优估计得到最大无重叠像方视场尺寸。
具体地，所述从预设模式多孔图像中测量得到最大无重叠像方视场尺寸S i包括：
使用具有预设模式的标准物在所述MAPIS系统中成像，获取预设模式多孔图像，所述预设模式具有预设尺寸S′ o，设定S′ o＝c*S o，其中，c为已知常数；
在所述预设模式多孔图像中测量出该预设模式成像的尺寸S′ i，那么S′ i＝c*S i。
因此，可根据公式(1)计算S i。
S i＝S o×S′ i/S′ o   公式(1)
在一种可实现的方式中，所述预设模式可以是由包含两个方向或者多个方向线条的单个模式按照周期长度有规律重复排布构成，如图10、图11、图12所示的模式，以便于测量所述预设模式多孔图像中预设模式的成像尺寸。可选的，所述周期长度是小孔周期8的正整数倍，例如在图10中，周期长度为相邻两个正方形各自的几何中心的距离，所述周期长度是小孔周期8的正整数倍。
可选地，物距OD和像距ID用下面的公式(2)和公式(3)表示，
ID＝ID 0+δ ID  公式(2)
OD＝OD 0+δ OD+Δ OD  公式(3)
其中，ID 0表示预设像距值，OD 0表示预设物距值，δ ID为像距在生产过程中的加工误差，δ OD为物距在生产过程中的加工误差，Δ OD为物距在出厂后的变化量。
如果能够准确获知MAPIS的物距OD和像距ID，或者已知ID 0、OD 0的设计参数和出厂后物距变化量Δ OD，并且加工误差δ ID、δ OD可以忽略，则可利用根据第一预设规则计算S i。
在一种可能的实现方式中，所述利用所述计算参数中的物距和像距计算得到最大无重叠像方视场尺寸，具体地根据如下公式(4)进行计算：
S i＝S o×ID/OD，公式(4)
对于公式(4)，由于本实施例中所用的图像采集设备为矩阵式等周期分布的MAPIS，而且小孔周期P pinhole在实际生产MAPIS时的加工误差可忽略，因此，可以用小孔周期P pinhole作为最大无重叠物方视场，即，S o＝P pinhole。
在某些情况下虽然不能准确获知物距OD和像距ID，在已知ID 0、OD 0的前提下，出厂后物距变化量Δ OD和加工误差δ ID、δ OD的取值范围如果可以较为准确地知道，可以用最优拼接法估计S i。即对具有明显对比度的实际目标物(如指纹)成像，尝试用S i的各种可能值进行拼接，对拼接效果进行评价，挑选最优值，即，对所述预设模式小孔像斑的拼接效果进行最优估计得到最大无重叠像方视场尺寸。S i的可能值依据物距变化量Δ OD和加工误差δ ID、δ OD的取值范围确定。
在进行拼接时，如果S i准确，则得到的拼接图像在各个小孔像斑相接的地方过渡自然平滑，相反，如果S i错误，则在各个小孔像斑相接的地方会存在明显的错位和瑕疵，如图13所示。通过对拼接效果的评估，选择最佳拼接效果对应的S i作为该MAPIS图像采集设备的S i。
在一种可实现的方式中，所述MAPIS的成像分辨率利用最大无重叠像方视场尺寸根据下式公式(5)获取。
R＝S i/(S pS o)＝ID/(S pOD)  公式(5)
其中，R为所述MAPIS的成像分辨率，S p为图像传感器的像元尺寸，所述S p是在矩阵式多孔成像系统设计时预设的参数。
在一种可实现的方式中，所述使用矩阵式多孔成像系统的计算参数计算得到拼接参数还包括：使用所述计算参数计算得到亮度矫正参数和畸变矫正参数。
由于小孔成像本身会引起每个小孔像斑由中心向周边的亮度衰减，因此，用MAPIS图像采集设备采集到的多孔图像上每个小孔像斑由中心向周边的亮度不一致，如图14所示。如果不对多孔图像进行亮度矫正，则拼接得到的目标图像会存在亮度不均的问题。
可选地，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前，还包括：根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；使用亮度矫正参数调整所述目标物小孔像斑中像素的灰度值，以消除所述子图中目标物小孔像斑由中心向外周的亮度衰减，使得所述子图中目标物小孔像斑由中心向外周的亮度均匀。
可选地，所述使用所述计算参数计算得到亮度矫正参数包括：
根据所述计算参数中的像距和图像传感器的像元尺寸计算得到亮度矫正参数，具体地，
对光路进行理论分析可知，使用均匀面光源成像时，在小孔像斑上的关系如公式(6)所示：
E x＝E ocos 4θ  公式(6)
其中，E x表示小孔像斑上与小孔中心距离为x的点X的照度，E o表示小孔像斑中心处的照度，θ为X点与小孔中心的连线和小孔光轴的夹角，即图16中的θ 2，小孔图像上每一点的θ可根据物距OD、像距ID、透明介质的折射率n 1、n 2及当前点在像面上与小孔中心的距离y i计算得到。记1/cos 4θ为亮度矫正参数，从而，X点校正后的亮度根据公式(7)计算获得：
I′ x＝I x/cos 4θ   公式(7)
其中，I′ x为校正后亮度，I x为校正前亮度。
或者，所述使用所述计算参数计算得到亮度矫正参数包括：用矩阵式多孔成像系统获取标准多孔图像；从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；从所述目标物小孔像斑中获取目标物像素；测量所述预设模式像素和所述目标物像素并对比分析，获得亮度矫正参数。
具体地，对面光源或预设模式成像的小孔像斑进行统计，获得亮度衰减模板，如图14所示；
依据公式(8)，矫正任一点X的亮度：
I′ x＝I x×T o/T x   公式(8)
其中，I′ x为校正后亮度，I x为校正前亮度，T o为亮度衰减模板中心的亮度值，T x为亮度衰减模板中X点的亮度值。此时，亮度校正参数变为T o/T x。
图15为用预设模式获取的未进行畸变矫正的小孔像斑。MAPIS的图像采集设备在工作时，光线从物体表面透过小孔到达图像传感器，由于穿透折射率不同的介质层 导致成像存在一定的几何畸变，如图15所示，如果不进行畸变矫正，拼接的图像会存在一定程度的畸变和混叠。
可选地，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前还包括：根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；使用畸变矫正参数调整所述子图中目标物小孔像斑中像素的位置，以消除所述子图中目标物小孔像斑由中心向外周的几何畸变。
在一种可实现的方式中，所述使用所述计算参数计算得到畸变矫正参数包括：用矩阵式多孔成像系统获取标准多孔图像；从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；从所述目标物小孔像斑中获取目标物像素；对所述预设模式像素与所述目标物像素进行几何特征匹配，获得匹配点对；获得匹配点对的位置；对所述匹配点对的位置差异进行分析测量，获得每个像素的畸变矫正参数。
具体地，图16为小孔像斑畸变原理图，如图16所示，假设从物面1到阻光层3为透明介质I，其折射率为n 1，从阻光层3到像面4为透明介质II，折射率为n 2，则有如下公式(9)、公式(10)和公式(11)成立：
y o＝OD·tanθ 1  公式(9)
y i＝ID·tanθ 2  公式(10)
n 1·sinθ 1＝n 2·sinθ 2  公式(11)
其中，y o为物方点距离小孔中心的距离，y i为像方点距离小孔中心的距离，OD、ID分别为MAPIS的物距和像距,θ 1表示入射角、θ 2表示出射角。
由公式(9)、公式(10)和公式(11)可以得到公式(12)：
y i/y o＝(ID*n 1*COSθ 1)/(OD*n 2*cosθ 2)  公式(12)
若n 1≠n 2，则y i不会随y o线性变化。从而距离小孔中心不同位置上物方点经小孔成像时，放大率不同，存在畸变。图17示出y i随y o变化曲线，如图17所示，当n 1＜n 2时，可以估计出小孔图像上各个位置的矫正参数。即
(x′ i，y′ i)＝C x，y(x i，y i)      公式(13)
C x，y＝(OD*n 2*cosθ 2)/(ID*n 1*cosθ 1)  公式(14)
其中，(x i，y i)是小孔像斑上某一点校正前的坐标，(x′ i，y′ i)是该点的校正后坐标。C x，y是矫正参数，与点(x i，y i)到小孔中心的距离有关。
或者，所述使用所述计算参数计算得到畸变矫正参数包括：利用所述计算参数中的物距、像距、透明介质折射率计算获得畸变矫正参数，具体地，根据公式(14)进行计算。若MAPIS系统的结构与图16不符，则需要根据实际结构调整公式(14)。
所述利用矩阵式多孔成像系统获取预设模式多孔图像，对目标物多孔图像中每个目标物小孔像斑包含的模式图像与预设模式图像对比，对差异进行分析测量，获得每一点的畸变矫正参数中，不妨假设获得的一个目标物小孔像斑如图15中所示，而预 设模式小孔像斑如图12所示。利用几何特征法匹配图15的目标物小孔像斑与图12的预设模式小孔像斑，可以得到逐点的对应关系；对于缺乏几何特征的区域，可以通过插值方法获得对应关系。设想通过几何特征匹配，目标物小孔像斑上的某一点(TI x，TI y)匹配了预设模式小孔像斑的(TO x，TO y)点，这里的坐标值以目标物小孔中心为原点。那么此时，公式(15)成立：
Figure PCTCN2018116904-appb-000001
从而可以依据公式(15)进行畸变矫正。
图18a为本实施例中对MAPIS获得的目标物多孔图像处理方法的流程图。结合图18a，说明本实施例的方法。在本申请中，所述目标物多孔图像是指目标物利用MAPIS获得的多孔图像，所述预设模式多孔图像是指具有预设模式的标准物利用MAPIS获得的多孔图像。
S100利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑。
S200基于矩阵式多孔成像系统的拼接参数，对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑。在本申请中，所述倒像纠正可以是对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正，也可是对所述目标物多孔图像中的一部分目标物小孔像斑进行倒像纠正。
S300基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。在本申请中，所述拼接可以是对所述倒像纠正后的目标物小孔像斑进行拼接，也可以是对所述倒像纠正后的目标物小孔像斑中的一部分进行拼接。
在S100中，所述目标物小孔像斑可以为任意形状。可选地，所述小孔像斑为圆形或者正方形等。一是便于矩阵式多孔成像系统的制造，二是规则形状的像斑更易于图像分割、倒像纠正以及拼接等操作。
在S200中，所述倒像纠正后的目标物小孔像斑与物方视场的方向一致，便于后续的图像拼接及其他处理和识别工作。
图18b为对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正的流程图，结合图18b，对所述多孔图像中的小孔像斑进行倒像纠正包括：
S201根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑。
S202获取所述目标物小孔像斑上的像素在以所述目标物小孔像斑的中心为原点的直角坐标系中的位置坐标。
在每个子图上，以目标物小孔像斑的中心为原点建立直角坐标系，每个像素的位置用其在该直角坐标系中对应的位置坐标表示。
S203将所述像素的位置坐标做关于所述原点的中心对称翻转，得到倒像纠正后的目标物小孔像斑。将所述目标物小孔像斑上像素做关于所述原点的中心对称翻转， 得到倒像纠正后的目标物小孔像斑。
示例地，图20为本实施例中心对称翻转的示意图。结合图20，对每个目标物小孔像斑，以目标物小孔像斑的中心为原点建立直角坐标系，每个像素对应一个位置坐标(x,y)以及一个灰度值i；对像素对应的位置坐标做关于所述原点的中心对称翻转，即(x,y,i)变为(-x,-y,i),得到倒像纠正后的目标物小孔像斑。
示例地，以图3为原始图像获得的倒像纠正后的目标物多孔图像如图24所示。
图18c为一种可实现的方式中，拼接倒像纠正后的目标物小孔像斑的流程，结合图18c，S300所述对倒像纠正后的目标物小孔像斑进行拼接包括：
S301从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸为最大无重叠像方视场尺寸；具体地，以每个倒像纠正后的目标物小孔像斑的几何中心为中心，从所述目标物小孔像斑中，取出尺寸恰好等于所述矩阵式多孔成像系统的最大无重叠像方视场尺寸S i的图像片段。
S302根据所述目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段，生成完整的目标图像。
另一种可实现的方式中，S300所述对倒像纠正后的目标物小孔像斑进行拼接包括：
从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸大于最大无重叠像方视场尺寸，具体地，以每个倒像纠正后的小孔像斑的中心为中心，从所述小孔像斑中取出尺寸大于所述矩阵式多孔成像系统的最大无重叠像方视场尺寸的图像片段，使相邻图像片段外周部分有信息重叠，所述信息是指所述图像片段上的图像信息；根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段。
在一种可能的实现方式中，所述根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段包括：对于相邻图像片段中不重叠的信息，保留原信息；对于相邻图像片段中重叠的信息，做加权平均或保留最优，特别的，按照预设规则保留最优，所述预设规则可以是优先选择与其邻域对比度最大的点，或优先选择亮度最高的点。
图18d为本实施例提供的另一种多孔图像处理方法流程图，结合图18d，所述方法包括：小孔倒像纠正和图像拼接，所述小孔倒像纠正是指对目标物小孔像斑进行倒像纠正，让你双眼皮图像拼接是指对倒像纠正后的目标物小孔像斑进行图像拼接。可选地，在小孔倒像纠正前还包括：小孔级图像分割、小孔亮度畸变矫正、小孔几何畸变矫正、小孔三维重建和深度计算中的一个或者多个步骤。
进一步可选地，在基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像后，还可以包括拼接图像增强和分割以及图像归一化中的一个或者多个步骤。
本发明人发现，如果在每个目标物小孔像斑上只截取尺寸恰好为S i的图像片段， 拼接得到的目标图像容易在相邻目标物小孔像斑相接的地方产生块效应。因此如果对每个目标物小孔像斑的取图范围做一定的扩展，使得到的图像片段的尺寸大于S i，就会使相邻的图像片段在内容上有部分重叠信息，再对重叠信息加权平均或按照预设规则保留最优，可以使目标图像内容的过渡更加平滑自然，避免拼接过程中的块效应。
在一种可能的实现方式中，所述图像处理方法还包括依据公式(7)或(8)的亮度矫正过程，使得所述目标物多孔图像中每个目标物小孔像斑由中心向外周的亮度衰减矫正到每个单独小孔像斑的亮度均匀。
可选地，所述亮度矫正在将所述目标物多孔图像分割为若干子图之前进行。
在一种可能的实现方式中，所述图像处理方法还包括依据公式(13)，公式(14)或(15)的畸变矫正过程，从而消除所述目标物多孔图像中每个目标物小孔像斑由中心向外周的几何畸变。
可选地，所述畸变矫正在亮度矫正之后，在将所述目标物多孔图像分割为若干子图之前进行。
以图1作为原始的目标物多孔图像，先对目标物多孔图像进行亮度矫正，结果如图22所示，再对亮度矫正后的目标物多孔图像进行畸变矫正，结果如图23所示。
在另一种可能的实现方式中，所述目标物多孔图像处理方法还包括图像增强处理和图像分割处理，使得目标图像易于识别。
优选地，所述图像增强处理可以采用直方图均衡化方法。
示例地，以图3为原始图像经过图像增强获得的图像如图25所示。
所述图像分割处理能够消除环境对目标物成像的干扰。
在另一种可实现的方式中，所述图像处理方法还包括归一化处理。归一化处理是将目标图像的亮度、对比度以及成像分辨率分别进行标准化归一化处理，使目标图像的平均亮度、对比度方差在预设范围内，成像分辨率为预设的标准值。归一化处理后的目标图像可方便后续的识别或其他应用。
可选地，根据下列公式(16)对灰度和灰度的方差进行归一化处理：
I′(x，y)＝(I(x，y)-μ)*σ ideal/σ+μ ideal   公式(16)
其中，
μ和σ为归一化处理前目标图像的灰度均值和灰度的方差，μ ideal和σ ideal是预设的灰度均值和灰度的方差。
I(x，y)是归一化处理前目标图像上(x，y)点的灰度值，I′(x，y)是归一化处理后目标图像上(x，y)点的灰度值。由于灰度对应于亮度，灰度的方差对应于对比度，从而上式将亮度和对比度归一化。
可选地，根据下列公式(17)对成像分辨率进行标准化处理：
I′(x，y)＝I(x×R ideal/R，y×R ideal/R) 公式(17)
其中，R是标准化前目标图像的成像分辨率，R ideal是预设的成像分辨率的标准值。
在另一种可能的实现方式中，所述图像处理方法在根据小孔位置将所述目标物多 孔图像分割为若干子图之前还包括基于多目视觉方法的三维重建过程。
基于多目视觉方法的三维重建过程，所述三维重建过程包括根据小孔排布信息及视差恢复出目标物上各点的深度信息。在多孔成像时，目标物上的一个点会通过多个成像小孔形成多个具有视差的像，利用这多个像的信息及矩阵式多孔成像系统的物距、像距、小孔间距等参数恢复出目标物上的点的深度信息，将目标物上所有点的深度信息进行归一化输出，得到完整的目标图像。
本申请还提供一种图像处理装置，图26为本实施例提供的多孔图像处理装置的结构示意图。结合图26，所述装置包括：目标物多孔图像获取模块100，用于利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑；倒像纠正模块200，用于基于矩阵式多孔成像系统的拼接参数，对所述目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑；目标图像生成模块300，用于基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。
图27为倒像纠正模块200的结构示意图，结合图27，在一种可实现的方式中，所述倒像纠正模块200包括：图像分割单元201，用于根据小孔位置将所述目标物多孔图像分割为若干子图，每个子图上包括一个完整的小孔像斑；像素确定单元202，用于在每个子图上，以所述小孔像斑的中心为原点建立直角坐标系，获取每个像素的位置和灰度值，每个像素的位置用其在该直角坐标系中对应的位置坐标表示；倒像纠正单元203，用于将所述小孔像斑上像素做关于所述原点的中心对称翻转，得到倒像纠正后的目标物小孔像斑。
在一种可实现的方式中，所述目标图像生成模块300包括：图像片段截取单元，用于从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸大于或者等于最大无重叠像方视场尺寸；图像片段拼接单元，用于根据所述多孔图像中各小孔像斑的相对位置拼合所述图像片段，生成完整的目标图像，如果相邻图像片段存在信息重叠，则对于相邻图像片段中不重叠的信息，保留原信息，对于相邻图像片段中重叠的信息，做加权平均或保留最优，特别的，按照预设规则保留最优，所述预设规则可以是优先选择与其邻域对比度最大的点，或优先选择亮度最高的点。
可选地，所述装置还包括拼接参数获取模块400，用于获取拼接参数。
图28为所述拼接参数获取模块400的结构示意图，结合图28，在一种可能的实现方式中，所述拼接参数获取模块400包括计算参数获取单元401，用于获取计算参数；拼接参数测算单元402，用于使用所述计算参数计算得到拼接参数。
所述计算参数获取单元401包括标准多孔图像获取子单元，用于获取标准多孔图像，所述标准多孔图像包括预设模式多孔图像或者面光源多孔图像；计算参数测算子单元，用于根据所述标准多孔图像测算所述计算参数。
所述标准多孔图像获取子单元包括预设模式多孔图像获取从单元，用于利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像，所述预设模式多孔图像含有 多个预设模式小孔像斑；面光源多孔图像获取从单元，用于利用矩阵式多孔成像系统在均匀面光源发光条件下形成面光源多孔图像，所述面光源多孔图像含有多个亮场小孔像斑。
所述预设模式多孔图像获取从单元包括以下预设模式多孔图像获取器中的至少一种：
第一预设模式多孔图像获取器，用于将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用外置光源照射所述具有预设模式的标准物使之透射成像，获得所述预设模式多孔图像；
第二预设模式多孔图像获取器，用于在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，利用具有预设模式的外置结构光源照射所述矩阵式多孔成像系统的图像采集器上表面，使所述具有预设模式的外置结构光源通过所述矩阵式多孔成像系统成像，获得所述预设模式多孔图像；
第三预设模式多孔图像获取器，用于将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用内置光源照射所述具有预设模式的标准物使之反射成像，获得所述预设模式多孔图像；
第四预设模式多孔图像获取器，用于在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，使具有显示功能的内置光源发出具有预设模式的光线照射所述矩阵式多孔成像系统的图像采集器上表面反射成像，获得所述预设模式多孔图像。
在一种可实现的方式中，所述拼接参数测算单元402包括：小孔位置获取子单元，最大无重叠像方视场尺寸获取子单元和成像分辨率获取子单元。
所述小孔位置获取子单元包括：面光源多孔图像获取从单元，用于利用矩阵式多孔成像系统获取面光源多孔图像；直角坐标系建立从单元，用于在所述面光源多孔图像上，以所述面光源多孔图像上任意一点为原点建立直角坐标系；小孔位置确定从单元，用于确定小孔位置，所述小孔位置为所述亮场小孔像斑的几何中心在所述直角坐标系中对应的坐标。
所述最大无重叠像方视场尺寸获取子单元包括以下从单元中的至少一种：最大无重叠像方视场尺寸测量从单元，用于从预设模式多孔图像中测量得到最大无重叠像方视场尺寸；最大无重叠像方视场尺寸计算从单元，用于利用所述计算参数中的物距和像距计算得到最大无重叠像方视场尺寸；最大无重叠像方视场尺寸最优估计从单元，用于对所述预设模式小孔像斑的拼接效果进行最优估计得到最大无重叠像方视场尺寸。
在一种可实现的方式中，所述拼接参数测算单元402还包括亮度矫正参数获取子单元，用于获取亮度矫正参数和畸变矫正参数获取子单元，用于获取畸变矫正参数。
在一种可实现的方式中，如图26所示，所述装置还包括归一化处理模块500，用于将拼接后的目标图像进行亮度、对比度及成像分辨率标准化归一化，使得所述目标图像的平均亮度以及对比度方差在预设范围内，使得成像分辨率为标准值。
在一种可实现的方式中，如图26所示，所述装置还包括基于多目视觉方法的三 维重建模块600，用于根据小孔排布信息及视差恢复出目标物上各点的深度信息。
以上结合具体实施方式和范例性实例对本申请进行了详细说明，不过这些说明并不能理解为对本申请的限制。本领域技术人员理解，在不偏离本申请精神和范围的情况下，可以对本申请技术方案及其实施方式进行多种等价替换、修饰或改进，这些均落入本申请的范围内。本申请的保护范围以所附权利要求为准。

Claims (26)

  1. 一种图像处理方法，其特征在于，包括：
    利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑；
    基于矩阵式多孔成像系统的拼接参数，对所述目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑；
    基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。
  2. 根据权利要求1所述的方法，其特征在于，在对所述目标物小孔像斑进行倒像纠正之前还包括获取拼接参数。
  3. 根据权利要求2所述的方法，其特征在于，所述获取拼接参数包括：
    获取计算参数；
    使用所述计算参数计算得到拼接参数。
  4. 根据权利要求3所述的方法，其特征在于，所述计算参数为在设计所述矩阵式多孔成像系统时预设的参数。
  5. 根据权利要求3所述的方法，其特征在于，所述获取计算参数包括：
    获取标准多孔图像，所述标准多孔图像包括预设模式多孔图像或者面光源多孔图像；
    根据所述标准多孔图像测算所述计算参数。
  6. 根据权利要求5所述的方法，其特征在于，获取预设模式多孔图像包括：
    利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像，所述预设模式多孔图像含有多个预设模式小孔像斑。
  7. 根据权利要求5所述的方法，其特征在于，获取面光源多孔图像包括：
    利用矩阵式多孔成像系统在均匀面光源发光条件下形成面光源多孔图像，所述面光源多孔图像含有多个亮场小孔像斑。
  8. 根据权利要求6所述的方法，其特征在于，所述预设模式由包含两个方向或者多个方向线条的单个模式按照周期长度重复排布构成。
  9. 根据权利要求8所述的方法，其特征在于，所述周期长度是小孔周期的正整数倍。
  10. 根据权利要求6所述的方法，其特征在于，所述利用矩阵式多孔成像系统获取预设模式对应的预设模式多孔图像包括：
    将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用外置光源照射所述具有预设模式的标准物使之透射成像，获得所述预设模式多孔图像；或者，
    在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，利用具有预设模式的外置结构光源照射所述矩阵式多孔成像系统的图像采集器上表面，使所述具有预设模式的外置结构光源通过所述矩阵式多孔成像系统成像，获得所述预设 模式多孔图像；或者，
    将具有预设模式的标准物放置于矩阵式多孔成像系统的图像采集器上表面，利用内置光源照射所述具有预设模式的标准物使之反射成像，获得所述预设模式多孔图像；或者，
    在矩阵式多孔成像系统的图像采集器上表面不放置任何物体，使具有显示功能的内置光源发出具有预设模式的光线照射所述矩阵式多孔成像系统的图像采集器上表面反射成像，获得所述预设模式多孔图像。
  11. 根据权利要求10所述的方法，其特征在于，使用显示屏作为所述外置光源、所述具有预设模式的外置结构光源、所述内置光源或所述具有显示功能的内置光源。
  12. 根据权利要求3所述的方法，其特征在于，使用所述计算参数计算得到拼接参数包括：
    使用所述计算参数计算得到小孔位置和最大无重叠像方视场尺寸；
    使用所述最大无重叠像方视场尺寸计算成像分辨率。
  13. 根据权利要求12所述的方法，其特征在于，所述使用所述计算参数计算得到小孔位置包括：
    利用矩阵式多孔成像系统获取面光源多孔图像；
    在所述面光源多孔图像上，以所述面光源多孔图像上任意一点为原点建立直角坐标系；
    确定小孔位置，所述小孔位置为亮场小孔像斑的几何中心在所述直角坐标系中对应的坐标。
  14. 根据权利要求12所述的方法，其特征在于，所述使用所述计算参数的计算最大无重叠像方视场尺寸包括：
    从预设模式多孔图像中测量得到最大无重叠像方视场尺寸；
    或者，
    利用所述计算参数中的物距和像距计算得到最大无重叠像方视场尺寸；
    或者，
    对所述预设模式小孔像斑的拼接效果进行最优估计得到最大无重叠像方视场尺寸。
  15. 根据权利要求1所述的方法，其特征在于，对所述目标物小孔像斑进行倒像纠正包括：
    根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；
    获取所述目标物小孔像斑上的像素在以所述目标物小孔像斑的中心为原点的直角坐标系中的位置坐标；
    将所述像素的位置坐标做关于所述原点的中心对称翻转，得到倒像纠正后的目标物小孔像斑。
  16. 根据权利要求1或15所述的方法，其特征在于，所述对倒像纠正后的 目标物小孔像斑进行拼接包括：
    从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸为最大无重叠像方视场尺寸；
    根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段。
  17. 根据权利要求1或15所述的方法，其特征在于，所述对倒像纠正后的目标物小孔像斑进行拼接包括：
    从所述目标物小孔像斑中取出一个图像片段，所述图像片段的中心与所述倒像纠正后的目标物小孔像斑的中心相同，所述图像片段的尺寸大于最大无重叠像方视场尺寸；
    根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段。
  18. 根据权利要求17所述的方法，其特征在于，根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段包括：
    对于相邻图像片段中不重叠的信息，保留原信息；
    对于相邻图像片段中重叠的信息，做加权平均或根据拼接效果保留最优。
  19. 根据权利要求3所述的方法，其特征在于，所述使用所述计算参数计算得到拼接参数还包括：
    使用所述计算参数计算得到亮度矫正参数和畸变矫正参数。
  20. 根据权利要求19所述的方法，其特征在于，使用所述计算参数计算得到亮度矫正参数包括：
    根据所述计算参数中的像距和图像传感器的像元尺寸计算得到亮度矫正参数；
    或者，
    用矩阵式多孔成像系统获取标准多孔图像；
    从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；
    从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；
    从所述目标物小孔像斑中获取目标物像素；
    测量所述预设模式像素和所述目标物像素并对比分析，获得亮度矫正参数。
  21. 根据权利要求19所述的方法，其特征在于，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前，还包括：
    根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；
    使用亮度矫正参数调整所述目标物小孔像斑中像素的灰度值，以消除所述子图中目标物小孔像斑由中心向外周的亮度衰减，使得所述子图中目标物小孔像斑由中心向外周的亮度均匀。
  22. 根据权利要求19所述的方法，其特征在于，所述使用所述计算参数计算得到畸变矫正参数包括：
    利用所述计算参数中的物距、像距、透明介质折射率计算获得畸变矫正参数；
    或者，
    用矩阵式多孔成像系统获取标准多孔图像；
    从所述标准多孔图像中获取预设模式小孔像斑或者亮场小孔像斑；
    从所述预设模式小孔像斑或者亮场小孔像斑中获取预设模式像素；
    从所述目标物小孔像斑中获取目标物像素；
    对所述预设模式像素与所述目标物像素进行几何特征匹配，获得匹配点对；
    获得匹配点对的位置；
    对所述匹配点对的位置差异进行分析测量，获得每个像素的畸变矫正参数。
  23. 根据权利要求19所述的方法，其特征在于，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前还包括：
    根据所述小孔位置将所述目标物多孔图像分割为若干子图，每个子图上有且只有一个完整的目标物小孔像斑；
    使用畸变矫正参数调整所述子图中目标物小孔像斑中像素的位置，以消除所述子图中目标物小孔像斑由中心向外周的几何畸变。
  24. 根据权利要求18所述的方法，其特征在于，在根据目标物多孔图像中各目标物小孔像斑的相对位置拼合所述图像片段之前还包括归一化处理，所述归一化处理包括：
    将拼接后的目标图像进行亮度、对比度及成像分辨率标准化归一化，使得所述目标图像的平均亮度以及对比度方差在预设范围内，使得成像分辨率为标准值。
  25. 根据权利要求1所述的方法，其特征在于，在对倒像纠正后的目标物小孔像斑进行拼接之前还包括：
    基于多目视觉方法的三维重建过程，所述三维重建过程包括根据小孔排布信息及视差恢复出目标物上各点的深度信息。
  26. 一种图像处理装置，其特征在于，包括：
    目标物多孔图像获取单元，用于利用矩阵式多孔成像系统获取目标物对应的目标物多孔图像，所述目标物多孔图像含有多个目标物小孔像斑；
    目标图像纠正单元，用于基于矩阵式多孔成像系统的拼接参数，对所述目标物多孔图像中的目标物小孔像斑进行倒像纠正，得到多个倒像纠正后的目标物小孔像斑；
    目标图像拼接单元，用于基于矩阵式多孔成像系统的拼接参数，对倒像纠正后的目标物小孔像斑进行拼接，生成目标图像。
PCT/CN2018/116904 2018-11-22 2018-11-22 一种图像处理方法及装置 WO2020103075A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/116904 WO2020103075A1 (zh) 2018-11-22 2018-11-22 一种图像处理方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/116904 WO2020103075A1 (zh) 2018-11-22 2018-11-22 一种图像处理方法及装置

Publications (1)

Publication Number Publication Date
WO2020103075A1 true WO2020103075A1 (zh) 2020-05-28

Family

ID=70773777

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/116904 WO2020103075A1 (zh) 2018-11-22 2018-11-22 一种图像处理方法及装置

Country Status (1)

Country Link
WO (1) WO2020103075A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113344782A (zh) * 2021-05-31 2021-09-03 浙江大华技术股份有限公司 图像拼接方法、装置、存储介质及电子装置

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110228142A1 (en) * 2009-10-14 2011-09-22 Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. Device, image processing device and method for optical imaging
CN204028936U (zh) * 2014-05-16 2014-12-17 深圳印象认知技术有限公司 超薄型指纹采集装置、可采集指纹的显示设备
CN106791498A (zh) * 2016-11-18 2017-05-31 成都微晶景泰科技有限公司 图像定位方法、透镜阵列成像方法及装置
CN107194326A (zh) * 2017-04-28 2017-09-22 广东欧珀移动通信有限公司 指纹采集方法及相关产品
CN107358216A (zh) * 2017-07-20 2017-11-17 京东方科技集团股份有限公司 一种指纹采集模组、显示装置及指纹识别方法
CN108242453A (zh) * 2016-12-23 2018-07-03 京东方科技集团股份有限公司 一种oled显示面板及显示装置
CN108388868A (zh) * 2018-02-27 2018-08-10 北京思比科微电子技术股份有限公司 一种近距离感应物体影像的装置
CN109754365A (zh) * 2017-11-07 2019-05-14 印象认知（北京）科技有限公司 一种图像处理方法及装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110228142A1 (en) * 2009-10-14 2011-09-22 Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. Device, image processing device and method for optical imaging
CN204028936U (zh) * 2014-05-16 2014-12-17 深圳印象认知技术有限公司 超薄型指纹采集装置、可采集指纹的显示设备
CN106791498A (zh) * 2016-11-18 2017-05-31 成都微晶景泰科技有限公司 图像定位方法、透镜阵列成像方法及装置
CN108242453A (zh) * 2016-12-23 2018-07-03 京东方科技集团股份有限公司 一种oled显示面板及显示装置
CN107194326A (zh) * 2017-04-28 2017-09-22 广东欧珀移动通信有限公司 指纹采集方法及相关产品
CN107358216A (zh) * 2017-07-20 2017-11-17 京东方科技集团股份有限公司 一种指纹采集模组、显示装置及指纹识别方法
CN109754365A (zh) * 2017-11-07 2019-05-14 印象认知（北京）科技有限公司 一种图像处理方法及装置
CN108388868A (zh) * 2018-02-27 2018-08-10 北京思比科微电子技术股份有限公司 一种近距离感应物体影像的装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113344782A (zh) * 2021-05-31 2021-09-03 浙江大华技术股份有限公司 图像拼接方法、装置、存储介质及电子装置
CN113344782B (zh) * 2021-05-31 2023-07-18 浙江大华技术股份有限公司 图像拼接方法、装置、存储介质及电子装置

Similar Documents

Publication Publication Date Title
US10690492B2 (en) Structural light parameter calibration device and method based on front-coating plane mirror
CN110717942B (zh) 图像处理方法和装置、电子设备、计算机可读存储介质
CN110392252B (zh) 用于生成相机的校正模型以校正像差的方法
US20160371855A1 (en) Image based measurement system
CN107025670A (zh) 一种远心相机标定方法
CN110689581A (zh) 结构光模组标定方法、电子设备、计算机可读存储介质
CN111028205B (zh) 一种基于双目测距的眼睛瞳孔定位方法及装置
CN114494045B (zh) 一种基于机器视觉的大型直齿轮几何参数测量系统及方法
CN108716890A (zh) 一种基于机器视觉的高精度尺寸检测方法
CN109862345B (zh) 视场角测试方法和系统
CN109727290A (zh) 基于单目视觉三角测距法的变焦相机动态标定方法
US20200342596A1 (en) Method for measuring objects in digestive tract based on imaging system
WO2019232793A1 (zh) 双摄像头标定方法、电子设备、计算机可读存储介质
CN113487510A (zh) 用于机器人自动配液的针尖位置检测方法、系统、设备
CN113450418A (zh) 基于复杂畸变模型的水下标定的改进方法、装置及系统
US20140218477A1 (en) Method and system for creating a three dimensional representation of an object
CN113963065A (zh) 一种基于外参已知的镜头内参标定方法及装置、电子设备
WO2020103075A1 (zh) 一种图像处理方法及装置
TW201326735A (zh) 寬度量測方法及系統
JP6346018B2 (ja) 眼球計測システム、視線検出システム、眼球計測方法、眼球計測プログラム、視線検出方法、および視線検出プログラム
CN107527323B (zh) 镜头畸变的标定方法及装置
CN109754365B (zh) 一种图像处理方法及装置
CN105758337A (zh) 一种获取透镜平面与图像传感器平面之间夹角的方法
JP6452236B2 (ja) 眼球識別装置及び眼球識別方法
EP3831060A1 (en) Method and system for mapping the non-uniformity of an image sensor

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18940647

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18940647

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 09.11.2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18940647

Country of ref document: EP

Kind code of ref document: A1