WO2010004677A1 - 画像処理方法、画像処理装置、画像処理プログラム、画像合成方法、および画像合成装置 - Google Patents
画像処理方法、画像処理装置、画像処理プログラム、画像合成方法、および画像合成装置 Download PDFInfo
- Publication number
- WO2010004677A1 WO2010004677A1 PCT/JP2009/002165 JP2009002165W WO2010004677A1 WO 2010004677 A1 WO2010004677 A1 WO 2010004677A1 JP 2009002165 W JP2009002165 W JP 2009002165W WO 2010004677 A1 WO2010004677 A1 WO 2010004677A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- color
- polarization
- information
- light source
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
- H04N23/84—Camera processing pipelines; Components thereof for processing colour signals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N25/00—Circuitry of solid-state image sensors [SSIS]; Control thereof
- H04N25/10—Circuitry of solid-state image sensors [SSIS]; Control thereof for transforming different wavelengths into image signals
- H04N25/11—Arrangement of colour filter arrays [CFA]; Filter mosaics
- H04N25/13—Arrangement of colour filter arrays [CFA]; Filter mosaics characterised by the spectral characteristics of the filter elements
- H04N25/134—Arrangement of colour filter arrays [CFA]; Filter mosaics characterised by the spectral characteristics of the filter elements based on three different wavelength filter elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Definitions
- the present invention relates to an image processing technique, and more particularly, to a technique for improving the accuracy of image division necessary for obtaining shape information of an object and image composition.
- the “appearance” of an object is composed of multiple components, such as a specular reflection component in which incident light reflected from the surface of the subject is observed as a “light”, and a diffuse reflection component observed by repeated scattering inside the object. It is known that it is constituted by.
- the image of the subject that we are observing is usually formed by the sum of various light components that reach the eye through a plurality of different processes.
- Such components are, for example, “specular reflection component” and “diffuse reflection component”.
- the luminance can be separated into, for example, a “specular reflection component” and a “diffuse reflection component”.
- a pixel in which the specular reflection component is dominant is treated as a pixel constituting the “specular reflection region”
- a pixel in which the diffuse reflection component is dominant is handled as a pixel constituting the “diffuse reflection region”.
- image component separation methods for separating an image into such components have been widely used in order to obtain information compression for digital archives and to acquire the shape and surface material of a subject (for example, Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document 3).
- image separation technique A technique for estimating and separating the ratio of each component for each pixel in an image is an image component separation technique (image separation technique).
- model-based image processing that expresses separated images as individual models is widely known.
- Model-based image synthesis is a technique widely used in the field of CG (Computer Graphics). This is because the appearance of the object is modeled as a function of the viewpoint position, the light source position, and the normal direction of the subject.
- CG Computer Graphics
- Various image processing for obtaining an image of a light source position (light source environment change image) or the like is possible.
- efficient data compression is also possible.
- a method is known in which an image is separated into a specular reflection component and a diffuse reflection component, and the specular reflection component is a Cook-Torrance model, and the diffuse reflection component is a Lambertian model.
- linearly polarized light is projected onto the subject, and the polarization filter installed between the camera and the subject is rotated in a plane perpendicular to the optical axis, and reflected by the subject.
- the light is separated into a specular reflection component and a diffuse reflection component (for example, Patent Document 1).
- this method has a problem that a specular reflection component that should not exist originally is detected in a region where the emission angle is sufficiently large (near the shielding edge). Therefore, sufficient accuracy cannot be achieved by the diffuse reflection component / specular reflection component separation method using polarization information.
- Non-Patent Document 4 a method using color information is widely known as a dichroic reflection model (for example, Non-Patent Document 4). This is because the specular reflection component and the diffuse reflection component are separated by utilizing the fact that the color vector of the specular reflection component is observed as the light source color vector while the color vector of the diffuse reflection component becomes the object color vector. .
- the light source color vector is known because it is white or can be observed by another method.
- this method cannot uniquely separate the specular reflection component and the diffuse reflection component unless the object color vector of the subject is known. For this reason, it cannot be used to accurately separate the specular reflection component and the diffuse reflection component.
- the technique using polarization information described in Patent Document 1 has a problem that a specular reflection component is observed in the vicinity of a shielding edge where a specular reflection component should not be observed.
- the specular reflection component and the diffuse reflection component are uniquely separated from the dichroic reflection model. I could't.
- the present invention has been made in order to solve the above-described problem, taking into consideration the presence of the polarization component of the diffuse reflection component, and performing component separation with higher accuracy than using the color information of the light source.
- An object is to provide an image processing technique.
- An image processing apparatus of the present invention is an image processing apparatus that separates components of an image of a subject by imaging the subject, and a light projecting unit that projects linearly polarized light emitted from a light source onto the subject; For each of the color polarization acquisition unit that acquires the color polarization image of the subject and the unit pixels that form the color polarization image, the luminance of the light that has passed through a polarizer having a polarization main axis direction of 3 or more and the direction of the polarization main axis Based on the correspondence relationship between the polarization information processing unit that generates color polarization information, the light source color information acquisition unit that acquires color information of the light source, the color polarization information and the color information of the light source, An image component separation unit that performs component separation of the image.
- a color information processing unit that generates a color image from the color polarization image is provided, and the image component separation unit performs component separation of the color image.
- the color polarization acquisition unit transmits light through three or more polarizers having different directions of polarization main axes, a color filter disposed at a position facing the polarizer, and the polarizer and the color filter. And an image sensor for receiving light.
- the polarization information processing unit generates at least one of a polarization minimum color component and a polarization amplitude color component as the color polarization information.
- the image component separation unit separates a color vector of each pixel constituting the subject image into a diffuse reflection component and a specular reflection component.
- the image component separation unit separates a color vector of each pixel constituting the subject image into a diffuse reflection non-polarization component, a diffuse reflection polarization component, and a specular reflection polarization component.
- the image component separation unit separates at least a part of the image of the subject into a shaded area.
- the image component separation unit separates the color polarization information by the light source color vector and the color vector of the minimum polarization color component.
- the image component separation unit separates the polarization amplitude color component by the light source color vector and the color vector of the minimum polarization color component.
- the image component separation unit separates the color image by the light source color vector and the color vector of the minimum polarization color component.
- the color polarization acquisition unit performs a synchronization process with the light projecting unit.
- the light projecting unit is arranged apart from the color polarization acquiring unit.
- An image processing system is an image processing system that includes a light projecting device and an image processing device, and that performs component separation of an image of the subject by imaging the subject, and the light projecting device emits light from a light source.
- a light projecting unit that projects the linearly polarized light that has been transmitted onto the subject, and the image processing device receives light transmitted through a polarizer in three or more directions having different polarization principal axes through a color filter.
- a color polarization acquisition unit that acquires a color polarization image of a subject, a color information processing unit that generates a color image from the color polarization image, and a unit pixel that constitutes the color polarization image are transmitted through the polarizer.
- a polarization information processing unit that generates color polarization information based on a correspondence relationship between the luminance of the measured light and the direction of the polarization main axis, and light source color information acquisition that acquires color information of the light source. And parts, based on the color and polarization information and the light source color information, and an the image component separation unit for performing component separation of a color image.
- An image separation system is an image separation system that includes a light projecting device and an image processing device, and performs component separation of an image of the subject by imaging the subject, and the light projecting device emits light from a light source.
- a communication unit that receives the light from the light projecting device, and a color polarization acquisition unit that acquires a color polarization image of the subject by receiving light passing through a polarizer in three or more directions having different polarization principal axes through a color filter;
- the color information processing unit that generates a color image from the color polarization image, and the luminance of the light transmitted through the polarizer and the polarization for each of the unit pixels constituting the color polarization image
- a polarization information processing unit that generates color polarization information, a
- the communication unit transmits and receives color information of the light source in addition to a signal notifying light projection, and the light source color information acquisition unit acquires color information of the light source from the communication unit.
- the image processing method of the present invention is an image processing method for separating components of an image of the subject by imaging the subject, and a light projecting step of projecting linearly polarized light emitted from a light source onto the subject;
- a color polarization acquisition step for acquiring a color polarization image of a subject by receiving light transmitted through a polarizer in three or more directions having different directions of polarization main axes through a color filter, and generating a color image from the color polarization image Color information regarding received polarization using a correspondence relationship between the luminance of the light transmitted through the polarizer and the direction of the polarization main axis for each of the color information processing step and the unit pixel constituting the color polarization image.
- a polarization information processing step for generating certain color polarization information; a light source color information acquisition step for acquiring light source color information; and Using information and the color information of the light source, and an image component separation step of performing component separation of the color image.
- the program according to the present invention is a program for an image processing apparatus that separates an image of a subject by imaging the subject, and causes a computer to execute the steps included in the image processing method.
- a model-based image synthesis device includes a parameter estimation device and an image synthesis device, and is a model-based image synthesis device that synthesizes images.
- the parameter estimation device includes an image imaging unit that images a subject; An image separation unit that performs component separation of the image captured by the image capturing unit, and light source information including at least one of the direction and position of the light source that irradiates the subject, luminance, color, and spectrum information.
- the image synthesis device includes a viewpoint / light source information acquisition unit that acquires a viewpoint of the image to be synthesized and light source information.
- a rendering unit that synthesizes an image in accordance with the viewpoint and light source information acquired by the viewpoint / light source information acquisition unit, using reflection parameters held in the parameter DB.
- the model-based image synthesis method of the present invention includes a parameter estimation step and an image synthesis step, and is a model-based image synthesis method for synthesizing images, wherein the parameter estimation step includes an image imaging step for imaging a subject,
- the image processing method according to Item 16 wherein an image separation step is performed to separate components of an image captured by the image capturing unit, a light source information estimation step that estimates light source information, and normal information or three-dimensional information on the surface of the subject.
- a shape information acquisition step for acquiring position information as shape information; and a parameter estimation step for estimating a reflection model parameter for each component divided by the image separation unit from the imaged subject.
- a viewpoint / light source information acquisition step for acquiring viewpoints and light source information of an image to be combined; Using the reflection parameter estimated by the acquisition step, and a rendering step of synthesizing an image conforming to the viewpoints and the light source information obtaining acquired viewpoint and light source information in steps.
- the image processing of the present invention it is possible to accurately separate and obtain the specular reflection component and the diffuse reflection component for each pixel by using two pieces of information of polarization information and color information. Further, by using image component separation using two types of information, polarization information and color information, for parameter estimation in a model-based image synthesis method, it is possible to synthesize an image faithful to the subject.
- FIG. 1 is a block diagram of an image processing apparatus according to a first embodiment of the present invention. It is a flowchart which shows the flow of a process of the image separation method which concerns on the 1st Embodiment of this invention.
- 1 is a configuration example of a camera equipped with an image processing apparatus according to a first embodiment of the present invention.
- 3 is a block diagram illustrating configurations of a color polarization acquisition unit, a color information processing unit, and a polarization information processing unit Z according to the first embodiment of the present invention.
- FIG. It is a schematic diagram which shows the basic composition of the color polarization acquisition part of this invention. It is a schematic diagram which shows the example of the pixel arrangement
- FIG. 12 is a diagram in which the image of FIG. 11 is divided into a plurality of areas according to brightness levels. It is a flowchart which shows the flow of a process of the color information processing part 103 and the polarization information processing part 104 in this invention. It is a schematic diagram for demonstrating the light source estimation method using a specular sphere.
- FIG. 1 is a configuration example of a camera equipped with an image processing apparatus according to the present embodiment. It is a block diagram in the image processing apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the flow of a process of the image separation method which concerns on the 2nd Embodiment of this invention. It is a figure which shows the structural example of the camera by which the image processing apparatus which concerns on the 2nd Embodiment of this invention is mounted.
- FIG. 1 is a graph which shows the relationship between the brightness
- (b) is an image of a to-be-photographed object.
- FIG. In order to explain the problems of the conventional method, it is a schematic diagram in which the relationship between the brightness Is of the specular reflection component and the intermediate vector ⁇ is obtained using a conventional image processing apparatus. It is the synthesized image created using the image separation method using the conventional polarization information. It is a schematic diagram which shows the model parameter hold
- FIG. 1 is a block diagram of the image processing apparatus according to the present embodiment.
- This image processing apparatus is an apparatus that separates components of an image of the subject by capturing an image of the subject.
- the image processing device and the light projecting unit 101 that projects linearly polarized light onto the subject have three or more different polarization principal axis angles.
- a color polarization acquisition unit 102 that acquires a color polarization image of a subject by receiving light transmitted through the polarizer through a color filter, and a color information processing that generates a color image from the image acquired by the color polarization acquisition unit 102
- the unit 103 and the image acquired by the color polarization acquisition unit 102 receive light for each unit pixel constituting the polarization image using a correspondence relationship with the luminance of light transmitted through the polarizer in three or more directions.
- Polarization information processing unit 104 that generates color polarization information that is color information related to polarization, and light source color information that acquires color information of a light source that projects a subject Using the polarization information generated by the acquisition unit 105 and the polarization information processing unit 104 and the light source color information acquired by the light source color information acquisition unit 105, component separation of the color image acquired by the color information processing unit 103 is performed.
- An image component separation unit 106 and an output unit that outputs a signal generated by the image separation unit 106 are provided.
- the “polarized image” is an image formed by light transmitted through a polarizing filter (polarizer) having a specific polarization principal axis angle. It is possible to obtain a different “polarized image” for each of a plurality of polarization principal axis angles.
- the “color polarization image” is an image formed by light of a plurality of colors having different principal wavelengths among light transmitted through a polarization filter (polarizer) having a specific polarization principal axis angle.
- red (R), green (G), and blue (B) color polarization images are obtained based on light transmitted through a polarizing filter (polarizer) having a certain polarization principal axis angle.
- a color vector composed of R, G, and B luminances is determined for each unit pixel.
- the color vector of the color image is also determined. “Component separation” in this specification corresponds to representing one color vector as the sum of a plurality of color vectors (components).
- FIG. 2 is a flowchart showing the image processing operation according to the present embodiment.
- the light projecting unit 101 projects polarized light onto the subject.
- the color polarization acquisition unit 102 acquires a color polarization image including color polarization information by receiving a subject with an imaging device through a patterned polarizer and a color filter.
- the patterned polarizer has a polarization principal axis angle (rotation angle of the polarization transmission axis) of three or more directions as will be described later.
- step S103 the color information processing unit 103 acquires color image information using information output from the color polarization acquisition unit 102.
- the polarization information processing unit 104 uses the information output from the color polarization acquisition unit 102 to generate a polarization minimum color component I min and a polarization amplitude color component I amp described later as polarization information (step S104).
- the light source color information acquisition unit 105 acquires the color vector I Light of the light source that irradiates the subject (step S105).
- the order of step S103, step S104, and step S105 is arbitrary, and may be executed in parallel or sequentially.
- step S106 the image component separation unit 106 separates the polarization amplitude color component I amp generated by the polarization information processing unit 104 into a polarization amplitude light source color vector component I amp1 and a polarization amplitude polarization minimum color vector component I amp2 . Further, in step S107, the image component separation unit 106 uses the polarization amplitude polarization minimum color component I min generated by the polarization information processing unit 104 as a diffuse reflection non-polarization component, and diffuse polarization reflects the polarization amplitude minimum polarization color vector component I amp2. As a polarization component, component separation is performed using a polarization amplitude light source color vector component I amp1 as a specular reflection polarization component.
- the color polarization acquisition unit 102, the color information processing unit 103, the polarization information processing unit 104, the light source color information acquisition unit 105, and the image component separation unit 106 are realized by the CPU 205 shown in FIG. 3 executing a program. Shall be. However, all or part of these functions may be realized by hardware.
- 3 includes information acquired by the color polarization acquisition unit 102, color image information acquired by the color information processing unit 103, polarization information acquired by the polarization information processing unit 104, and a light source color information acquisition unit.
- the light source color vector information acquired by 105 is stored.
- FIG. 3 is a block diagram showing a configuration example of such a camera.
- the camera in FIG. 3 includes a patterned polarizer 201, a color imaging device 208, a memory 204, a CPU 205, a light emitting device 206, and a polarizer 207.
- the color imaging device 208 includes a color filter 202 and an imaging device 203.
- the light projecting unit 101 uses the light emitting device 206 and the polarizer 207 to project polarized light onto the subject.
- the light emitting device 206 uses a polarizing filter (polarizer 207) in front of the flash.
- polarizer 207 As means for using polarized light, a liquid crystal polarizer or the like may be used.
- the color polarization acquisition unit 102 receives a subject with an imaging element through the patterned polarizer 201 and the color filter 202, thereby acquiring a color polarization image that is a color image including color polarization information.
- the color information processing unit 103 calculates color image information using information output from the color polarization acquisition unit 102. Further, the polarization information processing unit 104 calculates polarization information using the information output from the color polarization acquisition unit 102. This process will be described in detail.
- FIG. 4 is a block diagram illustrating configurations of the color polarization acquisition unit 102, the color information processing unit 103, and the polarization information processing unit 104.
- Color image information and polarization image information are acquired from the subject in real time, and polarization minimum color information that is a non-polarization component is output as polarization information.
- Incident light that has passed through the lens 220 and the diaphragm 221 enters the color polarization acquisition unit 102. From this incident light, the color polarization acquisition unit 102 can acquire both color image information and polarization image information in real time.
- Color information processing section 103 and a polarization information processing section 104 performs various processes on the signal, and outputs a color image lm, the polarization minimum color information I min.
- FIG. 5 is a schematic diagram showing a basic configuration of the color polarization acquisition unit 102.
- the color filter 202 and the patterned polarizer 201 are disposed so as to overlap the front surface of the image sensor pixel 203.
- the order of installing the color filter and the patterned polarizer is arbitrary.
- the incident light passes through the color filter 202 and the patterned polarizer 201 and reaches the image sensor, and the luminance is observed by the image sensor pixel 203.
- FIG. 6A is a view of a part of the imaging surface of the color polarization acquisition unit 102 as viewed from directly above the optical axis direction.
- FIG. 6A shows only 16 pixels (4 ⁇ 4) on the imaging surface.
- the four rectangular areas 301 to 304 shown in the figure indicate corresponding portions of the Bayer type color mosaic filter installed on the four pixel cells, respectively.
- a rectangular area 301 is a B (Blue) filter area and covers the pixel cells B1 to B4.
- B patterned polarizers having different polarization main axes are in close contact with the pixel cells B1 to B4.
- the “polarization main axis” is an axis parallel to the polarization plane (transmission polarization plane) of light transmitted through the polarizer.
- polarizer units having transmission polarization planes of different angles in the same color pixel are arranged adjacent to each other. More specifically, four types of polarizer units having different transmission polarization plane directions are arranged in pixels of the same color of R, G, and B. One polarizer unit corresponds to one minute polarization pixel. In FIG. 6A, a code such as G1 is given to each polarization pixel.
- FIG. 6B shows the polarization main axes assigned to the four finely polarized pixels with which the patterned polarizer for B is in close contact.
- the straight line described in each finely polarized pixel schematically shows the polarization main axis direction of the minute polarizing plate.
- patterned polarizers are in close contact with the pixels in the rectangular areas 302 and 304, respectively, and four R (Red: red) patterned elements are in the pixels of the rectangular area 303.
- the polarizer is in close contact.
- a position indicated by reference numeral “305” indicates a virtual pixel position in which four pixels in the imaging system are collectively displayed.
- the patterned polarizers of the rectangular regions 302 to 304 are also divided into portions having four different polarization main axes as shown in FIG.
- FIG. 7A is a diagram illustrating another example of the pixel array in the color polarization acquisition unit 102.
- G pixels are arranged in a cross shape in a 3 ⁇ 3 block inclined by 45 °, and R and B are alternately arranged in four pixels around the G pixel.
- FIG. 7B shows the fine structure of each color pixel, and each color pixel is composed of four types of fine polarization pixels.
- each color pixel includes a plurality of finely polarized pixels having different polarization principal axes, and the color mosaic arrangement itself is arbitrary.
- each finely polarized pixel is referred to as a “polarized pixel”.
- the 8A to 8C are graphs schematically showing the wavelength characteristics of the B, G, and R polarization pixels, respectively.
- the vertical axis of each graph is the intensity of transmitted light, and the horizontal axis is the wavelength.
- the polarization pixels for B, G, and R have polarization characteristics that transmit TM (Transverse Magnetic Wave) waves and reflect (do not transmit) TE (Transverse Electric Wave) waves in the B, G, and R wavelength bands.
- TM wave Transverse Magnetic Wave
- TE Transverse Electric Wave
- FIG. 8A shows the polarization characteristics 402 and 403 of the B-polarized image and the transmission characteristics 401 of the B color filter.
- Polarization characteristics 402 and 403 indicate the transmittances of the TM wave and the TE wave, respectively.
- FIG. 8B shows the polarization characteristics 405 and 406 of the G polarization image and the transmission characteristics 404 of the G color filter.
- Polarization characteristics 405 and 406 indicate the transmittances of the TM wave and the TE wave, respectively.
- FIG. 8C shows the polarization characteristics 408 and 409 of the R-polarized image and the transmission characteristics 407 of the R color filter.
- Polarization characteristics 408 and 409 indicate the transmittances of the TM wave and the TE wave, respectively.
- the characteristics as shown in FIG. 8A to FIG. 8C are, for example, “Kawashima, Sato, Kawakami, Nagashima, Ota, Aoki,“ Development of polarization imaging device and utilization technology using patterned polarizer ”, It can be realized by using a photonic crystal described in the IEICE General Conference 2006, No. D-11-52, P52, 2006. In the case of a photonic crystal, light having an electric field vector oscillation plane parallel to a groove formed on the surface thereof is TE wave, and light having an electric field vector oscillation plane is TM wave.
- FIGS. 8A to 8C a patterned polarizer showing polarization separation characteristics in each of the B, G, and R transmission wavelength bands is used. is there.
- FIG. 9 shows a case where the wavelength is shifted between the transmission region of the G color filter and the polarization separation region determined by the polarization characteristics 410 and 411. According to the polarizer exhibiting such characteristics, the intended operation of the present invention cannot be performed.
- the luminance dynamic range and the number of bits of the image sensor be as large as possible (for example, 16 bits) in order to reliably acquire a polarization component included in a particularly bright specular reflection portion of the subject and a polarization component included in the shadow region of the subject. .
- the color polarization acquisition unit 102 performs synchronization processing with the light projection control unit 101.
- the color polarization acquisition unit 102 may perform imaging immediately after the light emission timing of the light projecting unit 101.
- the luminance information acquired for each polarization pixel with the configuration shown in FIG. 5 is processed by the polarization information processing unit 104 in FIG. Hereinafter, this process will be described.
- the observation luminance when the rotation angle ⁇ of the polarization main axis is ⁇ i is I ( ⁇ i).
- i is an integer greater than or equal to 1 and less than or equal to N, and “N” is the number of samples.
- FIG. 10 shows luminance 501 to 504 corresponding to samples ( ⁇ i, Ii ( ⁇ i)) of 4 pixels.
- the relationship between the angle ⁇ i of the polarization main axis and the luminance 501 to 504 is expressed by a sine function curve.
- a sine function curve In FIG. 10, four points of luminance 501 to 504 are described so as to be positioned on one sine function curve. However, when the sine function curve is determined based on more observation luminances, one of the observation luminances is shown. It is possible that the part slightly deviates from the sine function curve.
- polarization information in this specification means amplitude modulation degree and phase information in a sinusoidal curve indicating the dependence of luminance on the polarization principal axis angle.
- the reflected light luminance I with respect to the principal axis angle ⁇ of the patterned polarizer is approximated as follows using the internal four pixel luminances for each of the same color regions 301 to 304 shown in FIG. 6A as samples. To do.
- Equation 1 can be expanded as follows.
- a and B are shown by the following (Formula 3) and (Formula 4), respectively.
- the three parameters A, B, and C of the sine function approximation are determined for one color.
- the image processing apparatus of the present embodiment outputs the minimum polarization luminance I min and the polarization amplitude luminance I amp , but the output polarization information is other information as long as it is information obtained from the sine function of FIG. It may be a pair.
- the polarization degree image ⁇ , the polarization phase image ⁇ max , the polarization estimation error E, the value of the maximum luminance I max of the sine function, and combinations thereof may be output as polarization information.
- the degree of polarization ⁇ represents the degree to which the light of the relevant pixel is polarized
- the polarization phase ⁇ max represents the principal axis angle of the partial polarization of the light of the relevant pixel.
- the polarization estimation error is the total difference between the luminance observed for the 4-pixel sample and the luminance determined from the approximation obtained by the approximation.
- the principal axis angle of polarized light is the same between 0 and 180 ° ( ⁇ ).
- the color information processing unit 103 uses the information output from the color polarization acquisition unit 102 to calculate color luminance.
- the luminance of the light transmitted through the polarizer is different from the original luminance of the light before entering the polarizer.
- a value obtained by averaging the observed luminances of all polarized polarization main axes corresponds to the original luminance of light before entering the polarizer.
- an image composed of a set of unit pixels having such color luminance is expressed as a color image I.
- a normal color mosaic image can be generated by obtaining the luminance in each polarization pixel.
- a color image I is generated by converting each pixel into a color image having RGB pixel values based on the mosaic image. Such conversion is realized using a known interpolation technique such as a Bayer mosaic interpolation method.
- the polarization minimum color component I is expressed as “color information” expressed by a color vector composed of three luminances of RGB separately from the polarization minimum luminance I min and the polarization amplitude luminance I amp which are polarization information. min and polarization amplitude color component I amp can be obtained. At this time, the following relationship is established between the color image I and the color information.
- FIG. 11A shows only the luminance of the color image I (x, y) of the grapefruit (subject).
- FIGS. 11B and 11C show examples of images of the minimum polarization luminance I min (x, y) and the polarization amplitude luminance I amp (x, y) for the subject in FIG. 11A, respectively. ing. Originally, these images are color images, but are monochromeized.
- FIG. 12 is a diagram schematically showing each figure of FIG. 11 (a diagram in which the shading is clarified). In this figure, each area (A01 to B06) corresponds to each area in FIG.
- the color image I (x, y) and the luminance and polarization information of each pixel are obtained using the four polarization pixels shown in FIG. 6B, the individual luminance and polarization information are shown in FIG. It can be considered that the value at the virtual pixel point 305 located at the center of the four polarization pixels shown in FIG. Therefore, the resolutions of the color image and the polarization image are both reduced to 1 ⁇ 2 ⁇ width 1 ⁇ 2 of the resolution of the image sensor.
- the color polarization acquisition unit 102 first acquires a color image and a polarization information image in real time.
- steps S201 to S203 observation values of a plurality of polarization luminances at the R, G, and B pixels of the color image are acquired.
- the order of steps S201 to S203 is arbitrary and may be executed in parallel. Specifically, four types of polarization luminance are acquired in the R, G, and B color mosaic pixels.
- the signal indicating the polarization luminance is sent to the polarization information processing unit 104 and processed as follows in steps S204 to S208.
- a sine function parameter is calculated based on the varying luminance obtained from each of the R pixel, the G pixel, and the B pixel.
- the sine function parameters are defined by A, B, and C in Equation 1 above. Since the processes in steps S204 to S206 are also independent of each other, they can be performed in any order and may be executed in parallel.
- the polarization information processing unit 104 obtains the minimum luminance and amplitude component of the sine function, thereby generating the polarization minimum luminance I min that is a non-polarization component and the polarization amplitude luminance I amp that is a polarization component.
- the color information processing unit 103 executes the process of step S208. Specifically, the average luminance of R, G, and B is obtained using the above-described equation 12, and a color luminance image I (x, y) is generated.
- the polarizing element may be a film-type polarizing element, a wire grid type, or a polarizing element based on other principles.
- a fobion element or the like may be used instead of using a color filter.
- the configuration example of the camera equipped with the image processing apparatus according to the present embodiment in FIG. 3 is configured such that the color filter 202 and the imaging device 203 are used as the color imaging device 208.
- the light source color information acquisition unit 105 acquires the color information of the light source that irradiates the subject.
- the color information of the light source is known.
- the color information of the light emitting device 206 may be held in the memory 204 of the image processing device, and the light source color information acquisition unit 105 may call the light source color information from the memory 204.
- the light source color information acquisition unit 105 for example, a target with a known shape and surface reflectance for estimating the light source information near the subject.
- color information can be estimated from the image captured by the color imaging device 208 (for example, “Masayuki Kambara, Naokazu Yokoya,” vision-based expansion considering optical consistency by real-time estimation of the light source environment) Reality ", IEICE Technical Report, Pattern Recognition / Media Understanding, PRMU2002-190, pp.7-12, 2003”). This process will be described in detail.
- the light source color information is acquired using, for example, a sphere 601 that can be regarded as a mirror surface shown in FIG.
- a specular sphere 601 is installed near the subject, and its position and normal direction are known.
- the specular sphere 601 is imaged by the color polarization acquisition unit 102 and the color information processing unit 103 acquires a color image.
- the shooting environment is reflected in the specular sphere 601.
- a high-luminance pixel is detected as a specular reflection pixel from the captured specular image. Since the reflectance of the specular surface is also known, the color vector information of the light source can be acquired by detecting the color information of the specular reflection pixel.
- light source color information obtained by shooting before may be used. This is effective when the light source environment does not change like an indoor surveillance camera. In such a case, the target may be photographed when the camera is installed to obtain the light source information. Processing may be performed assuming that the light source is white. This is effective when taking an image under fluorescent lamp illumination, or taking an image outdoors in the daytime.
- the image component separation unit 106 uses the polarization information generated by the polarization information processing unit 104 and the light source color vector information acquired by the light source color information acquisition unit 105 to generate a color image acquired by the color information processing unit 103. Separate the components.
- This method depends on the difference in polarization characteristics between specular reflection and diffuse reflection as follows. -Since the specular reflection component is generated by surface reflection, the polarization characteristic of incident light is maintained. Therefore, it is observed as a polarized component of luminance observed by the camera. ⁇ Diffusion reflection has repeated scattering, so the polarization characteristics of incident light are lost. Therefore, it is observed as a non-polarized component of luminance observed by the camera.
- These polarization characteristics are based on the following two conditions. (Condition 1) When linearly polarized light is projected, the specular reflection component is observed as a polarization component. (Condition 2) When linearly polarized light is projected, the diffuse reflection component is observed as a non-polarized component.
- the image of FIG. 11B is treated as indicating a diffuse reflection component
- the image of FIG. 11C is treated as indicating a specular reflection component.
- a specular reflection component that should not originally exist is detected in a region where the emission angle is sufficiently large (near the shielding edge). From this, it can be determined that the diffuse reflection component / specular reflection component separation method using polarization information cannot achieve sufficient accuracy.
- the horizontal axis in FIG. 16 represents the incident angle
- the vertical axis represents the specular reflection component polarization degree
- the horizontal axis in FIG. 17 represents the emission angle
- the vertical axis represents the diffuse reflection component polarization degree.
- This figure shows how much outgoing light and reflected light are polarized when non-polarized light is projected onto the subject surface. That is, as the degree of polarization is closer to 0, the outgoing light (reflected light) becomes non-polarized, and as the degree of polarization is closer to 1, it approaches closer to linearly polarized light.
- This figure shows the following. -The polarization degree of the diffuse reflection component is sufficiently small except in a region where the emission angle is sufficiently large. In a region where the exit angle is sufficiently large, the degree of polarization of the diffuse reflection component is sufficiently larger than the degree of polarization of the specular reflection component.
- the image component separation unit 106 in the present embodiment solves the above problem by using color information.
- the image component separation unit 106 in the present embodiment will be described in detail.
- the color vector I (x, y) of the subject in the color image observed at the pixel (x, y) can be decomposed as follows.
- I Light represents the color vector of the light source
- I Object (x, y) represents the object color vector at the pixel (x, y).
- the color vector is a three-dimensional vector that expresses the brightness of each color of R, G, and B as a vector.
- C′1 (x, y) and C′2 (x, y) indicate the weights of the light source color vector and the object color vector. From this equation, it can be seen that the specular reflection component I s (x, y) and the diffuse reflection component I d (x, y) at the pixel (x, y) can be separated as in the following equation.
- the image component separation unit 106 solves this problem by using the polarization minimum color information I min obtained from the polarization information processing unit 104 as the object color vector.
- the image component separation unit 106 in the present embodiment separates the specular reflection component and the diffuse reflection component by using the following expression.
- the color image I (x, y) uses the fact that it is the sum of the doubled polarization amplitude color component I amp and the polarization minimum color information I min . That is, the image component separation unit 106 separates the polarization amplitude color component I amp into a light source color vector component and a polarization minimum color vector component. By performing this processing, the image component separation unit 106 converts the color image I (x, y) into the polarization amplitude light source color vector component I amp1 (x, y) and the polarization amplitude polarization as follows according to Expression 17. Separate into a minimum color vector component I amp2 (x, y) and a polarization minimum color component I min (x, y).
- I (x, y) is separated into I amp1 (x, y), I amp2 (x, y), and I min (x, y) for each pixel.
- I amp1 (x, y), I amp2 (x, y), and I min (x, y) in the pixel at the position (x, y) (514, 432) shown as an example in FIG. [7269, 7132, 6505], [4151, 1705, 821], and [11845, 4865, 2345], respectively.
- the three numerical values in [] are the luminances of the R, G, and B colors as described above.
- I amp1 (x, y), I amp2 (x, y), and I min (x, y) at the pixel at position (x, y) (780, 330) are [37, 37, respectively]. , 33], [1455, 530, 244], and [4207, 1532, 706].
- the polarization amplitude color component I amp (x, y) is expressed as follows.
- R amp (x, y), G amp (x, y), and B amp (x, y) are the R component, G component, and B component of the polarization amplitude color component I amp (x, y)
- R Light , G Light , and B Light are the R component, G component, and B component of the light source color vector I Light
- R min (x, y), G min (x, y), and B min (x, y) are the R component, G component, and B component of the polarization minimum color component I min (x, y).
- Equation 19 the color vector I (x, y) of the subject is obtained from the color image processing unit 103.
- the light source color vector I Light has already been obtained from the light source color information acquisition unit 105.
- the polarization minimum color information I min (x, y) has already been obtained from the polarization information processing unit 104. Therefore, in Equation 19, the number of equations is 3, while the unknowns are two, C1 (x, y) and C2 (x, y). Therefore, Expression 20 can obtain C1 (x, y) and C2 (x, y) from the following expressions by using the method of least squares.
- the image component separation unit 106 further separates each component separated by the polarization information processing unit 104 and the image component separation unit 106 in accordance with the component separation standard of FIG.
- FIG. 11E shows a polarization amplitude light source color vector component I amp1 (x, y) which is a “specular reflection polarization component”.
- I amp (x, y) in FIG. 11C is “specular reflection component” and I min (x, y in FIG. ) As a “diffuse reflection component”. Therefore, as described above, in a region where the exit angle is sufficiently large (near the shielding edge), there arises a problem that a specular reflection component that should not exist originally is detected. However, in the image separation method of the present embodiment, such a problem occurs. No problem arises.
- the light source color information acquisition unit 105 may acquire the color information of the light source by using the subject information acquired by the color information processing unit 103 and the polarization information processing unit 104. As described above, in the dichroic reflection model, the subject color vector I (x, y) is separated into a specular reflection component equal to the color vector of the light source and a diffuse reflection component having an object color vector.
- the diffuse reflection component is considered negligible.
- the light source color information acquisition unit 105 detects the pixel with the highest luminance in the minimum polarization luminance I min or the polarization amplitude luminance I amp acquired by the color information processing unit 103 and the polarization information processing unit 104, and The color vector of the minimum polarization luminance I min or the polarization amplitude luminance I amp may be estimated as the color vector of the light source.
- the image component separation unit 106 may separate the color image into a diffuse reflection component and a specular reflection component, for example, instead of separating the color image into a diffuse reflection non-polarization component, a diffuse reflection polarization component, and a specular reflection polarization component. . This can be done by modifying Equation 17 as follows.
- FIG. 20 is a diagram showing component separation criteria in this case.
- FIG. 21 is a flowchart showing the flow of processing of the image separation method according to this embodiment.
- steps common to FIG. 2 are denoted by the same reference numerals as in FIG. 2, and detailed description thereof is omitted here.
- the pixel with the highest luminance is detected, and the color vector of that pixel is estimated as the color vector of the light source.
- the image component separation unit 106 uses the minimum polarization color component I min generated by the polarization information processing unit 104 as the diffuse reflection non-polarization component and the minimum polarization amplitude polarization color generated by the minimum polarization color component I min .
- the component separation is performed using the vector component I amp2 as the diffuse reflection component and the polarization amplitude light source color vector component I amp1 as the specular reflection component (step S108).
- FIG. 11E shows a “specular reflection component”
- FIG. 11F shows a “diffuse reflection component”.
- the image component separation unit 106 does not separate the polarization amplitude color component I amp into the light source color vector component and the minimum polarization color vector component, but converts the color image acquired by the color polarization acquisition unit 102 into the light source color vector component and the minimum polarization color component. It may be separated into color vector components.
- FIG. 22 is a diagram showing the component separation criteria in this case.
- FIG. 23 is a flowchart showing the flow of processing of the image separation method according to this embodiment. In FIG. 23, steps common to those in FIG. 2 are denoted by the same reference numerals as those in FIG. 2, and detailed description thereof is omitted here.
- the polarization information processing unit 104 generates the minimum polarization color component I min as polarization information using the information output from the color polarization acquisition unit 102 (step S113).
- the order of step S103 and step S113 is arbitrary, and may be executed in parallel or sequentially.
- the light source color information acquisition unit 105 acquires the color vector I Light of the light source that irradiates the subject (step S105).
- the color vector of the light source can be obtained using the various methods described above.
- the image component separation unit 106 separates the color image I acquired by the color information processing unit 103 into the color image light source color vector component I1 and the color image polarization minimum color vector component I2 (step S114). Further, the image component separation unit 106 performs component separation using the color image polarization minimum color vector component I2 as a diffuse reflection non-polarization component and the color image light source color vector component I1 as a specular reflection component (step S115).
- the image component separation unit 106 may perform image separation in consideration of shadows. This is because, in shadows, light rays become complicated due to the influence of multiple reflection and the like, and the reliability of polarization information is lost. Therefore, it is considered that the image separation accuracy deteriorates in the shadow region.
- FIG. 24 is a diagram showing component separation criteria in this case.
- FIG. 25 is a flowchart showing the flow of processing of the image separation method according to this embodiment. In FIG. 25, steps common to those in FIG. 2 are denoted by the same reference numerals as those in FIG. 2, and detailed description thereof is omitted here.
- the image component separation unit 106 determines whether the luminance of the pixel is equal to or less than a threshold value in order to estimate whether the pixel is a shadow (step S109). Since the shadow area has low luminance, a pixel whose luminance of the color image obtained by the color information processing unit 103 is equal to or less than a threshold value can be determined as a shadow.
- the threshold value may be determined experimentally. For example, 256 may be set for a 16-bit monochrome image.
- the luminance for detecting the shadow the polarization maximum information I max and the minimum polarization luminance I min acquired by the polarization information processing unit 104, or the average value and weighting of the maximum polarization luminance I max and the minimum polarization luminance I min are obtained.
- I max + I min is an image equivalent to an image captured when no polarizer is installed under a linearly polarized light source. Therefore, by performing image processing using I max + I min , it is possible to perform the same processing as when normal polarization is not used. Further, when the minimum polarization luminance I min is used, the luminance becomes very high, and the influence of the specular reflection component that is likely to cause overexposure can be reduced, which is very effective. If the luminance of the pixel is smaller than the threshold (Yes in step S0109), it is determined that the pixel is a shadow component (step S110), and the process ends.
- the image component separation unit 106 uses the polarization amplitude color component I amp as the polarization amplitude light source color as described above.
- the component is separated into the vector component I amp1 and the polarization amplitude polarization minimum color vector component I amp2 (step S106), and the polarization minimum color component I min generated by the polarization information processing unit 104 is used as the diffuse reflection non-polarization component and polarization amplitude polarization minimum.
- Component separation is performed using the color vector component I amp2 as a diffuse reflection polarization component and the polarization amplitude light source color vector component I amp1 as a specular reflection polarization component (step S107).
- the light emitting device 206 is desirably arranged as far as possible from the imaging device 203.
- the region where the specular reflection component is observed is a region where the incident angle is near 0 degrees. This is because the specular reflection component is observed in the vicinity of the regular reflection region.
- the specular reflection component is hardly polarized as shown in FIG. Therefore, the polarization minimum luminance I min (x, y) includes many specular reflection components, and the image component separation accuracy by the image generation / separation unit 106 is reduced.
- FIG. 26 is a block diagram illustrating another configuration example of the image processing apparatus according to the present embodiment. This is a block diagram for processing in which the light source color information acquisition unit 105 acquires the color information of the light source by using the subject information acquired by the color information processing unit 103 and the polarization information processing unit 104.
- the same components as those in FIG. 1 are denoted by the same reference numerals as those in FIG. 1, and detailed description thereof is omitted here.
- FIG. 27 is a flowchart showing a process flow of the image separation method according to the present embodiment. In FIG. 27, steps common to those in FIG. 2 are denoted by the same reference numerals as in FIG. 2, and detailed description thereof is omitted here.
- the difference from FIG. 2 is that the order of step S103, step S104, and step S105 is arbitrary in FIG. 2, but only the order of step S103 and step S104 is arbitrary in FIG. Step S103 and step S104 may be executed in parallel or sequentially.
- a photonic crystal is used for the patterned polarizer 201, but without using the patterned polarizer, shooting is performed while rotating the polarizing plate mounted in front of the lens of the imaging device. You may make it acquire the brightness
- This method is disclosed in, for example, Japanese Patent Application Laid-Open No. 11-212433.
- subject tracking technology widely used in image processing may be used (for example, JianboanShi and Carlo Tomasi, “Good“ Features to Track ”, IEEE Conferon Computer Vision and Pattern Recognition, pages 593-600, 1994).
- the color polarization acquisition unit 102 may acquire color image information and polarization information not by the same imaging device 203 but by individual imaging devices.
- FIG. 28 shows a configuration example of a camera on which the image processing apparatus according to the present embodiment is mounted when the color polarization acquisition unit 102 is configured by separate image sensors 203 and 203-2 for color image information and polarization information. Is shown. In this case, it is desirable to make the optical axes of the two imaging devices 203 and 203-2 equal by using a beam splitter or the like.
- image separation using polarization information and color information can be performed by using the image separation method of the present invention.
- image separation can also be performed by separating the specular reflection component and the diffuse reflection component in consideration of the polarization component of the diffuse reflection component.
- FIG. 29 shows a block diagram of the image separation system according to the present embodiment. 29, the same reference numerals as those in FIG. 1 are given to the same components as those in FIG. 1, and detailed description thereof is omitted here.
- the difference from the first embodiment is that the light projecting device 107 and the image processing device 108 are separated.
- FIG. 30 is a flowchart showing the flow of processing of the image separation method in the image processing apparatus according to this embodiment. In FIG. 30, steps common to FIG. 2 are denoted by the same reference numerals as in FIG. 2, and detailed description thereof is omitted here.
- FIG. 31 shows a configuration example of the camera and the light projecting device 107 on which the image processing apparatus 108 according to the present embodiment is mounted. In FIG. 31, the same components as those in FIG. 3 are denoted by the same reference numerals as those in FIG. 3, and detailed description thereof is omitted here.
- the image separation system includes a light projecting device 107 and an image processing device 108.
- the image separation system performs component separation of the image of the subject by imaging the subject.
- a light projecting unit 101 that projects polarized light onto the subject is provided.
- the image processing apparatus 108 receives light transmitted through a polarizer in three or more directions having different polarization principal axis angles through a color filter, thereby obtaining a color polarization acquisition unit 102 that acquires a color polarization image of the subject, and a color A color information processing unit 103 that generates a color image from an image acquired by the polarization acquisition unit 102 and a unit pixel that constitutes the polarization image from an image acquired by the color polarization acquisition unit 102 in each of the three directions or more.
- the polarization information processing unit 104 that generates color polarization information that is color information related to the received polarization and the color information of the light source that projects the subject are acquired.
- the light source color information acquisition unit 105 that performs the polarization information generated by the polarization information processing unit 104 and the light source that is acquired by the light source color information acquisition unit 105.
- an image component separating unit 106 for component separation of a color image obtained by the color information processing section 103.
- the light projecting device 107 and the image processing device 108 may perform synchronization processing, and the image processing device 108 may perform imaging using a synchronization signal from the light projecting device 107. This process will be described.
- FIG. 32 shows a block diagram of the image separation system according to the present embodiment.
- the same reference numerals as those in FIG. 29 are given to the same components as those in FIG. 29, and detailed description thereof is omitted here.
- the difference from FIG. 29 is that both the light projecting device 107 and the image processing device 108 have communication units 109 and 110.
- FIG. 33 is a flowchart showing the flow of processing of the image separation method in the light projecting device 107 and the image processing device 108 according to this embodiment.
- steps common to those in FIG. 2 are denoted by the same reference numerals as those in FIG. 2, and detailed description thereof is omitted here.
- FIG. 33 steps common to those in FIG. 2 are denoted by the same reference numerals as those in FIG. 2, and detailed description thereof is omitted here. Further, FIG.
- 34 shows a configuration example of a camera and a light projecting device on which the image processing apparatus according to this embodiment is mounted. 34, the same reference numerals as those in FIG. 31 are attached to the same components as those in FIG. 31, and detailed description thereof is omitted here.
- step S101 the light projecting device 107 projects polarized light onto the subject by the light projecting unit 101. Thereafter, the light projecting device 107 transmits a signal notifying light projection to the image processing device 108 through the communication device 209 (step S111).
- the image processing apparatus 108 receives the signal indicating the light projection by the communication apparatus 210 (step S112), as described above, the color polarization acquisition unit 102 passes the subject through the patterned polarizer 201 and the color filter 202 with the image sensor. By receiving the light, a color polarization image that is a color image including the color polarization information is acquired (step S102). As described above, polarization information, a color image, and a light source color vector are acquired through steps S103 to S107, and finally image component separation is performed.
- the communication unit 109 may not only transmit a signal notifying light projection to the image processing apparatus 108 but also transmit color vector information of the light source to the image processing apparatus 108.
- the light source color information processing unit 105 may acquire light source color vector information from the signal received by the communication unit 110.
- the image separation system separates the light projecting device 107 and the image processing device 108, and synchronizes the light projection and imaging by communication, thereby achieving a high-accuracy image while being a smaller image processing device 108. It is possible to achieve component separation.
- Model-based image composition using image separation The image component separation according to the present invention is particularly effective for model-based image composition processing used in digital archives and the like.
- Model-based image synthesis is important as an interactive method for presenting captured data because the light source direction and line-of-sight direction of the captured image can be arbitrarily changed. It is also very effective for giving a sense of realism. Therefore, model-based image composition processing can be used for applications such as virtual museums and AR (Augmented Reality).
- model-based image composition processing can be used for applications such as virtual museums and AR (Augmented Reality).
- the captured image is divided into a specular reflection component and a diffuse reflection component, and different models are used for each component. There was a problem that the image quality deteriorated due to insufficient separation of the diffuse reflection components. This problem is caused by the following.
- the image composition process of the present invention comprises a parameter estimation device 800 and an image composition device 801.
- the following five pieces of input information are used.
- ⁇ Diffuse reflection image of the subject ⁇ Specular reflection image of the subject ⁇ Three-dimensional shape information of the subject ⁇ Light source position / color / illuminance ⁇ Viewpoint / light source information in the composite image
- FIG. 35 is a block diagram showing the configuration of the parameter estimation device 800 and the image composition device 801 according to an embodiment of the present invention.
- 35 uses an imaging device, an image imaging unit 802 that performs imaging, an image separation unit 803 that separates an image into a specular reflection component and a diffuse reflection component by the above-described image separation method, and a subject
- a light source information estimation unit 804 for estimating light source information such as direction, position, luminance, color, spectrum information, etc., and shape information acquisition for acquiring normal information or three-dimensional position information on the surface of the subject as shape information
- a parameter estimation unit 806 for estimating a reflection model parameter, and a parameter DB (which holds the reflection parameter estimated by the parameter estimation unit 806) Database) and a 807.
- the image composition device 801 uses the viewpoint / light source information acquisition unit 808 that acquires the viewpoint and light source information of the image to be combined, and the model parameter information held in the parameter DB 807, and uses the viewpoint / light source information acquisition unit 808. And a rendering unit 809 that synthesizes an image in accordance with the viewpoint and the light source information acquired in (1).
- FIG. 36 and FIG. 37 are flowcharts showing the processing flow of the parameter estimation method and the image composition method in the image composition method according to the present embodiment.
- the image capturing unit 802 calculates color luminance using the color information processing unit 103 described above (step S301).
- the polarization information processing unit 104 may be used to obtain a weighted sum (I max + I min or I max + 2 ⁇ I min ) of the maximum polarization luminance I max and the minimum polarization luminance I min .
- the image separation unit 803 separates the image captured by the image capturing unit 802 into a diffuse reflection component and a specular reflection component by the above-described image separation method (step S302).
- the light source information estimation unit 804 acquires the direction of the light source, and further color information and illuminance information as the light source information (step S303).
- the shape information acquisition unit 805 acquires surface normal information, which is the shape information of the subject, or three-dimensional position information of the subject (step S304).
- the parameter estimation unit 806 obtains the light source information estimated by the light source information estimation unit and the shape information acquired by the shape information acquisition unit for the diffuse reflection image and the specular reflection image separated by the image separation unit 803. By using these, each reflection parameter is estimated by a different method (step S305).
- the parameter DB 807 holds the reflection parameter of the diffuse reflection component and the reflection parameter of the specular reflection component obtained by the parameter estimation unit 806, and the shape information acquired by the shape information acquisition unit S305 as model parameters (step S306).
- the rendering unit 809 calls the model parameter information held in the parameter DB 807 (step S307).
- the viewpoint / light source information acquisition unit 808 acquires the viewpoint of the image to be combined, the direction of the light source, color information, and illuminance information (step S308).
- the rendering unit 809 uses the model parameter information stored in the parameter DB 807 to synthesize an image in accordance with the viewpoint and light source information acquired by the viewpoint / light source information acquisition unit 808 (step S309).
- FIG. 38 shows a configuration example of a camera equipped with the image composition device according to the present embodiment.
- the image separation unit 803, the light source information estimation unit 804, the shape information acquisition unit 805, the parameter estimation unit 806, and the rendering unit 809 are realized by the CPU 205 executing a program.
- the viewpoint / light source information acquisition unit 808 is executed by the user interface unit 212.
- the memory 204 also includes a polarization image captured by the image capturing unit 802, a specular reflection component and diffuse reflection component image acquired by the image separation unit 803, light source information estimated by the light source information estimation unit, and a shape information acquisition unit 805.
- the reflection parameter information estimated by the parameter estimation unit 807, and the viewpoint / light source information acquired by the viewpoint / light source information acquisition unit are stored as model parameters.
- the parameter estimation device 800 will be described.
- the image capturing unit 802 acquires a color image of a subject using the color polarization acquisition unit 102 and the color information processing unit 103 using an imaging device such as a CCD or a CMOS. It is desirable that the image captured in this way be recorded with sufficient luminance resolution without causing the specular reflection component and the diffuse reflection component with extremely high luminance to be saturated at the same time. For this reason, it is desirable to use an imaging device capable of imaging a wide dynamic range, such as a cooled CCD camera or multiple exposure imaging.
- Such an image capturing unit is a weighted sum (I max + I min or I max + 2 ⁇ I min ) of the maximum polarization luminance I max and the minimum polarization luminance I min acquired by the polarization information processing unit 103 described above. Also good.
- I max + I min is an image equivalent to an image captured when no polarizer is installed under a linearly polarized light source. Therefore, by performing image processing using I max + I min , it is possible to perform the same processing as when normal polarization is not used.
- the eyelid image separation unit 803 separates the image captured by the image capturing unit 802 into a diffuse reflection component and a specular reflection component by the above-described image separation method.
- the light source information estimation unit 804 acquires light source direction, color information, and illuminance information as light source information.
- a mirror surface having a known shape for estimating light source information may be disposed near the subject and estimated from the image of the mirror image captured by the image capturing unit 802 (for example, “Nobuyuki Kambara, Naokazu Yokoya, “Vision-based Augmented Reality Considering Optical Consistency Based on Real-Time Estimation of Light Source Environment”, IEICE Technical Report, Pattern Recognition / Media Understanding, PRMU2002-190, pp. 7-12, 2003 ”). This process will be described in detail.
- the light source information estimation unit 804 uses the sphere 601 that can be regarded as a mirror surface shown in FIG.
- a specular sphere 601 is installed near the subject, and its position and normal direction are known.
- the specular sphere 601 is imaged by the image imaging unit 802. At this time, the shooting environment is reflected in the specular sphere 601.
- the position where the reflection occurs is a position where the line-of-sight direction and the direction to the reflection target have a regular reflection relationship with respect to the normal direction on the specular sphere. Therefore, if the position of the specular sphere and its normal direction are known, the direction of the reflection target can be detected from the image reflected on the specular surface.
- the direction of the light source can be acquired by detecting high-luminance pixels from the captured specular image. Furthermore, if the reflectance of the mirror surface is known, light source illuminance information such as light source color information and radiance can be acquired as described above.
- the light source information not only the light source direction but also the light source position information may be obtained instead of the light source direction.
- This uses, for example, the above-mentioned two specular spheres, or a stereo image processing technique widely known in the image processing field using a plurality of images picked up by changing the position of the image pickup device. do it.
- the light source information obtained by shooting before may be used. This is effective when the light source environment does not change like an indoor surveillance camera. In such a case, the specular sphere may be photographed when the camera is installed to obtain the light source information.
- the light source information estimation unit 804 may estimate a light source direction using a subject instead of using a reference object such as a sphere 601 that can be regarded as a mirror surface. This method will be described. First, the pixel with the highest luminance is selected in the image area where the subject is imaged. As will be described later, since the normal information of the surface, which is the shape information of the subject, is estimated by the shape information acquisition unit 805, the normal direction of the pixel with the highest luminance is known. Here, assuming that the light source in the specular reflection direction is reflected in the pixel with the highest luminance, the direction of the light source can be acquired from the normal direction as in the method using the reference object described above.
- the incident illuminance may be measured as the light source information.
- a method of using the incident illuminance information will be described later.
- the shape information acquisition unit 805 acquires surface normal information, which is subject shape information, or three-dimensional position information of the subject.
- an existing method such as a slit light projection method, a pattern light projection method, or a laser radar method may be used.
- the acquisition of shape information is not limited to these methods.
- stereo vision using multiple cameras motion stereo method using camera movement, illuminance difference stereo method using images taken while changing the position of the light source, and using subjects such as millimeter waves and ultrasound
- illuminance difference stereo method using images taken while changing the position of the light source and using subjects such as millimeter waves and ultrasound
- a method using the polarization characteristics of the reflected light for example, USP 5,028,138, “Daisuke Miyazaki, Katsushi Ikeuchi,“ Method for estimating surface shape of transparent object by polarization ray tracing method ”) , Journal of the Institute of Electronics, Information and Communication Engineers, vol. J88-D-II, No. 8, pp. 1432-1439, 2005 ”).
- an illuminance difference stereo method and a method using polarization characteristics will be described.
- the illuminance difference stereo method is a method for estimating the normal direction and reflectance of a subject using three or more images having different light source directions. For example, “H. Hayagawa,“ Photometric Stereo under a light source with reflection, with a point of reflection, with a point of reflection, with a point of reflection. ” By acquiring points with the same rate as known information and using them as constraint conditions, the following parameters are estimated while the position information of the light source is unknown.
- Subject information Normal direction and reflectance of each point on the image
- Light source information Light source direction and illuminance at the observation point of the subject
- only diffuse reflection images separated by the diffuse reflection / specular reflection separation method described above are used. Use the photometric stereo method. Originally, since this method assumes that the subject is completely diffusely reflected, a large error occurs in a subject where specular reflection exists. However, the estimation error due to the presence of the specular reflection component can be eliminated by using only the separated diffuse reflection component.
- Diffuse reflection images with different light source directions are expressed by a luminance matrix I d as follows.
- i df (p) indicates the luminance at the pixel p of the diffuse reflection image in the light source direction f.
- the number of pixels of the image is P pixels, and the number of images taken in different light source directions is F.
- ⁇ dp represents the reflectance (albedo) of the pixel p
- n p represents the normal direction vector of the pixel p
- t f represents the incident illuminance of the light source f
- L f represents the direction vector of the light source f.
- R is a surface reflection matrix
- N is a surface normal matrix
- L is a light source direction matrix
- T is a light source intensity matrix
- S is a surface matrix
- M is a light source matrix
- Equation 24 can be expanded as follows.
- E represents a unit matrix.
- U ′ is a P ⁇ 3 matrix
- U ′′ is a P ⁇ (F-3) matrix
- ⁇ ′ is a 3 ⁇ 3 matrix
- ⁇ ′′ is an (F-3) ⁇ (F-3) matrix
- V ′ is 3
- the ⁇ F matrix, V ′′ is an (F-3) ⁇ F matrix.
- U ′′, V ′′ are considered to be orthogonal bases of signal components U ′, V ′, that is, noise components.
- Equation 26 can be transformed as follows.
- Equation 27 shape information and light source information can be obtained simultaneously, but the following 3 ⁇ 3 matrix A indeterminacy remains.
- A is an arbitrary 3 ⁇ 3 matrix.
- the reflectance is known to be equal at six or more points on the screen. For example, if the reflectances of arbitrary 6 points k1 to k6 are equal, From Equation 25, Equation 28, and Equation 30, further, Then, Formula 31 becomes as follows.
- Equation 33 can be solved.
- the matrix A can be solved by using singular value decomposition in Expression 32.
- shape information and light source information are obtained from Equation 28 and Equation 29.
- Subject information Normal direction vector and reflectance of each point on the imageLight source information: Light source direction vector and radiance at the observation point of the subject
- the reflectance of the subject obtained by the above processing and the radiance of the light source are relative, and in order to obtain the absolute value, the reflectance is known at six or more points on the screen. Different known information is required.
- the normal information of the surface is acquired by the method using the illuminance difference stereo method and the polarization characteristic.
- a method such as slit light projection or stereo vision
- three-dimensional position information of the subject is acquired.
- the normal information on the surface of the subject is tilt information in a minute space of the three-dimensional position information of the subject, and both are shape information of the subject.
- the shape information acquisition unit 805 acquires the surface normal information or the subject three-dimensional position information, which is the subject shape information.
- the parameter estimation unit 806 estimates each reflection parameter for the diffuse reflection component and the specular reflection component separated by the image separation unit 803 by different methods. First, the process of the diffuse reflection component will be described.
- the parameter estimation unit 806 estimates the albedo of the subject using the diffuse reflection component separated by the image separation unit 803. Since albedo is not affected by light source information, processing that is robust to light source fluctuations can be realized by performing processing using an albedo image. This process will be described. First, the reflection characteristics of an object will be described. Assuming a dichroic reflection model, the brightness of an object is expressed by the following equation as the sum of a diffuse reflection component and a specular reflection component.
- I is the luminance of the subject imaged by the imaging apparatus
- I a is the ambient light component
- I d is the diffuse reflection component
- I s is the specular reflection component.
- the ambient light component is indirect light in which light from a light source is scattered by an object or the like. It is scattered throughout the space and gives a slight brightness to shadows where direct light does not reach. For this reason, it is usually handled as noise.
- the image can be separated into a diffuse reflection component and a specular reflection component.
- ⁇ i represents an angle formed between the normal direction vector of the subject and the light source direction vector.
- the angle ⁇ i is known by the light source information estimation unit 804 and the shape information acquisition unit 805.
- the albedo ⁇ dp of the subject can be obtained from Equation 35.
- a pseudo albedo obtained by multiplying the albedo by the radiance of the light source may be obtained and used according to the following equation.
- the parameter estimation unit 806 estimates parameters representing the subject using the normal information of the subject acquired by the shape information acquisition unit 805 and the diffuse reflection image and the specular reflection image separated by the image separation unit 803. .
- a method of using a Cook-Torrance model widely used in the field of Computer-Graphics will be described.
- the specular reflection image is modeled as follows.
- E i is the incident illuminance
- ⁇ s ⁇ is the bidirectional reflectance of the specular reflection component at wavelength ⁇
- n is the normal direction vector of the subject
- V is the line-of-sight vector
- L is the light source direction vector
- H is the line-of-sight vector
- ⁇ represents the angle between the intermediate vector H and the normal direction vector n (see FIG. 39).
- F ⁇ is a Fresnel coefficient which is a ratio of reflected light from the dielectric surface obtained from the Fresnel equation
- D is a microfacet distribution function
- G is a geometric attenuation factor representing the influence of light shielding by unevenness on the object surface.
- n ⁇ is the refractive index of the subject
- m is a coefficient indicating the roughness of the subject surface
- I j is the radiance of the incident light
- K s is a coefficient of the specular reflection component.
- ⁇ d is the reflectance (albedo) of the diffuse reflection component
- dpx, dpy are the lengths in the x direction and y direction of one pixel of the imaging device
- r is the distance from the observation point O of the imaging device.
- K d is a coefficient that satisfies the following relational expression.
- FIG. 40 is a schematic diagram for explaining the constant Sr.
- the diffuse reflection component energy reflected at the observation point O spreads in a hemispherical shape.
- the ratio S r between the energy reaching one imaging element of the imaging device and the total energy reflected at the observation point O is expressed by Expression 47.
- the parameter estimation unit 806 estimates the reflection parameter of the specular reflection component from Expression 36 to Expression 47.
- the known parameters for parameter estimation and the parameters to be estimated are as follows.
- (Known parameters) ⁇ Ambient light component Ia ⁇ Diffuse reflection component I d ⁇ Specular reflection component I s ⁇ Normal direction vector n of the subject ⁇
- Light source direction vector L ⁇ Gaze vector V ⁇ Intermediate vector H ⁇
- An angle ⁇ between the intermediate vector H and the normal direction vector n Length of one pixel of the imaging device 701 in the x and y directions dpx, dpy ⁇ Distance r between imaging device 701 and observation point O (Parameters to be estimated) ⁇ Incident illuminance E i ⁇ Specular reflection coefficient k s ⁇ Subject surface roughness m ⁇ Subject refractive index ⁇
- the coefficient k d of the diffuse reflection component and the reflectance (albedo) ⁇ d of the diffuse reflection component are also unknown parameters. However, since only the parameter of the specular reflection component is estimated, no estimation processing is performed here.
- FIG. 41 is a diagram showing a processing flow of the parameter estimation unit 806. The process consists of the following two stages.
- the incident illuminance E i is obtained using the light source information (step S401).
- the position information of the light source acquired by the light source information estimation unit 804, the distance information between the imaging device and the subject obtained by the shape information acquisition unit 805, and the light source illuminance obtained by the light source information estimation unit 804 are used. This is obtained from the following equation.
- the illuminance meter 211 is installed in the imaging device 701 as described above.
- I i is the incident illuminance of the light source 702 measured by the illuminometer 211
- R 1 is the distance between the imaging device 701 and the light source 702
- R 2 is the distance between the light source 702 and the observation point O
- ⁇ 1 is the observation point.
- the angle formed between the normal direction vector n and the light source direction vector LC in O, and ⁇ 2 represents the angle formed between the optical axis direction and the light source direction vector LA in the imaging device 701 (see FIG. 42).
- the simplex method is a method in which variables are assigned to the vertices of a figure called simplex, and the function is optimized by changing the size and shape of the simplex (Noboru Ohta, “Basics of Color Reproduction Optics”, pp. 90-92, Corona).
- a simplex is a set of (n + 1) points in an n-dimensional space.
- n is the number of unknowns to be estimated, and is “3” here. Therefore, the simplex is a tetrahedron.
- the vertex of the simplex is represented by a vector x i and a new vector is defined as follows:
- the simplex method is based on the expectation that the function value in the mirror image becomes small by selecting the largest function value among the vertices of the simplex. If this expectation is correct, the minimum value of the function is obtained by repeating the same process. That is, while updating the parameter given as the initial value by three types of operations, the parameter update is repeated until the error from the target indicated by the evaluation function becomes less than the threshold.
- m, ⁇ , k s as parameters
- the difference ⁇ I s between the specular reflection component image calculated from Expression 36 and the specular reflection component image obtained by the image separation unit 803 represented by Expression 55 as an evaluation function.
- i s (i, j) ′ and i s (i, j) are the calculated specular reflection image estimated value I s ′ and the pixel of the specular reflection component image I s obtained by the image separation unit 803, respectively.
- the luminance of (i, j), M s (i, j), is a function that takes 1 when the pixel (i, j) has a specular reflection component and 0 otherwise.
- FIG. 43 is a flowchart for explaining the flow of this process.
- 0 is substituted into counters n and k for storing the number of repetitive calculation updates, and initialization is performed (step S411).
- the counter n is a counter that stores how many times the initial value is changed
- k is a counter that stores how many times the candidate parameter is updated by the simplex with respect to a certain initial value.
- step S412 random values are used to determine initial values of estimation parameter candidate parameters m ′, ⁇ ′, and k s ′ (step S412).
- the generation range of the initial value was determined as follows.
- the candidate parameter obtained in this way is substituted into Equation 36 to obtain an estimated value I s ′ of the specular reflection image (step S413). Further, a difference ⁇ I s between the calculated specular reflection image estimated value I s ′ and the specular reflection component image obtained by the image separation unit 803 is obtained from Expression 55, and this is used as an evaluation function of the simplex method (step S414). ). If ⁇ I s obtained in this way is sufficiently small (Yes in step S415), the parameter estimation is successful, and candidate parameters m ′, ⁇ ′, k s ′ are selected as the estimation parameters m, ⁇ , k s , and the process ends. To do. On the other hand, if ⁇ I s is large (No in step S415), the candidate parameters are updated by the simplex method.
- step S416 1 is added to the counter k storing the number of updates (step S416), and the size of the counter k is determined (step S417). If the counter k is sufficiently large (No in step S417), the repetitive calculation has been sufficiently performed, but since it has fallen to the local minimum, it is determined that the optimum value is not reached even if the update is repeated as it is. Change the value to get out of the local minimum. Therefore, 1 is added to the counter n, and 0 is added to the counter k (step S421).
- step S422 it is determined whether or not the value of the counter n is higher than the threshold value, and it is determined whether the processing is continued as it is or whether the processing is terminated because the processing is impossible (step S422).
- n is larger than the threshold value (No in step S422), it is determined that this image cannot be estimated, and the process ends.
- the initial value is again selected from random numbers within the range of expression 56 (step S412), and the process is repeated. For example, 100 may be selected as the threshold for such k.
- step S417 if the counter k is equal to or smaller than the threshold value in step S417 (Yes in step S417), the candidate parameters are changed using the equations 52 to 54 (step S418). This process will be described later.
- step S419) it is determined whether the candidate parameter thus transformed is meaningful as a solution. That is, by repeating the simplex method, there is a possibility that the deformed parameter has a physically meaningless value (for example, the roughness parameter m is a negative value, etc.), which is removed. For example, the following conditions may be given, and if this condition is satisfied, it may be determined as a meaningful parameter, and if not satisfied, it may be determined as a meaningless parameter.
- a physically meaningless value for example, the roughness parameter m is a negative value, etc.
- the refractive index ⁇ is a value determined by the material of the subject.
- plastic is 1.5 to 1.7 and glass is 1.5 to 1.9, and these values may be used. That is, when the subject is plastic, the refractive index ⁇ may be 1.5 to 1.7.
- step S419 If the deformed parameter satisfies Expression 57 (Yes in step S419), the candidate parameter is considered to be a meaningful value, so it is set as a new candidate parameter (step S420) and the update process is repeated (step S413). On the other hand, if the deformed parameter does not satisfy Expression 57 (No in step S419), the updating process for the initial value is terminated, and the update is performed with the new initial value (step S421).
- FIG. 44 is a flowchart showing the flow of this process.
- the candidate parameters m ′, ⁇ ′, and k s ′ are expressed as vectors, which are set as parameters x. That is,
- step S431 First, using the equation 50 to equation 52, then calculate the parameters x r performing the mirror operation, calculating the difference [Delta] I s (x r) of the specular reflection component image with x r by the expression 55 (step S431 ).
- ⁇ I s (x r ) obtained in this way is compared with ⁇ I s (x s ) having the second worst evaluation function (step S432).
- the evaluation value ⁇ I s (x r ) obtained by performing the mirror image operation and ⁇ I s ( x l) comparing step S433.
- ⁇ I s (x r ) is equal to or larger than ⁇ I s (x l ) (No in step S433), x h having the lowest evaluation value is changed to x r (step S434), and the process is terminated. To do.
- step S433 if ⁇ I s (x r) is smaller than ⁇ I s (x l) (Yes in step S433), it performs expansion processing by using the equation 54, the parameters x e, the specular reflection component image with x e A difference ⁇ I s (x e ) is calculated (step S435). Next, ⁇ I s (x e ) obtained in this way is compared with ⁇ I s (x r ) by mirror image operation (step S436). If ⁇ I s (x e ) is smaller than ⁇ I s (x r ) (Yes in step S436), x h having the worst evaluation value is changed to x e (step S437), and the process ends.
- step S436 if ⁇ I s (x e ) is greater than or equal to ⁇ I s (x r ) (No in step S436), x h having the worst evaluation value is changed to x r (step S434), and the process is terminated. .
- step S432 If ⁇ I s (x r ) is larger than ⁇ I s (x s ) in step S432 (No in step S432), the evaluation value ⁇ I s (x r ) on which the mirror image operation has been performed and the current evaluation value The bad ⁇ I s (x h ) is compared (step S438).
- ⁇ I s (x r ) is smaller than ⁇ I s (x h ) (Yes in step S438), x h having the worst evaluation value is changed to x r (step S439), and Expression 53 is used.
- Te calculates a parameter x c performing the contraction operation, the difference between the specular reflection component image with x c [Delta] I s a (x c) (step S440).
- ⁇ I s (x r ) is greater than or equal to ⁇ I s (x h ) (No in step S438)
- a difference ⁇ I s (x c ) with the reflection component image is calculated (step S440).
- ⁇ I s (x c ) thus obtained is compared with ⁇ I s (x h ) having the worst evaluation value (step S441). If ⁇ I s (x c ) is smaller than ⁇ I s (x h ) (Yes in step S441), x h having the worst evaluation value is changed to x c (step S442), and the process is terminated.
- the unknown parameters m, ⁇ , and k s in the specular reflection image are estimated.
- FIG. 45 (a) is a graph in which the horizontal axis represents the intermediate vector ⁇ and the vertical axis represents the luminance Is of the specular reflection component for the region A in FIG. 45 (b).
- white ⁇ is a plot of the luminance Is observed for the region A.
- the black squares are obtained by estimating and plotting each parameter of the Cook-Torrance model.
- FIG. 46 is a graph using a conventional image separation method using polarization information.
- FIG. 47 shows a composite image created using a conventional image separation method using polarization information. As described above, when parameter estimation fails, the texture of the composite image is greatly different from the real thing.
- the model used for parameter estimation does not need to be a Cook-Torrance model.
- a Torrance-Sparrow model for example, a Torrance-Sparrow model, a Phong model, a simple Torrance-Sparrow model (for example, “K. Ikechi and K. a Sato, min“ Determining reflectance ” of an object using range and brightness images ”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.13, no.11, 3.
- the parameter estimation method does not need to be a simplex method, and for example, a general parameter estimation method such as a gradient method or a least square method may be used.
- processing may be performed for each pixel, or image separation may be performed to estimate an equal parameter set for each region.
- a sample in which known parameters such as a normal direction vector n, a light source direction vector L, or a line-of-sight vector V of the subject fluctuate is acquired by moving the light source, the imaging device, or the subject. It is desirable.
- processing is performed for each region, it is desirable to perform optimal parameter estimation by changing the image separation so that the variation in parameters obtained for each region is reduced.
- the parameter DB 807 holds the reflection parameter of the diffuse reflection component and the reflection parameter of the specular reflection component obtained by the parameter estimation unit 806 and the shape information acquired by the shape information acquisition unit 805 as model parameters.
- FIG. 48 is a schematic diagram showing model parameters held in the parameter DB 307.
- the parameter estimation device 800 estimates the diffuse reflection component parameter and the specular reflection component parameter, and holds the estimated parameter information in the parameter DB 807.
- the viewpoint / light source information acquisition unit 808 acquires the viewpoint and light source information of the image to be combined.
- the user may input the viewpoint position, the light source position / light source illuminance, and the ambient light component information.
- the light source information may be estimated using the light source information estimation unit 804.
- the rendering unit 809 uses the model parameter information held in the parameter DB 807 to synthesize an image in accordance with the viewpoint and light source information acquired by the viewpoint / light source information acquisition unit 808.
- the rendering unit 809 performs rendering separately for the diffuse reflection component and the specular reflection component, respectively, and combines the rendered diffuse reflection component, specular reflection component, and ambient light component information to synthesize an image.
- the diffuse reflection component will be described.
- the albedo image is obtained by dividing the diffusion component image by the inner product of the light source vector and the normal direction vector of the subject. Therefore, the light source direction vector information acquired by the viewpoint / light source information acquisition unit 808 is used for the albedo image (estimated by the parameter estimation unit 806) and the shape information (acquired by the shape information acquisition unit 805) held in the parameter DB.
- the diffuse reflection component can be synthesized.
- the diffusion component image is synthesized by obtaining the inner product of the light source direction vector acquired by the viewpoint / light source information acquisition unit 808 and the normal direction vector of the subject and further multiplying the albedo image.
- a diffuse reflection component image is synthesized for each light source, and the images are added together to synthesize a single diffuse reflection component image. To do.
- the specular reflection component includes the light source direction acquired by the viewpoint / light source information acquisition unit 808 in the specular reflection parameter (estimated by the parameter estimation unit 806) and the shape information (acquired by the shape information acquisition unit 805) held in the parameter DB. By using vector information, it can be synthesized. Specifically, the specular reflection component image is synthesized by substituting the estimated parameters into Expression 36 to Expression 44.
- the rendering unit 809 By rendering the diffuse reflection component image, the specular reflection component image, and the ambient light component information acquired by the viewpoint / light source information acquisition unit 808, the rendering unit 809 obtains the viewpoint acquired by the viewpoint / light source information acquisition unit 808. And synthesize images according to light source information.
- model-based image composition used in digital archives and the like can be performed with high accuracy.
- image processing it is possible to more accurately separate the specular reflection component and the diffuse reflection component by using two pieces of information of polarization information and color information. Since image separation processing can be realized from an image shot so as to take a snapshot, it is useful for various digital still cameras, digital movie cameras, surveillance cameras, and the like.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Studio Devices (AREA)
- Image Processing (AREA)
- Color Television Image Signal Generators (AREA)
Abstract
Description
まず、本発明の第1の実施形態における画像処理装置の概要について説明する。
・偏光度ρ
・鏡面反射成分は表面反射によって生じるため、入射光の偏光特性が保持されている。そのため、カメラで観測された輝度の偏光成分として観測される。
・拡散反射はスキャッタリングを繰り返しているため、入射光の偏光特性が失われている。そのため、カメラで観測された輝度の非偏光成分として観測される。
これらの偏光特性は、以下の2条件を前提としている。
(条件1)直線偏光光が投光されている場合、鏡面反射成分は偏光成分として観測される。
(条件2)直線偏光光が投光されている場合、拡散反射成分は非偏光成分として観測される。
・出射角が十分に大きい領域以外では、拡散反射成分の偏光度は十分小さい。
・出射角が十分に大きい領域では、拡散反射成分の偏光度は、鏡面反射成分の偏光度に比べて十分に大きい。
・出射角が十分に大きい領域(遮蔽エッジ近傍)では、(条件2)が成立しない。
図29は、本実施形態に係る画像分離システムにおけるブロック図を示している。図29において、図1と共通の構成要素には図1と同一の符号を付しており、ここではその詳細な説明は省略する。第1の実施形態との違いは、投光装置107と画像処理装置108を分離させたことである。また、図30は本実施形態に係る画像処理装置における、画像分離方法の処理の流れを示すフローチャートである。図30において、図2と共通のステップには図2と同一の符号を付しており、ここではその詳細な説明は省略する。さらに、図31は、本実施形態に係る画像処理装置108が搭載されたカメラと投光装置107の構成例を示している。図31において、図3と共通の構成要素には図3と同一の符号を付しており、ここではその詳細な説明は省略する。
本発明の画像成分分離は、デジタルアーカイブなどに使われているモデルベースの画像合成処理に特に有効である。モデルベースの画像合成は、撮像した画像の光源方向や視線方向を任意に変化させることができるため、撮像データのインタラクティブな提示方法として重要である。また、臨場感を与えるためにも非常に有効である。そのため、モデルベースの画像合成処理は、バーチャルミュージアムやAR(Augmented Reality)などの応用に利用できる。本来、このようなモデルベースの画像合成では撮像画像を鏡面反射成分と拡散反射成分に領域分離を行ない、それぞれの成分ごとに異なったモデルを利用してきたが、前述のように、鏡面反射成分と拡散反射成分の分離が不十分であったため、画質が劣化するという問題があった。この問題は、以下のことが原因である。それぞれのモデルにおけるパラメータ推定の際、実際とは異なった鏡面反射成分と拡散反射成分を利用しているため、実際とは異なるパラメータを推定してしまう。このように実際とは異なったパラメータを利用して画像を合成した場合、前述のように、視点変換画像や光源変化画像を作成した際に出力される画像に大きな誤差が生じてしまう。
○被写体の拡散反射画像
○被写体の鏡面反射画像
○被写体の3次元形状情報
○光源位置・色・照度
○合成画像における視点・光源情報
・被写体情報:画像上の各点の法線方向と反射率
・光源情報:被写体の観察点における光源方向と照度
ここでは、前述の拡散反射・鏡面反射分離手法によって分離された拡散反射画像のみを利用した照度差ステレオ法を行う。本来、この手法は被写体が完全拡散反射をしていることを仮定しているため、鏡面反射が存在する被写体では大きな誤差が生じてしまう。しかし、分離した拡散反射成分のみを利用することで、鏡面反射成分の存在による推定誤差を無くすことができる。
・被写体情報:画像上の各点の法線方向ベクトルと反射率
・光源情報:被写体の観察点における光源方向ベクトルと放射輝度
○被写体の拡散反射画像
○被写体の鏡面反射画像
○被写体の3次元形状情報
○光源位置・照度
この処理を説明する。まず、物体の反射特性について説明する。2色性反射モデルを仮定すると、物体の輝度は、拡散反射成分と鏡面反射成分との和として以下の式で表現される。
(既知パラメータ)
○環境光成分Ia
○拡散反射成分Id
○鏡面反射成分Is
○被写体の法線方向ベクトルn
○光源方向ベクトルL
○視線ベクトルV
○中間ベクトルH
○中間ベクトルHと法線方向ベクトルnの角度β
○撮像装置701の1画素のx方向、y方向の長さdpx, dpy
○撮像装置701と観察点Oとの距離r
(推定すべきパラメータ)
○入射照度Ei
○鏡面反射成分の係数ks
○被写体表面の粗さm
○被写体の屈折率ηλ
さらに、この方法で用いる3種類の操作を以下のように定める。
1.鏡像:
102 カラー偏光取得部
103 カラー情報処理部
104 偏光情報処理部
105 光源色情報取得部
106 画像成分分離部
107 投光装置
108 画像処理装置
109 通信部
110 通信部
Claims (19)
- 被写体を撮像することによって前記被写体の画像の成分分離を行う画像処理装置であって、
光源から発せられた直線偏光光を前記被写体に投光する投光部と、
前記被写体のカラー偏光画像を取得するカラー偏光取得部と、
前記カラー偏光画像を構成する単位画素のそれぞれについて、偏光主軸の方向が3以上の偏光子を透過した光の輝度と前記偏光主軸の方向との対応関係に基づいて、カラー偏光情報を生成する偏光情報処理部と、
前記光源の色情報を取得する光源色情報取得部と、
前記カラー偏光情報と前記光源の色情報とに基づいて、前記画像の成分分離を行う画像成分分離部と、
を備える画像処理装置。 - 前記カラー偏光画像からカラー画像を生成するカラー情報処理部を備え、
前記画像成分分離部は、前記カラー画像の成分分離を行う、請求項1に記載の画像処理装置。 - 前記カラー偏光取得部は、
偏光主軸の方向が異なる3方向以上の偏光子と、
前記偏光子に対向する位置に配置されたカラーフィルタと、
前記偏光子およびカラーフィルタを透過してきた光を受ける撮像素子とを備える、請求項1に記載の画像処理装置。 - 前記偏光情報処理部は、前記カラー偏光情報として、偏光最小カラー成分、偏光振幅カラー成分の少なくとも1つを生成する請求項3に記載の画像処理装置。
- 前記画像成分分離部は、前記被写体の画像を構成する各画素の色ベクトルを、拡散反射成分および鏡面反射成分に分離する請求項4に記載の画像処理装置。
- 前記画像成分分離部は、前記被写体の画像を構成する各画素の色ベクトルを、拡散反射非偏光成分、拡散反射偏光成分および鏡面反射偏光成分に分離する請求項4に記載の画像処理装置。
- 前記画像成分分離部は、前記被写体の画像の少なくとも一部を陰影領域に分離する請求項5または6に記載の画像処理装置。
- 前記画像成分分離部は、前記カラー偏光情報を、前記光源色ベクトルと、前記偏光最小カラー成分の色ベクトルとによって分離する請求項5または6に記載の画像処理装置。
- 前記画像成分分離部は、前記偏光振幅カラー成分を、前記光源色ベクトルと、前記偏光最小カラー成分の色ベクトルとによって分離する請求項8に記載の画像処理装置。
- 前記画像成分分離部は、前記カラー画像を、前記光源色ベクトルと、前記偏光最小カラー成分の色ベクトルとによって分離する請求項5または6に記載の画像処理装置。
- 前記カラー偏光取得部は、前記投光部と同期処理を行う請求項1に記載の画像処理装置。
- 前記投光部は、カラー偏光取得部から離して配置することを特徴とする請求項3に記載の画像処理装置。
- 投光装置と画像処理装置とを備え、被写体を撮像することによって前記被写体の画像の成分分離を行う画像処理システムであって、
前記投光装置は、
光源から発せられた直線偏光光を前記被写体に投光する投光部を有し、
前記画像処理装置は、
偏光主軸の方向が異なる3方向以上の偏光子を透過してくる光をカラーフィルタを通して受光することで、被写体のカラー偏光画像を取得するカラー偏光取得部と、
前記カラー偏光画像からカラー画像を生成するカラー情報処理部と、
前記カラー偏光画像を構成する単位画素のそれぞれについて、前記偏光子を透過した光の輝度と前記偏光主軸の方向との対応関係に基づいて、カラー偏光情報を生成する偏光情報処理部と、
前記光源の色情報を取得する光源色情報取得部と、
前記カラー偏光情報と前記光源色情報に基づいて、前記カラー画像の成分分離を行う画像成分分離部と、を有する画像処理システム。 - 投光装置と画像処理装置とを備え、被写体を撮像することによって前記被写体の画像の成分分離を行う画像分離システムであって、
前記投光装置は、
光源から発せられた直線偏光光を前記被写体に投光する投光部と、
投光を知らせる信号を前記画像処理装置へ送信する通信部を有し、
前記画像処理装置は、
前記投光を知らせる信号を前記投光装置から受信する通信部と、
偏光主軸の方向が異なる3方向以上の偏光子を透過してくる光をカラーフィルタを通して受光することで、被写体のカラー偏光画像を取得するカラー偏光取得部と、
前記カラー偏光画像からカラー画像を生成するカラー情報処理部と、
前記カラー偏光画像を構成する単位画素のそれぞれについて、前記偏光子を透過した光の輝度と前記偏光主軸の方向との対応関係に基づいて、カラー偏光情報を生成する偏光情報処理部と、
光源の色情報を取得する光源色情報取得部と、
前記光源の色情報を取得する光源色情報取得部と、
前記カラー偏光情報と前記光源の色情報に基づいて、前記カラー画像の成分分離を行う画像成分分離部と、を有する画像処理システム。 - 前記通信部は、投光を知らせる信号に加え、前記光源の色情報を送受信し、前記光源色情報取得部は、前記通信部から前記光源の色情報を取得する請求項14に記載の画像処理システム。
- 被写体を撮像することによって前記被写体の画像の成分分離を行う画像処理方法であって、
光源から発せられた直線偏光光を前記被写体に投光する投光ステップと、
偏光主軸の方向が異なる3方向以上の偏光子を透過してくる光をカラーフィルタを通して受光することで、被写体のカラー偏光画像を取得するカラー偏光取得ステップと、
前記カラー偏光画像からカラー画像を生成するカラー情報処理ステップと、
前記カラー偏光画像を構成する単位画素のそれぞれについて、前記偏光子を透過した光の輝度と前記偏光主軸の方向との対応関係を用いて、受光した偏光に関するカラー情報であるカラー偏光情報を生成する偏光情報処理ステップと、
光源の色情報を取得する光源色情報取得ステップと、
前記カラー偏光情報と前記光源の色情報を利用して、前記カラー画像の成分分離を行う画像成分分離ステップとを含む画像処理方法。 - 被写体を撮像することによって前記被写体の画像を成分分離する画像処理装置のためのプログラムであって、
請求項16に記載の画像処理方法に含まれるステップをコンピュータに実行させるプログラム。 - パラメータ推定装置と画像合成装置とを備え、画像を合成するモデルベース画像合成装置であって、
前記パラメータ推定装置は、
被写体を撮像する画像撮像部と、
請求項16に記載の画像処理方法によって、前記画像撮像部によって撮像された画像の成分分離を行う画像分離部と、
被写体に照射する光源の方向や位置、輝度、色、スペクトル情報の少なくとも1つを含む光源情報を推定する光源情報推定部と、
被写体の表面の法線情報または3次元位置情報を形状情報として取得する形状情報取得部と、
撮像された被写体から、前記画像分離部で分割された成分ごとに前記光源情報推定部で推定された光源情報と前記形状情報取得部で取得された形状情報をモデル化することで反射モデルパラメータを推定するパラメータ推定部と、
前記パラメータ推定部において推定された反射パラメータを保持するパラメータDBと、を有し、
前記画像合成装置は、
合成する画像の視点や光源情報を取得する視点・光源情報取得部と、
前記パラメータDBに保持されている反射パラメータを利用して、前記視点・光源情報取得部で取得された視点や光源情報に則した画像を合成するレンダリング部と
を有するモデルベース画像合成装置。 - パラメータ推定ステップと画像合成ステップとを備え、画像を合成するモデルベース画像合成方法であって、
前記パラメータ推定ステップは、
被写体を撮像する画像撮像ステップと、
請求項16に記載の画像処理方法によって、前記画像撮像部によって撮像された画像の成分分離を行う画像分離ステップと、
光源の情報を推定する光源情報推定ステップと、
被写体の表面の法線情報または3次元位置情報を形状情報として取得する形状情報取得ステップと、
撮像された被写体から、前記画像分離部で分割された成分ごとに反射モデルパラメータを推定するパラメータ推定ステップと
を含み、
前記画像合成ステップは、
合成する画像の視点や光源情報を取得する視点・光源情報取得ステップと、
前記形状情報取得ステップによって推定された反射パラメータを利用して、前記視点・光源情報取得ステップで取得された視点や光源情報に則した画像を合成するレンダリングステップと
を含むモデルベース画像合成方法。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US12/672,940 US8025408B2 (en) | 2008-07-08 | 2009-05-15 | Method, apparatus and program for image processing and method and apparatus for image synthesizing |
CN2009801064169A CN101960859B (zh) | 2008-07-08 | 2009-05-15 | 图像处理方法、图像处理装置、图像合成方法和图像合成装置 |
JP2009549317A JP4469021B2 (ja) | 2008-07-08 | 2009-05-15 | 画像処理方法、画像処理装置、画像処理プログラム、画像合成方法、および画像合成装置 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2008-178025 | 2008-07-08 | ||
JP2008178025 | 2008-07-08 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2010004677A1 true WO2010004677A1 (ja) | 2010-01-14 |
Family
ID=41506805
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2009/002165 WO2010004677A1 (ja) | 2008-07-08 | 2009-05-15 | 画像処理方法、画像処理装置、画像処理プログラム、画像合成方法、および画像合成装置 |
Country Status (4)
Country | Link |
---|---|
US (1) | US8025408B2 (ja) |
JP (1) | JP4469021B2 (ja) |
CN (1) | CN101960859B (ja) |
WO (1) | WO2010004677A1 (ja) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102742258A (zh) * | 2010-07-21 | 2012-10-17 | 松下电器产业株式会社 | 图像处理装置 |
JP2013009082A (ja) * | 2011-06-23 | 2013-01-10 | Nippon Telegr & Teleph Corp <Ntt> | 映像合成装置、映像合成方法、および映像合成プログラム |
JP2013045232A (ja) * | 2011-08-23 | 2013-03-04 | Ricoh Co Ltd | 車両運転支援装置、道路の路肩を検出する方法および該方法に基づく車両運転支援方法 |
CN103037752A (zh) * | 2010-09-24 | 2013-04-10 | 松下电器产业株式会社 | 图像处理装置 |
JP2013520675A (ja) * | 2010-02-25 | 2013-06-06 | ヴォロテック・リミテッド | 様々な偏光角度を有する光フィルタ及び処理アルゴリズム |
JP2013245944A (ja) * | 2012-05-23 | 2013-12-09 | Nippon Telegr & Teleph Corp <Ntt> | 投影型3次元形状復元装置、投影型3次元形状復元方法、投影型3次元形状復元プログラムおよびそのプログラムを記録した記録媒体 |
JP2014042210A (ja) * | 2012-08-23 | 2014-03-06 | Nippon Telegr & Teleph Corp <Ntt> | 投影型3次元形状復元装置、投影型3次元形状復元方法及び投影型3次元形状復元プログラム |
JP2015059849A (ja) * | 2013-09-19 | 2015-03-30 | 凸版印刷株式会社 | 色と三次元形状の計測方法及び装置 |
AU2011242408B2 (en) * | 2010-04-21 | 2016-12-15 | Commonwealth Scientific And Industrial Research Organisation | Shape and photometric invariants recovery from polarisation images |
JP2019020991A (ja) * | 2017-07-14 | 2019-02-07 | シャープ株式会社 | 撮像装置、認証装置、および撮像方法 |
US10397544B2 (en) | 2010-08-19 | 2019-08-27 | Nissan Motor Co., Ltd. | Three-dimensional object detection device and three-dimensional object detection method |
CN111247794A (zh) * | 2017-10-30 | 2020-06-05 | 索尼半导体解决方案公司 | 固态摄像装置和电子设备 |
JP2020139821A (ja) * | 2019-02-27 | 2020-09-03 | 株式会社新菱 | 検査装置、検査システム及び検査方法 |
Families Citing this family (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8718336B2 (en) * | 2007-03-30 | 2014-05-06 | Fujifilm Corporation | Image correction apparatus, image correction method, skin diagnosis method, and computer program product |
JP5018486B2 (ja) * | 2008-01-10 | 2012-09-05 | セイコーエプソン株式会社 | 画像処理回路、画像処理装置 |
US8184194B2 (en) * | 2008-06-26 | 2012-05-22 | Panasonic Corporation | Image processing apparatus, image division program and image synthesising method |
US8025408B2 (en) * | 2008-07-08 | 2011-09-27 | Panasonic Corporation | Method, apparatus and program for image processing and method and apparatus for image synthesizing |
JP5648483B2 (ja) * | 2009-01-27 | 2015-01-07 | 日本電気株式会社 | カラー画像処理方法、カラー画像処理装置およびカラー画像処理プログラム |
IL201110A (en) | 2009-09-22 | 2014-08-31 | Vorotec Ltd | Device and method of navigation |
JP5795384B2 (ja) * | 2010-12-09 | 2015-10-14 | サムスン エレクトロニクス カンパニー リミテッド | 映像処理装置、照明処理装置及びその方法 |
US9576349B2 (en) * | 2010-12-20 | 2017-02-21 | Microsoft Technology Licensing, Llc | Techniques for atmospheric and solar correction of aerial images |
US9110282B2 (en) * | 2011-03-30 | 2015-08-18 | The Regents Of The University Of California | Nanometer-scale optical imaging by the modulation tracking (MT) method |
JP5906464B2 (ja) | 2012-02-02 | 2016-04-20 | パナソニックIpマネジメント株式会社 | 撮像装置 |
JP6251962B2 (ja) | 2012-03-01 | 2017-12-27 | 日産自動車株式会社 | カメラ装置及び画像処理方法 |
JP6251963B2 (ja) * | 2012-03-01 | 2017-12-27 | 日産自動車株式会社 | カメラ装置及び画像処理方法 |
JP6131545B2 (ja) * | 2012-03-16 | 2017-05-24 | 株式会社ニコン | 画像処理装置、撮像装置および画像処理プログラム |
US9536345B2 (en) * | 2012-12-26 | 2017-01-03 | Intel Corporation | Apparatus for enhancement of 3-D images using depth mapping and light source synthesis |
KR101473768B1 (ko) * | 2013-11-28 | 2014-12-24 | 한국과학기술연구원 | 편광 라이트 필드 카메라를 이용한 스페큘러 및 디퓨즈 영상 생성 장치 및 그것의 제어방법 |
JP2015144194A (ja) * | 2014-01-31 | 2015-08-06 | ソニー株式会社 | 固体撮像素子および電子機器 |
KR102211862B1 (ko) * | 2014-04-09 | 2021-02-03 | 삼성전자주식회사 | 이미지 센서 및 이를 포함하는 이미지 센서 시스템 |
GB201414144D0 (en) * | 2014-08-08 | 2014-09-24 | Imagination Tech Ltd | Relightable texture for use in rendering an image |
WO2016136086A1 (ja) * | 2015-02-27 | 2016-09-01 | ソニー株式会社 | 撮像装置と画像処理装置と画像処理方法 |
JP6801647B2 (ja) * | 2015-05-22 | 2020-12-16 | ソニー株式会社 | 画像処理装置および方法 |
JP6760298B2 (ja) * | 2015-09-30 | 2020-09-23 | ソニー株式会社 | 情報取得装置と情報取得方法 |
JP6556013B2 (ja) * | 2015-10-08 | 2019-08-07 | キヤノン株式会社 | 処理装置、処理システム、撮像装置、処理方法、プログラム、および記録媒体 |
US9741163B2 (en) | 2015-12-22 | 2017-08-22 | Raytheon Company | 3-D polarimetric imaging using a microfacet scattering model to compensate for structured scene reflections |
JP6816768B2 (ja) * | 2016-07-22 | 2021-01-20 | ソニー株式会社 | 画像処理装置と画像処理方法 |
JP6877936B2 (ja) * | 2016-09-28 | 2021-05-26 | キヤノン株式会社 | 処理装置、処理システム、撮像装置、処理方法、プログラム、および記録媒体 |
KR102149180B1 (ko) * | 2017-07-07 | 2020-08-28 | 한국전자통신연구원 | 증강현실을 위한 가상 콘텐츠 합성 방법 및 이를 위한 장치 |
US11598715B2 (en) * | 2017-11-24 | 2023-03-07 | Sony Corporation | Detection apparatus and method of producing electronic apparatus |
CN111198445B (zh) * | 2018-11-16 | 2021-10-01 | 华为技术有限公司 | 一种分光偏振成像的设备和方法 |
US10659751B1 (en) * | 2018-12-14 | 2020-05-19 | Lyft Inc. | Multichannel, multi-polarization imaging for improved perception |
US11403811B1 (en) * | 2019-02-28 | 2022-08-02 | Apple Inc. | Estimating optical parameters of surfaces |
EP3709623A1 (en) * | 2019-03-15 | 2020-09-16 | Aptiv Technologies Limited | Method for simulating a digital imaging device |
CN110346116B (zh) * | 2019-06-14 | 2021-06-15 | 东南大学 | 一种基于图像采集的场景照度计算方法 |
CN110536067B (zh) * | 2019-09-04 | 2021-02-26 | Oppo广东移动通信有限公司 | 图像处理方法、装置、终端设备及计算机可读存储介质 |
KR102113166B1 (ko) * | 2019-10-14 | 2020-05-20 | (주)큐엘 | 이미지 글래스의 제조 시스템 및 방법 |
US20230342963A1 (en) * | 2019-12-13 | 2023-10-26 | Sony Group Corporation | Imaging device, information processing device, imaging method, and information processing method |
CN116245931B (zh) * | 2023-03-16 | 2024-04-19 | 如你所视(北京)科技有限公司 | 物品属性参数的确定方法、装置、设备、介质和产品 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005333336A (ja) * | 2004-05-19 | 2005-12-02 | Olympus Corp | 撮像装置 |
JP2006254331A (ja) * | 2005-03-14 | 2006-09-21 | Fuji Photo Film Co Ltd | 被写体の反射光成分を検出する画像処理方法およびこれを実行する装置 |
WO2007006630A1 (de) * | 2005-07-08 | 2007-01-18 | Grau Guenter | Verfahren zur herstellung von polarisationsfiltern sowie anwendung auf polarisationsempfindliche photosensoren und polarisationserzeugende wiedergabegeräte |
JP2007174689A (ja) * | 2005-09-09 | 2007-07-05 | Matsushita Electric Ind Co Ltd | 画像処理装置 |
Family Cites Families (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5028138A (en) | 1989-05-23 | 1991-07-02 | Wolff Lawrence B | Method of and apparatus for obtaining object data by machine vision form polarization information |
JP3727758B2 (ja) | 1997-07-17 | 2005-12-14 | 日本放送協会 | 映り込み防止形撮像装置 |
JPH11211433A (ja) | 1998-01-30 | 1999-08-06 | Toppan Printing Co Ltd | 成形品の表面形状測定方法及び測定システム |
DE19824460A1 (de) * | 1998-05-30 | 1999-12-02 | Zeiss Carl Jena Gmbh | Anordnung und Verfahren zur mikroskopischen Erzeugung von Objektbildern |
JP4294264B2 (ja) | 2002-04-25 | 2009-07-08 | 有限会社オートクローニング・テクノロジー | 集積型光学素子 |
WO2004110264A1 (ja) | 2003-06-11 | 2004-12-23 | Kose Corporation | 肌の評価方法および画像のシミュレーション方法 |
JP4347615B2 (ja) | 2003-06-11 | 2009-10-21 | 株式会社コーセー | 画像のシミュレーション方法 |
JP2007201541A (ja) * | 2006-01-23 | 2007-08-09 | Ricoh Co Ltd | 画像処理装置、撮像装置、画像処理方法、および画像処理プログラム |
US20080151194A1 (en) * | 2006-01-31 | 2008-06-26 | Avner Segev | Method and System for Illumination Adjustment |
JP4077869B2 (ja) * | 2006-05-29 | 2008-04-23 | 松下電器産業株式会社 | 光源推定装置、光源推定システムおよび光源推定方法、並びに、画像高解像度化装置および画像高解像度化方法 |
KR20090041402A (ko) * | 2006-07-18 | 2009-04-28 | 더 트러스티스 오브 더 유니버시티 오브 펜실바니아 | 편광을 사용해서 중첩되어 드리워진 그림자 성분들의 분리와 콘트라스트 강화 및 그림자 내의 표적 탐지를 위한 방법 |
CN100510700C (zh) * | 2006-12-30 | 2009-07-08 | 清华大学深圳研究生院 | 线偏振光成像方法及装置 |
EP2202688B1 (en) * | 2007-02-13 | 2013-11-20 | Panasonic Corporation | System, method and apparatus for image processing and image format |
US8718336B2 (en) * | 2007-03-30 | 2014-05-06 | Fujifilm Corporation | Image correction apparatus, image correction method, skin diagnosis method, and computer program product |
CN101584222B (zh) * | 2007-05-31 | 2011-10-12 | 松下电器产业株式会社 | 图像处理装置 |
US8025408B2 (en) * | 2008-07-08 | 2011-09-27 | Panasonic Corporation | Method, apparatus and program for image processing and method and apparatus for image synthesizing |
JP5487597B2 (ja) * | 2008-11-13 | 2014-05-07 | セイコーエプソン株式会社 | 画像処理装置、画像表示装置及び画像処理方法 |
US8437537B2 (en) * | 2009-03-27 | 2013-05-07 | Mitsubishi Electric Research Laboratories, Inc. | Method and system for estimating 3D pose of specular objects |
-
2009
- 2009-05-15 US US12/672,940 patent/US8025408B2/en active Active
- 2009-05-15 CN CN2009801064169A patent/CN101960859B/zh active Active
- 2009-05-15 WO PCT/JP2009/002165 patent/WO2010004677A1/ja active Application Filing
- 2009-05-15 JP JP2009549317A patent/JP4469021B2/ja active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005333336A (ja) * | 2004-05-19 | 2005-12-02 | Olympus Corp | 撮像装置 |
JP2006254331A (ja) * | 2005-03-14 | 2006-09-21 | Fuji Photo Film Co Ltd | 被写体の反射光成分を検出する画像処理方法およびこれを実行する装置 |
WO2007006630A1 (de) * | 2005-07-08 | 2007-01-18 | Grau Guenter | Verfahren zur herstellung von polarisationsfiltern sowie anwendung auf polarisationsempfindliche photosensoren und polarisationserzeugende wiedergabegeräte |
JP2007174689A (ja) * | 2005-09-09 | 2007-07-05 | Matsushita Electric Ind Co Ltd | 画像処理装置 |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013520675A (ja) * | 2010-02-25 | 2013-06-06 | ヴォロテック・リミテッド | 様々な偏光角度を有する光フィルタ及び処理アルゴリズム |
AU2011242408B2 (en) * | 2010-04-21 | 2016-12-15 | Commonwealth Scientific And Industrial Research Organisation | Shape and photometric invariants recovery from polarisation images |
CN102742258A (zh) * | 2010-07-21 | 2012-10-17 | 松下电器产业株式会社 | 图像处理装置 |
CN102742258B (zh) * | 2010-07-21 | 2016-10-26 | 松下知识产权经营株式会社 | 图像处理装置 |
US10397544B2 (en) | 2010-08-19 | 2019-08-27 | Nissan Motor Co., Ltd. | Three-dimensional object detection device and three-dimensional object detection method |
JPWO2012039086A1 (ja) * | 2010-09-24 | 2014-02-03 | パナソニック株式会社 | 画像処理装置 |
JP5450826B2 (ja) * | 2010-09-24 | 2014-03-26 | パナソニック株式会社 | 画像処理装置 |
US8913113B2 (en) | 2010-09-24 | 2014-12-16 | Panasonic Corporation | Image processing apparatus |
CN103037752A (zh) * | 2010-09-24 | 2013-04-10 | 松下电器产业株式会社 | 图像处理装置 |
JP2013009082A (ja) * | 2011-06-23 | 2013-01-10 | Nippon Telegr & Teleph Corp <Ntt> | 映像合成装置、映像合成方法、および映像合成プログラム |
JP2013045232A (ja) * | 2011-08-23 | 2013-03-04 | Ricoh Co Ltd | 車両運転支援装置、道路の路肩を検出する方法および該方法に基づく車両運転支援方法 |
JP2013245944A (ja) * | 2012-05-23 | 2013-12-09 | Nippon Telegr & Teleph Corp <Ntt> | 投影型3次元形状復元装置、投影型3次元形状復元方法、投影型3次元形状復元プログラムおよびそのプログラムを記録した記録媒体 |
JP2014042210A (ja) * | 2012-08-23 | 2014-03-06 | Nippon Telegr & Teleph Corp <Ntt> | 投影型3次元形状復元装置、投影型3次元形状復元方法及び投影型3次元形状復元プログラム |
JP2015059849A (ja) * | 2013-09-19 | 2015-03-30 | 凸版印刷株式会社 | 色と三次元形状の計測方法及び装置 |
JP2019020991A (ja) * | 2017-07-14 | 2019-02-07 | シャープ株式会社 | 撮像装置、認証装置、および撮像方法 |
CN111247794A (zh) * | 2017-10-30 | 2020-06-05 | 索尼半导体解决方案公司 | 固态摄像装置和电子设备 |
CN111247794B (zh) * | 2017-10-30 | 2023-03-24 | 索尼半导体解决方案公司 | 固态摄像装置和电子设备 |
JP2020139821A (ja) * | 2019-02-27 | 2020-09-03 | 株式会社新菱 | 検査装置、検査システム及び検査方法 |
JP7288273B2 (ja) | 2019-02-27 | 2023-06-07 | 株式会社新菱 | 検査装置、検査システム及び検査方法 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2010004677A1 (ja) | 2011-12-22 |
CN101960859B (zh) | 2013-04-10 |
US20100303344A1 (en) | 2010-12-02 |
CN101960859A (zh) | 2011-01-26 |
JP4469021B2 (ja) | 2010-05-26 |
US8025408B2 (en) | 2011-09-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4469021B2 (ja) | 画像処理方法、画像処理装置、画像処理プログラム、画像合成方法、および画像合成装置 | |
US8184194B2 (en) | Image processing apparatus, image division program and image synthesising method | |
JP4787871B2 (ja) | カラー偏光撮像装置 | |
US7948514B2 (en) | Image processing apparatus, method and computer program for generating normal information, and viewpoint-converted image generating apparatus | |
JP6456156B2 (ja) | 法線情報生成装置、撮像装置、法線情報生成方法および法線情報生成プログラム | |
US8564700B2 (en) | Image processing device, image processing method, and imaging device using a patterned polarizer | |
JP4324238B2 (ja) | 画像処理システム、方法及び装置 | |
JP4563513B2 (ja) | 画像処理装置及び擬似立体画像生成装置 | |
JP4762369B2 (ja) | 画像処理装置 | |
US8648907B2 (en) | Image processing apparatus | |
JP6422924B2 (ja) | 撮像装置および被写体情報取得方法 | |
US11676293B2 (en) | Methods for depth sensing using candidate images selected based on an epipolar line | |
KR20230031122A (ko) | 이미지 센서의 색상 왜곡을 평가하는 평가 장치의 동작하는 방법 | |
JP2017083778A (ja) | 画像投影装置、方法、及びプログラム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200980106416.9 Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2009549317 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12672940 Country of ref document: US |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09794125 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: 09794125 Country of ref document: EP Kind code of ref document: A1 |