JP2017194352A - Image processing device, image processing method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine an appropriate geometric condition for measuring a sense of luminosity.SOLUTION: Provided is an image processing device comprising: first image input means for inputting a first captured image in which a subject irradiated by a first light source is imaged; surface normal calculation means for calculating, for each pixel in the first captured image, a surface normal when it constitutes a luminescent spot and calculating a surface normal image; division means for dividing the captured image into a plurality of regions for each of prescribed angle ranges; luminescent spot probability calculation means for quantizing the first captured image and generating a luminescent spot image, and calculating, for each region, the ratio of a total area of luminescent spots included in the region to a total area of the region on the basis of the captured image and the luminescent spot image, and calculating luminescent spot probability that corresponds to the prescribed angle range; and geometric condition determination means for determining, upon imaging the subject illuminated by a second light source, a geometric condition for the second light source that allows each of luminescent spots in the obtained captured image to be separated, on the basis of the luminescent spot probability.SELECTED DRAWING: Figure 3

Description

  The present invention relates to an image processing technique for measuring the brightness of a subject.

  Conventionally, a method for measuring the glitter feeling of a coated plate having a glitter feeling such as a metallic material is known. Patent Document 1 discloses a method for quantitatively evaluating the glitter feeling by dividing each of the glitter feeling and the particle feeling. In the method disclosed in Patent Document 1, first, the glittering material-containing coating film surface is irradiated with light, and the coating film surface is photographed with a camera at an angle at which regular reflection light does not enter. Then, based on the luminance for each section obtained by analyzing the photographed image, an evaluation value representing the glitter feeling and the particle feeling is calculated.

JP 2000-304696 A

  The position and number of the glittering material-containing coating film surface are different depending on the observation angle. Therefore, in order to measure the glitter feeling, it is necessary to arrange the light source to irradiate and the camera to shoot under the geometric conditions that allow the position and number of light brightness to be appropriately imaged. However, the method disclosed in Patent Document 1 cannot always obtain an image suitable for evaluating the light luminance. SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to photograph a subject under an appropriate geometric condition for measuring glitter.

  In order to solve the above-described problems, the present invention is an image processing apparatus for a measuring apparatus that measures the brightness of a subject's bright spots, and is obtained by imaging a subject irradiated with a first light source. First image input means for inputting the captured image, and surface normal calculation means for calculating a surface normal image for each pixel in the first captured image by calculating a surface normal when it becomes a bright spot. And dividing means for dividing the captured image into a plurality of regions for each predetermined angle range; and generating a bright spot image representing a bright spot by quantizing the first captured image; and Based on the bright spot image, the ratio of the total area of bright spots included in the area to the total area of the area is calculated for each area, and the ratio for each area is used to calculate the ratio within the predetermined angle range. Bright spot probability calculator that calculates the corresponding bright spot probability And each of the bright spots in the captured image obtained when the subject illuminated by the second light source having a light distribution characteristic having higher directivity than the first light source is captured based on the bright spot probability. Geometric condition determining means for determining a geometric condition of the second light source that can be separated.

  According to the present invention, it is possible to photograph a subject under an appropriate geometric condition for measuring glitter.

Measuring device that measures the subject and the subject. The figure which shows the structure of a measuring device. FIG. 2 is a block diagram showing a detailed logical configuration of an image processing unit 215. The flowchart of the process which the image process part 215 performs. The figure explaining a surface normal image. The figure explaining the division | segmentation method of a surface normal image. The flowchart of a bright spot probability calculation process. The flowchart of a geometric condition determination process. The figure explaining a bright spot feature-value calculation. The figure explaining a bright spot image. The figure which shows the bright spot area ratio corresponding to an average elevation angle. The figure which shows the position of the 2nd illumination part 203. FIG. The figure which shows an example of a captured image. The flowchart of the process which the bright spot probability calculation 312 performs. The figure explaining the position of a 2nd illumination part. The figure which shows the concept of a surface normal range. The conceptual diagram of the area | region used for a bright spot probability calculation.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. Note that the configurations shown in the following embodiments are merely examples, and the present invention is not limited to the illustrated configurations.

<First Embodiment>
In the first embodiment, a method of irradiating a subject using a point light source and a parallel light source and measuring the glitter feeling based on the captured image will be described. FIG. 1A schematically shows a coated plate having a metal piece as the subject 101 coated on the surface. The surface of the subject 101 includes a plurality of metal pieces. Assume that the imaging unit 102 is arranged in a direction facing the subject 101, for example. As shown in FIG. 1A, in some metal pieces, reflected light irradiated from a certain irradiation angle reflects to the angle of the imaging device 102. At this time, when the imaging unit 102 captures the subject 101, a captured image 103 having a plurality of bright spots is obtained. In the present embodiment, the number, size, and intensity of the bright spots of the subject to be measured are measured as feature quantities indicating glitter. However, if the subject is not irradiated and photographed at an appropriate position, a plurality of bright spots that should be recorded in the photographed image as separate bright spots are collected and imaged as one bright spot. Therefore, in this embodiment, first, when imaging a subject irradiated with a parallel light source based on a pre-photographed image of a subject irradiated with a point light source, each of the bright spots of the subject can be separated in the captured image. The incident angle condition of the light source is calculated.

FIG. 2 is a diagram illustrating a configuration of a measurement apparatus applicable to the first embodiment. The imaging unit 201 includes a lens, a diaphragm, a shutter, an optical low-pass filter, a color filter, and a sensor such as a CMOS or a CCD. The imaging unit 201 detects the amount of reflected light reflected from the subject, and outputs a digital image to the bus 215 that is a data transfer path through A / D conversion. FIG. 1B shows the positional relationship between the imaging unit 201, the light source, and the subject in the measurement apparatus of the first embodiment. A point on the subject 101 is set as the origin (0, 0, 0). The imaging unit 201 is arranged at the position of the upper part C (0, 0, d) of the subject 101. The first illumination unit 202 and the second illumination unit 203 irradiate the subject with light. The second illumination unit 203 has a light distribution characteristic that is more directional than the first illumination unit 202. Here, the first illumination unit 202 is a point light source, and is arranged at a position of L ₁ (dsin θ ₁ , 0, dcos θ ₁ ) as shown in FIG. The second illumination unit 203 is two parallel light sources, and as shown in FIG. 1B, the position of the point L ₂ (dsin θ ₂ , 0, dcos θ ₂ ) and the point L ₃ (dsin θ ₃ , 0, dcos θ _3). ) And is arranged.

  The ROM 204 and RAM 205 provide the CPU 206 with programs, data, work areas, and the like necessary for imaging and image processing. The CPU 206 reads out and executes a program stored in the ROM 204 or RAM 205 using the RAM 205 as a work memory, and controls each component via the bus 215. Thereby, various processes described later are executed.

  The imaging control unit 207 controls focusing on the imaging unit 201, opening a shutter, adjusting a diaphragm, and the like. The operation unit 208 corresponds to a button, a mode dial, or the like. Upon receiving an instruction from the CPU 206, the imaging control unit 207 causes the imaging unit 201 to acquire a pre-captured image or a captured image and store the acquired image in a predetermined memory. Via the operation unit 208, the user can input various instructions such as power activation and processing parameter setting. The display unit 210 displays captured images and characters received from the image processing unit 215. The display unit 210 may have a touch screen function. In that case, the display unit 210 also functions as an operation unit 208 for inputting a user instruction. In the present embodiment, the display unit 210 uses a liquid crystal display. The external memory control unit 213 is an interface for connecting to a PC (personal computer) and other media 214 (for example, hard disk, memory card, CF card, SD card, USB memory).

  The image processing unit 215 calculates a glitter feature amount indicating the glitter feeling of the subject 101 based on the digital image obtained from the imaging unit 201. In the present embodiment, the image processing unit 215 is configured to be realized by the CPU 206 executing a predetermined program. The image processing unit 215 outputs information indicating the calculated glitter feature quantity to the bus 215.

  FIG. 3 is a block diagram illustrating a detailed logical configuration of the image processing unit 215. The pre-captured image input unit 306 inputs a pre-captured image obtained by capturing an image of the subject illuminated by the first illumination unit 202 under the point light source. The input pre-captured image is used not for calculating the bright spot feature quantity of the subject but for determining the position of the illumination that irradiates the subject. Hereinafter, the imaging of the pre-captured image is referred to as pre-imaging. The light source position information input unit 307 inputs information indicating the position of the first illumination unit 202 during pre-imaging. The imaging unit position information input unit 308 inputs imaging unit position information indicating the position of the imaging unit 201 at the time of pre-imaging. The imaging parameter input unit 309 inputs imaging parameters of the imaging unit 201 during pre-imaging.

  The surface normal calculation unit 310 calculates a surface normal in the case of a bright spot in the pre-captured image based on the light source position information, the imaging unit position information, and the imaging parameters, and indicates the surface normal direction for each pixel. A surface normal image is calculated. The calculated surface normal image is output to the region dividing unit 311. The area dividing unit 311 divides the pre-captured image for each predetermined threshold Δθ with respect to the elevation angle θ of the bright spot in the surface normal direction based on the surface normal image. The area dividing unit 311 outputs a plurality of images obtained by extracting only the divided areas as divided images. The bright spot probability calculation unit 312 calculates the probability that a bright spot will occur in each divided area from the pre-captured image and each divided area data. The calculated bright spot probability is output to the geometric condition determination unit 313. The geometric condition determination unit 313 is configured to capture the subject irradiated by the second illumination unit 203 (parallel light source) based on the bright spot probability. In the image obtained by separating each bright spot, the incident angle of the parallel light source can be obtained. Determine the conditions. The picked-up image input unit 314 inputs an image obtained by picking up an image of a subject irradiated with the second illumination unit 203 arranged at a position that satisfies the geometric condition. The input captured image is output to the bright spot feature amount calculation unit 315. The bright spot feature quantity calculation unit 315 calculates the intensity, number, and size of bright spots in the captured image as the bright spot feature quantity.

FIG. 4 is a flowchart of processing executed by the image processing unit 215 in the first embodiment. The CPU 206 reads out a program executable by the computer describing the flowchart shown in FIG. 4 from the ROM 204, writes the program on the RAM 205, and executes the program to execute the processing. In step S401, the pre-captured image input unit 306 inputs, as a pre-captured image, an image captured by the imaging unit 201 of the subject irradiated by the first illumination unit 202. The input pre-captured image is stored in the storage area of the ROM 204 or RAM 205. In step S <b> 402, the light source position information input unit 307 inputs light source position information indicating the position where the first illumination unit 202 (point light source) that has irradiated the subject at the time of pre-imaging is disposed. Here, as described above, the point L ₁ (dsin θ ₁ , 0, dcos θ ₁ ) is input as the light source position information of the first illumination unit 202. The light source position information input unit 207 stores the light source position information in a predetermined storage area.

  In step S403, the imaging unit position information input unit 308 inputs imaging unit position information indicating a position where the imaging unit 202 is arranged at the time of pre-imaging. In the present embodiment, the point C (0, 0, d) is input as the imaging unit position information. The imaging unit position information input unit 308 stores the input imaging unit position information in a predetermined storage area. In step S404, the imaging parameter input unit 309 inputs imaging parameters indicating the lens focal length of the imaging unit 202 during pre-imaging. The imaging parameter dormitory unit 209 stores the input imaging parameters in a predetermined storage area.

In step S405, the surface normal geometric condition calculation unit 310 calculates a surface normal image indicating the geometric condition of the surface normal of the bright spot in the pre-captured image based on the light source position information, the imaging unit position information, and the imaging parameters. . As shown in FIG. 5A, a metal piece having a surface normal in the direction in which the incident angle of the point light source and the reflection angle in the direction of the imaging unit 201 are equal appears as a bright spot in the pre-captured image. For the pixel k in the pre-photographed image, the surface normal direction _nk where the incident angle and the reflection angle of the point light source (first illumination unit 202) are equal is calculated according to the following equation (1).

W _k is the three-dimensional coordinate of the pixel k in the pre-captured image data calculated from the imaging unit position information and the imaging parameters. The three-dimensional coordinate Wk of the pixel k is calculated by a well-known projective transformation process using the imaging unit position information input in step S403 and the imaging parameter input in step S404. i _k is a vector of size 1 representing the point light source direction at the position indicated by the three-dimensional coordinates of the pixel k. v _k is a vector of size 1 representing the direction of the imaging unit 201 at the position indicated by the three-dimensional coordinates of the pixel k. In the present embodiment, it is assumed that the alignment characteristic of the glitter material is isotropic with respect to the azimuth angle, and the elevation angle θ corresponding to the pixel k is calculated using the following equation (2). The elevation angle θ calculated here is a geometric condition in the surface normal direction that the metal piece can take when it becomes a bright spot for each pixel in the pre-captured image.

The result of calculating the elevation angle θ for each pixel in the captured image is defined as a surface normal image. FIG. 5B is a diagram illustrating a surface normal image. The surface normal image is an image represented by an 8-bit gray scale. FIG. 5B shows a pixel value 0 in black and a pixel value 255 in white. In the surface normal image, since the elevation angle θ = 0 at the pixel position W _k0 on the X axis and at the same distance from the point light source (first illumination unit 202) and the imaging unit 201, the pixel value is zero. This means that the surface normal n _k0 = (0, 0, 1) of the pixel position W _k0 . The surface normal calculation unit 310 stores the calculated surface normal image in a predetermined storage area.

  In step S406, the region dividing unit 311 divides the surface normal image into m regions for each predetermined angular range Δθ of the surface normal of the bright spot. FIG. 6 is a diagram for explaining a method for dividing a surface normal image. First, each pixel in the surface normal image calculated in step S405 is quantized at intervals of Δθ. For example, Δθ = 1 °. FIG. 6A schematically shows the result of quantizing the surface normal image. Based on the quantized surface normal image, the image is separated into images for each quantized value as shown in FIG. Note that the pixel values of the pixels in each divided image are not the quantized values but the elevation angles before the quantization. Output multiple divided images. The area dividing unit 311 stores the calculated plurality of divided images in a predetermined storage area.

  In step S407, the bright spot probability calculation unit 312 calculates the probability that a bright spot will occur for each divided image. Details of the bright spot probability calculation processing will be described later. The bright spot probability calculation unit 312 stores the calculated bright spot probability for each divided image in a predetermined storage area. In step S408, the geometric condition determination unit 313 determines, as the geometric condition, the incident angle condition of the second illumination unit 203 that allows each bright spot to be separated in the captured image, based on the bright spot probability for each divided image. Details of the geometric condition determination processing will be described later. The geometric condition determination unit 313 outputs the calculated geometric condition to the second illumination unit 203.

In step S409, the captured image input unit 314 inputs a captured image obtained by the imaging unit 201 capturing the subject irradiated by the second illumination unit 203 arranged at a position that satisfies the geometric condition output in step S408. In this embodiment, inputs the second signature portion 203 (parallel light) captured image obtained by capturing an image of a subject and irradiated disposed at a position on the geometrical conditions are satisfied point L ₃ calculated as shown in FIG. The input captured image is stored in a predetermined storage area.

  In step S410, the bright spot feature amount calculation unit 315 calculates the intensity, the number, and the size of the bright spot of the subject as the bright spot feature amount based on the captured image acquired in step S409. FIG. 9 is a diagram for explaining the calculation of the bright spot feature quantity. FIG. 9A shows the captured image acquired in step S409. First, the bright spot feature quantity calculation unit 315 binarizes the captured image using a predetermined threshold value, and generates a binarized image as shown in FIG. 9B. Here, the threshold is assumed to be 128. A bright spot is labeled in the binarized image shown in FIG. Specifically, a region where pixels having a pixel value of 255 are connected in the binarized image is detected, and numbers are sequentially assigned to the detected regions. FIG. 9C shows the result of labeling each bright spot. The bright spot feature quantity calculation unit 315 calculates the average number of bright spots in the captured image and the area of each bright spot. Further, based on the captured image shown in FIG. 9A and the binarized image shown in FIG. 9B, the pixel value of the pixel in the region that is not a bright spot is set to 0 as shown in FIG. 9D. The average intensity of the bright spots is calculated by generating the image and dividing the sum of the pixel values by the number of bright spots. The bright spot feature amount 315 stores the calculated average intensity, number, and average area of bright spots as a bright spot feature amount in a predetermined storage area, and ends the process.

Next, details of the bright spot probability calculation process executed by the bright spot probability calculation unit 312 in step S407 will be described. FIG. 7 is a flowchart of the bright spot probability calculation process. In step S701, the bright spot probability calculation unit 312 calculates an area R (θ _m ) of a region having a pixel value in a predetermined elevation angle range in each divided image shown in FIG. In step S <b> 406, the region dividing unit 311 separates each pixel group having pixel values in a predetermined elevation angle range. The bright spot probability calculation unit 312 calculates the area of a pixel group having pixel values in a corresponding predetermined elevation angle range in each divided image. Specifically, pixels having pixel values in a corresponding predetermined elevation angle range are counted, and the number of pixels is regarded as an area. When the number of divided areas is N, m is a natural number of 1 to N. The angle θ _m is an average elevation angle calculated by the average value of the elevation angles of each pixel included in each divided region. Here, the bright spot probability calculation unit 312 stores the calculated area area in a predetermined storage area.

In step S702, the bright spot probability calculation unit 312 binarizes the pre-captured image shown in FIG. 10A with a predetermined threshold (128 here), so that the bright spot has a clear bright spot position. Generate an image. FIG. 10B shows an example of a bright spot image. Further, the bright spot image is divided as shown in FIG. 10C in the same manner as the pixel group having a pixel value in a predetermined elevation angle range corresponding to each divided image. The total area of the bright spots is calculated as the bright spot total area S (θ _m ). Specifically, a pixel having a pixel value of 255 is detected as a bright spot in the binarized image, and the total number of bright spots is calculated by counting the number of pixels having a pixel value of 255. The bright spot probability calculation unit 312 stores the calculated bright spot total area S (θ _m ) for each average elevation angle in a predetermined storage area.

In step S703, the bright spot probability calculation unit 312 uses the following formula (3) based on the ratio between each area R (θ _m ) and the bright spot total area S (θ _m ) to obtain the bright spot area ratio P (Θ _m ) is calculated.

FIG. 11 is a graph showing the bright spot area ratio corresponding to the average elevation angle. The horizontal axis represents the elevation angle, and the vertical axis represents the bright spot area ratio. In the present embodiment, as shown in FIG. 11, the bright spot area ratios are plotted at intervals of Δθ with respect to the elevation angle θ. A function P (θ _m ) indicating the bright spot area ratio is calculated by approximating the plotted results as a function. The bright spot probability calculation unit 312 stores the bright spot area ratio function P (θ _m ) in a predetermined storage area.

  In step S <b> 704, the bright spot probability calculation unit 312 inputs the following expression (4) as a Beckman distribution model D (θ) that models the distribution of the minute surface normal representing the roughness of the surface of the subject 101.

α is an average inclination of the minute surface and is a parameter representing the surface roughness. As shown in the following equation (5), the Beckmann distribution is 1 when integrated in the upper hemisphere region Ω _N with the subject's macroscopic surface normal direction N = (0, 0, 1) as the zenith direction. Has been standardized.

Therefore, the Beckman distribution model D (θ) can be regarded as a probability that a bright spot having a surface normal of the elevation angle θ is generated. The bright spot probability calculation unit 312 stores the input Beckman distribution model D (θ) in a predetermined storage area.

In step S705, the bright spot probability calculation unit 312 uses the following equation (6) to calculate the surface roughness parameter α when fitting the bright spot area ratio function P (θ _m ) with the Beckman distribution model data D (θ). Is calculated.

t is a coefficient that normalizes the bright spot area ratio function P (θ _m ). The bright spot probability calculation unit 312 stores the calculated surface roughness parameter α as information indicating the bright spot probability in a predetermined storage area, and ends the process.

Hereinafter, the details of the geometric condition determination process executed by the geometric condition determination unit 313 in step S408 will be described. FIG. 8 is a flowchart of the geometric condition determination process. In step S <b> 801, the geometric condition determination unit 313 determines the surface normal direction of the bright spot where each bright spot can be separated in the captured image. Specifically, the following equation (7) is used to calculate the elevation angle θ of the surface normal n where the probability that a bright spot having the same surface normal exists in the vicinity of the eight connected points of each bright spot is equal to or less than a predetermined threshold. Calculate the range. Here, the threshold value P _th is assumed to be 0.1.

In the present embodiment, the Beckman distribution model D _α (θ) when the surface roughness parameter calculated in step S407 is α is used as a function representing the bright spot probability. As shown by the hatched portion in FIG. 11, the range of the elevation angle θ having the bright spot probability equal to or less than the predetermined threshold value P _th is θ> θ _{th, which} is the range of the surface normal n where each bright spot can be separated. The geometric condition determination unit 313 stores the calculated range of the elevation angle θ of the surface normal n in a predetermined storage area.

In step S <b> 802, the geometric condition determination unit 313 determines a geometric condition that enables each bright spot to be separated in a captured image obtained by capturing an image of a subject illuminated by a parallel light source. Specifically, the range of the incident angle θ _L of the parallel light source is calculated using the following formula (8) in the range of the elevation angle θ of the surface normal n calculated in step S801.

In the present embodiment, it is assumed that the camera direction at the position indicated by the three-dimensional coordinate corresponding to each pixel position in the captured image is a constant value of the camera direction v _k1 = (0, 0, 1) at the center of the angle of view. From Equation (8), v _k1 = (0, 0, 1), θ _L = 2θ. Therefore, the second illumination unit 203 has a geometric condition where θ _L > 2θ _{th in} the range of the incident angle θ _L. . As described above the second illumination unit 203 has two parallel light sources arranged in a point L ₃ of elevation theta ₂ or in a point L ₂ and elevation theta _3. As shown in FIG. 12, the elevation angle 2θ _t satisfies θ ₂ <2θ _th <θ ₃ . So by irradiating the subject 101 and the second illumination unit 203 into a parallel light source disposed in the position of the big points L ₃ angles from 2 [Theta] _th, by imaging an object 101, transferring the bright spots properly photographed image Can do. Geometric-condition determination section 313, the calculated geometric condition information indicating a range of the incident angle theta _L parallel light source and stored in a predetermined storage area, the process ends.

FIG. 13A is a captured image obtained as a result of imaging a subject that is arranged at the position of the point L _{2 in} the second illumination unit 203 and is illuminated by a parallel light source. Since the incident θ2 of the parallel light source arranged at the position of the point L2 is small, some of the bright spots are connected to the adjacent bright spots, and the measurement accuracy of the number and area of the bright spots is lowered. FIG. 13B is a captured image obtained by capturing an image of a subject illuminated by the parallel light source arranged at the position of the point L _{3 in} the second illumination unit 203. Since irradiation is performed from an arrangement where each bright spot is separable, the number and area of bright spots can be obtained with high accuracy. As described above, according to the present embodiment, the geometric condition for arranging the parallel light source (second illumination unit 203) is determined based on the result of pre-imaging using the point light source (first illumination unit 202). This makes it possible to acquire the brightness of the subject with high accuracy.

  In the first embodiment, it is assumed that the orientation characteristics of the glittering material are isotropic with respect to the azimuth angle, and the region obtained by dividing the surface normal image with respect to the elevation angle at Δθ intervals is calculated. However, a region divided for both the elevation angle and the azimuth angle may be calculated. For example, the surface normal image may be divided into blocks.

  In this embodiment, the bright spot probability is calculated by fitting the Beckman distribution model data to the bright spot area ratio function, but the model is not particularly limited to the Beckman distribution model. For example, a function normalized by fitting the luminous spot area ratio with a polynomial function and integrating the elevation angle θ in the range of 0 ° to 90 ° to 1 may be calculated as the bright spot probability function.

  In the present embodiment, as the geometric condition information, the parallel light source direction in which each bright spot is separable with respect to the camera direction at the pixel at the center of the angle of view of the captured image is calculated. May be calculated. In addition, a common range in the parallel light source direction for each calculated pixel may be calculated as geometric condition information.

  In the present embodiment, the brightness of the subject is acquired using a measuring device in which one first illumination unit 202 that is a point light source and two second illuminations 203 that are parallel light sources are arranged. There is no limitation on the number or position of the positions. For example, in a measuring device in which one point light source and three or more parallel light sources are arranged, a plurality of subjects irradiated by each of a plurality of parallel light sources satisfying the geometric condition are imaged, and the subject corresponding to the plurality of light source directions The brightness may be measured. Alternatively, one parallel light source may be arranged on the drive stage as the second illumination unit 203 and moved to a position that satisfies the geometric condition to measure the glitter of the subject.

  In this embodiment, the brightness of the subject is measured using one imaging unit 201. However, the brightness of the subject is measured using two cameras, the pre-imaging imaging unit and the imaging unit. Also good. For example, in order to widen the surface normal range for calculating the probability of bright spots during pre-imaging, images are taken with a camera equipped with a wide-angle lens. In order to obtain the image, the image may be captured by a camera equipped with a telephoto lens. Alternatively, a zoom lens may be attached to one imaging unit 201, and imaging may be performed by changing the magnification during pre-imaging and imaging.

  In this embodiment, the bright spot probability is calculated from the bright spot area ratio in each region divided according to the surface normal direction of the bright spot. However, for example, the histogram of the area of each bright spot in the divided region is calculated, and the bright spot probability is calculated from the variance value of the histogram using a pre-calculated dispersion value and bright spot probability lookup table. You can also.

Second Embodiment
In the first embodiment, a method has been described in which the point light source is regarded as an ideal diffused light source that does not have the size of light to be irradiated, and the brightness of the subject is measured. In the second embodiment, a case where a point light source having a finite size is used as the first illumination unit 202 will be described. In addition, about the structure similar to 1st Embodiment, the same symbol is attached | subjected and detailed description is abbreviate | omitted. In the second embodiment, the bright spot probability calculation process in step S407 is different from the first embodiment. FIG. 14 is a flowchart of processing executed by the bright spot probability calculation 312 applicable to the second embodiment.

In step S1401, the bright spot probability calculation unit 312 inputs information indicating the light intensity of the point light source used as the first illumination unit 202. In the second embodiment, as shown in FIG. 15, a circular light source having a radius r parallel to the XY plane is used as a point light source. The bright spot probability calculation unit 312 stores information indicating the size of the input point light source in a predetermined storage area. In step S <b> 1402, the bright spot probability calculation unit 312 captures, for each pixel constituting the pre-captured image, a surface normal range in which specular reflection light in which the light emitted from the point light source is reflected in the specular reflection direction on the subject 101 is imaged. Is calculated. Specifically, the surface normal range Δθ _S is calculated using the following equation (9) by the cosine theorem.

L ₁₋ and L ₁₊ are intersection points of a circular point light source L ₁ having a radius r and a plane α perpendicular to the XY plane shown in FIG. 15 and passing through the three-dimensional coordinates W _k of the pixel _k and the light source center position L _1. is there. Therefore, the distances | W _k L ₁₋ , |, | W _k L ₁₊ | between the point Wk−the point L ₁₋ and the point Wk−the point L ₁₊ can be calculated by the following equations (10) and (11).

W _kx and W _ky are the X coordinate and Y coordinate of the three-dimensional coordinate W _k of the pixel k. A surface normal range Δθ _S shown in FIG. 16 is calculated using the equations (9) to (11). The surface normal range Δθ _S is an image represented by an 8-bit gray scale. In the surface normal range, the specularly reflected light of the light emitted from the point light source is an image in which the pixel value increases in the vicinity of the light source position L ₁ where the surface normal range reflected in the imaging unit 201 increases. The bright spot probability calculation unit 312 stores the calculated surface normal range Δθ _S in which regular reflection occurs in a predetermined storage area.

In step S1403, the bright spot probability calculation unit 312 calculates a region smaller than the predetermined angle range Δθ obtained by dividing the elevation angle θ in step S406 with respect to the surface normal range Δθ _S as a region used for the bright spot probability calculation. In the present embodiment, it satisfies the .DELTA..theta.s <[Delta] [theta] in the region distant from the light source position L ₁ shown in FIG. 17, and is calculated as the area used for calculating bright spot probability. The calculated area is stored in a predetermined storage area. As described above, when pre-imaging a subject irradiated with a point light source having a finite size, a region where the calculation accuracy of the bright spot area ratio is reduced due to the size of the light source is excluded. This makes it possible to calculate with high accuracy a geometric condition that can appropriately measure the bright spot of the subject. In this embodiment, a circular surface light source having a radius r is used as a point light source having a finite size. However, a square or spherical light source may be used, and the shape and size of the light source are not particularly limited.

<Other embodiments>
In the above-described embodiment, each configuration of the image processing unit 215 has been described by taking a software process executed by the CPU as an example. However, the image processing unit 215 may be realized by hardware by a part of the processing unit or all the dedicated processing circuits.

306 Pre-captured image input unit 310 Surface normal geometric condition calculation unit 311 Region division unit 312 Bright spot probability calculation unit 313 Geometric condition determination unit 314 Captured image input unit 315 Bright spot feature amount calculation unit

Claims (8)

An image processing device for a measuring device that measures the radiance of a luminescent spot of a subject,
First image input means for inputting a first captured image obtained by imaging a subject irradiated with a first light source;
Surface normal calculation means for calculating a surface normal image by calculating a surface normal for each pixel in the first captured image,
Dividing means for dividing the captured image into a plurality of regions for each predetermined angle range;
A bright spot image representing a bright spot is generated by quantizing the first captured image, and each area is included in the area with respect to the total area of the area based on the captured image and the bright spot image. A bright spot probability calculating means for calculating a bright spot probability corresponding to the predetermined angle range using a ratio for each region,
Based on the probability of bright spots, each of the bright spots can be separated in the obtained captured image when the subject illuminated by the second light source having a light distribution characteristic having higher directivity than the first light source is captured. An image processing apparatus comprising: geometric condition determining means for determining a geometric condition of the second light source. Second input means for inputting a second captured image obtained by imaging the subject illuminated by the second light source arranged based on the geometric condition determined by the geometric condition determining means;
Feature amount calculating means for calculating a bright spot feature amount indicating the feature of the subject's glitter feeling based on the second captured image;
The image processing apparatus according to claim 1, further comprising:   The image processing apparatus according to claim 1, wherein the first light source is a point light source, and the second light source is a parallel light source.   The geometric condition determination means calculates the range of the elevation angle θ of the surface normal n where the probability that a bright point having the same surface normal exists in the vicinity of the eight connected points of each bright point is equal to or less than a predetermined threshold value. The image processing apparatus according to claim 1, wherein a geometric condition is determined.   The said bright spot probability calculation means calculates the said bright spot probability using the Beckman distribution model showing distribution of the surface normal in the said subject | subject. Image processing device.   A program that causes a computer to function as the image processing apparatus according to claim 1 by being read and executed by a computer. An image processing apparatus according to any one of claims 1 to 4,
An imaging unit for imaging a subject from a predetermined position;
A measuring device having a first light source and a second light source different from the first light source. An image processing method for a measuring device for measuring a glitter feeling composed of a bright spot of a subject,
Inputting a first captured image obtained by imaging a subject irradiated with a first light source;
For each pixel in the first captured image, calculate a surface normal when it becomes a bright spot, calculate a surface normal image,
The captured image is divided into a plurality of regions for each predetermined angle range;
A bright spot image representing a bright spot is generated by quantizing the first captured image, and each area is included in the area with respect to the total area of the area based on the captured image and the bright spot image. Calculating the ratio of the total area of the bright spots to be calculated, using the ratio for each region, calculating the bright spot probability corresponding to the predetermined angle range,
Based on the probability of bright spots, each of the bright spots can be separated in the obtained captured image when the subject illuminated by the second light source having a light distribution characteristic having higher directivity than the first light source is captured. An image processing method comprising: determining a geometric condition of the second light source.