KR101021027B1 - Stereoscopic 3D Measuring System and Method for Efficiency Evaluation of Skin Products - Google Patents

Abstract

The present invention relates to a three-dimensional image measurement system and method for measuring and evaluating the improvement of skin wrinkles, in particular to measure the efficacy of cosmetics, functional cosmetics, quasi-drugs, functional foods, dietary supplements, pharmaceuticals, etc. (skin improvement products) In order to evaluate and evaluate the skin wrinkles before and after the use of skin improvement products such as cosmetics with two cameras to obtain a stereo image, and reconstruct the two stereo images into stereoscopic 3D, the length and width of the skin wrinkles In addition, to provide a three-dimensional image measuring system and method for skin wrinkle evaluation for measuring and determining the efficacy of cosmetics or medicines by measuring the depth and volume.

The stereoscopic image measuring system for skin wrinkle evaluation according to the present invention includes two DSLR cameras for photographing a planar image (2D); A tripod provided with a case for allowing the two DSLR cameras to be installed on the same plane at a predetermined interval; It is connected to the two DSLR cameras and stores and displays stereo images taken by two DSLR cameras, and matches two stereo images taken by two DSLR cameras to obtain stereoscopic (3D) skin wrinkle data. Characterized in that it comprises a; computer installed software for generating and measuring.

In addition, the stereoscopic image measuring method for skin wrinkle evaluation according to the present invention includes a stereo image obtaining step (Acquire Images), a 3D image generating step, and a 3D image measuring step. The image generation step includes: a match matching point step, a f matrix calculation step, an image correction step (S230), a dense matching step, and a stereoscopic image reconstruction (3D) Reconstruction).

3D, 3D, Skin, Cosmetics, Medicines

Description

Stereoscopic 3D Measuring System and Method for Efficiency Evaluation of Skin Products}

The present invention relates to a three-dimensional image measuring system and a three-dimensional image measuring method for measuring and evaluating the improvement of skin wrinkles, in particular of cosmetics, functional cosmetics, quasi-drugs, functional foods, dietary supplements, pharmaceuticals (skin improvement products) In order to measure and judge the efficacy, the skin wrinkles before and after the use of skin improvement products such as cosmetics are photographed with two cameras to obtain stereo images, and the stereo images are reconstructed into stereoscopic 3D to determine the skin wrinkles. To provide a stereoscopic image measuring system and method for skin wrinkle evaluation for measuring and determining the efficacy of cosmetics or medicines by measuring depth and volume as well as length and width.

In recent years, as beauty of life is improved and quality of life is increasing, interest in skin beauty is rapidly increasing. In particular, clean skin without white wrinkles has emerged as a major criterion for distinguishing beautiful women, so that the new word “skin beauty” is created through the internet and the mass media. And investing money.

Therefore, companies that manufacture cosmetics, foods, and medicines (hereinafter collectively referred to as 'skin improvement products') that are used to improve skin beauty or skin quality are also taking a lot of time and effort to develop more effective products. Investing. What is essential to this effort, however, is the precise determination of the efficacy after using the skin remedial products. To this end, a technique for comparing and evaluating a skin image or skin wrinkle image obtained by using a skin improvement product and a skin image or skin wrinkle image obtained after using a skin improvement product is required as a prior task.

However, the conventional drug efficacy measurement method has a difficulty in accurately measuring the efficacy according to the use of cosmetics because it uses a two-dimensional visual information (2D). For example, the 2D skin wrinkle image shows the length and width of the wrinkles but not the depth of the wrinkles. Therefore, in order to measure the depth of wrinkles, the use of expensive 3D measuring equipment becomes inevitable. However, these 3D measuring equipment is not only a very expensive product in terms of cost, but also has a problem in that accurate measurement is difficult in areas requiring close-up photography such as wrinkles on the skin.

Meanwhile, a method of obtaining a 3D image, that is, a stereoscopic image, may be divided into a real 3D image and a stereoscopic 3D image. However, since it is expensive to make a real 3D image using a stereoscopic camera, binocular parallax is mainly used. The stereoscopic image obtained by the binocular disparity method is called a stereo-scopic image.

The human eyes are about 65mm in the horizontal direction, and each time you look at an object, the left and right eyes are slightly different from each other in the retina, and the brain interprets this tiny difference to feel a three-dimensional image. This is called binocular disparity and is the most important factor in feeling the three-dimensional effect. The left and right eyes see different two-dimensional images, and these two images are transmitted to the brain through the retina, and the brain accurately fuses each other to reproduce the perspective and reality of the 3D image. The present invention relates to the reconstruction of two stereo images into 3D images using a stereoscopic 3D method.

The principle of producing 3D stereoscopic image is a photographing equipment including two cameras to acquire two stereo images of a single object using binocular parallax, and two two-dimensional images captured by two cameras. There is a need for a wrinkle data generation program that edits or composes and converts them into 3D images.

That is, the process of generating a 3D stereoscopic image is a process of creating an image pair from the same object with two cameras located at spatially separated locations as if two eyes recognize the 3D, and a plane that is spatially different from each other. It consists of the process of reconstructing the three-dimensional image from the two-dimensional image pair obtained from. In other words, if image points corresponding to the same point in the three-dimensional space are matched with each other, a 3D stereoscopic image may be obtained using a geometric model.

The present invention is to solve the problems of the prior art, the main object of the present invention using a low-cost measurement device, to obtain two stereo images, and to match the two stereo images obtained to generate 3D skin wrinkle data To provide a three-dimensional image (3D) measurement system and method for skin wrinkle evaluation to measure and determine the efficacy of the cosmetic or drug by measuring the depth and volume as well as the length and width of the wrinkles before and after use of the cosmetic or drug will be.

The present invention is to reconstruct a three-dimensional image from a two-dimensional stereo image. This is called image-based modeling. In order to acquire a 3D image, information and conditions according to 3D are required, and the obtained image may be generated as a 3D image through a series of processing processes. After all, the depth sensing (image depth) of the real object is an element for generating a 3D image.

As a method of sensing the depth of the real object, which is information for acquiring a 3D image, there are a parallel stereoscopic imaging and a cross-visual stereoscopic imaging. Here, in the stereoscopic stereo system, when two lens projection center lines are placed in parallel, an image of the same object is an epipolar line generated by this object point (a line formed by two lens centers and an image plane). Based on. At this time, the depth of the real object can be detected by the distance on the lens center and the focal length and the distance between the projection objects on the image plane through the two lenses. After all, in order to obtain a stereoscopic image, at least two images based on the parallel stereoscopic method or the cross-visual stereo type are required, because the depth of the stereoscopic object and the image calculation according to the depth are required.

The stereoscopic image measuring system for skin wrinkle evaluation according to the present invention includes two DSLR cameras for photographing a planar image (2D); A tripod with a case in which the two DSLR cameras are installed in parallel at a predetermined interval apart; It is connected to the two DSLR cameras and stores and displays stereo images taken by two DSLR cameras, and matches two stereo images taken by two DSLR cameras to obtain stereoscopic (3D) skin wrinkle data. It is configured to include; computer that generates a program is installed.

In addition, the stereoscopic image measuring method for skin wrinkle evaluation according to the present invention includes a stereo image obtaining step (Acquire Images), a 3D image generating step, and a 3D image measuring step.

In the present invention, the 3D image generation step, the Match Matching Point step, F Matrix Computation step, Image Rectification step (S230), Dense Matching step 3D reconstruction.

In the present invention, in order to effectively find the feature points of the skin wrinkle image, the matching points are searched using edges instead of corners.

The stereoscopic image measuring system and method for skin wrinkle evaluation according to the present invention is used to acquire two stereo images by macro shooting through remote control. The stereo images are then combined to generate 3D wrinkle data and measured for depth, length, and volume of skin wrinkles. Due to the measurement of the 3D wrinkle data, it is possible to measure more accurately than conventional measuring equipment that cannot be photographed close-up, it can bring a cost reduction effect on the existing equipment.

Hereinafter, a low-cost 3D measurement system and method for skin wrinkle evaluation according to the present invention will be described in detail with reference to the accompanying drawings.

First, Figure 1 shows the configuration of a low-cost 3D measurement system for skin wrinkle evaluation according to the present invention. The low-cost stereoscopic image (3D) measurement system for skin wrinkle evaluation according to the present invention is a three-dimensional imaging device for recording two stereo images at the same time, and stores two stereo data images obtained from the three-dimensional imaging device It includes a display box and a computer (PC) equipped with software for registering two stereo images to generate a 3D skin wrinkle image.

In the stereoscopic image photographing apparatus used in the present invention, two ordinary DSLR cameras are attached to a camera case to take a photograph. The photographing lens used at this time uses a macro lens suitable for skin wrinkle measurement. Therefore, close-up photography, which is a problem of the existing measuring equipment, is possible, and each of the corresponding components is composed of low-cost parts, which can bring about cost reduction effects.

That is, a person feels a three-dimensional feeling by looking at an object at different angles from the left eye and the right eye. Based on this principle, three-dimensional images can be artificially implemented by showing two stereo images taken separately from the left and right eyes at the left and right eyes, respectively. According to the same principle, in order to produce a stereoscopic image, two subjects should be photographed using one camera.

2 shows a stereoscopic image photographing apparatus used in the present invention. As shown, the stereoscopic imaging apparatus of the present invention is composed of a tripod and a case for installing two DSLR cameras on the same plane. Preferably, the case is provided with a case bolster, camera bold and the like to maintain parallel.

FIG. 3 is a view of a screen on which software, which is one of software mounted on the computer (PC), for simultaneously recording two stereo images remotely. That is, in order to obtain a 3D skin wrinkle image, two stereo images must be obtained for one object (eg, skin wrinkles). Since a person who is a subject for obtaining skin wrinkle images is always in motion, two stereo images must be taken simultaneously. You can get an image. To this end, the present invention shoots using remote control software that can simultaneously shoot using a camera software development kit (SDK). The software not only allows you to shoot remotely from your PC, but also allows you to shoot two DSLR cameras simultaneously.

4 is a schematic flowchart showing a 3D measurement method for skin wrinkle evaluation according to the present invention. As shown, the 3D measurement method for skin wrinkle evaluation according to the present invention is largely, a stereo image acquisition step (Acquire Images) (S100), 3D image generation step (S200) and 3D image measurement step (S300) It is composed.

First of all, in order to generate a 3D skin wrinkle image, the process of acquiring a stereo image is important. The present invention uses a low-cost stereoscopic image capturing device for stereo images. As described above, the stereoscopic imaging apparatus according to the present invention includes two two DSLR cameras installed at a predetermined distance apart. For example, the two cameras are installed about 65 mm apart.

 As shown in Figure 5, the stereo image acquisition step (S100) is to take the image of the skin wrinkles of the subject using a low-cost stereoscopic imaging apparatus. That is, two cameras (camera 1 and camera 2) installed in the case are simultaneously remotely photographed to obtain two 2D skin wrinkle images. At this time, the distance between the subject and the camera for close-up photography is within 30cm, preferably 15 to 20cm. The two 2D skin wrinkle images thus measured cannot measure depth. 6 shows two stereo images actually taken.

Subsequently, the 3D image generating step (S200) is a step of generating a 3D skin wrinkle image having a depth by using the two 2D stereo images photographed.

The 3D image generating step (S200) is again a match matching point step (S210), F Matrix Computation step (S220), Image Rectification step (S230), dense matching ( A dense matching step (S240) and a stereoscopic image reconstruction (3D reconstruction) step (S250).

First, a find matching point step (S210) is a step of finding a match point to reconstruct two 2D skin wrinkle images into a 3D image. In general, the matching point search method is mainly a corner method. However, in the case of the skin image used in the present invention, it is very difficult to find the feature point that is the reference point of the registration point. That is, it is difficult to use the corner as a method of searching for a registration point in the skin image.

Accordingly, the present invention uses a method using an edge rather than a corner to effectively find a matching point. The present invention detects the edges of two skin wrinkle images using Canny Edge Detection, an edge detector.

Edge detection refers to clearly extracting positional information based on an algorithm of important visual information such as the boundary of an object in an image. In most cases, edge extraction masks are sensitive to noise, so even small noises are considered as edges, but canny masks are intended to extract strong edges by making them insensitive to noise.

Canny Edge Detector is a low noise filter preprocessing for eliminating noise, process of gradient operator to calculate edge strength and direction, and sobel operator Based on the data obtained by the method, the edge strength is determined using a thresholding technique (Thresholding in Canny Edge Detector), and the edge direction is formed by connecting the edge directions.

In other words, the Kenny edge detector first uses a Gaussian lowpass filter to remove noise. This filter is controlled by the value of sigma (σ). Then it goes through the Sobel operator. The Sobel operator consists of two high-pass filters, with the results obtained in each filter in vector form. The norm of this vector is taken to produce an edge energy image and then thresholded for this image to produce an edge map. 7 is a photograph showing edge detection results using an original skin wrinkle image and a canny mask.

After edge detection, unnecessary wrinkles are removed using the opening operation, which is performed in the order of morphology calculations, erosion, and dilation, in order to efficiently compare edges.

Morphophology is an image processing technique that analyzes and processes the shape of an image. Morphological processing is used to extract the image elements necessary to express or describe the shape of the boundary, block, and skeleton of the image.

Morphological erosion associations and expansion operations are the basis of other operations. Erosion is a reduction in the size of an object with respect to its background. Therefore, when there is a spark noise between the object and the background of the image, it is useful to remove this noise or to remove a very small object from the whole image. It is also useful when the projections in the image are reduced and the internal projections are increased to separate objects that touch each other. Dilation, on the other hand, is a technique for expanding the size of the object and reducing the background by increasing the protrusions inside the object and increasing the outside protrusions. Therefore, it is used to fill a space such as a hole generated inside an object or to connect a short broken area.

 The Opening operation uses the expansion operation immediately after the erosion operation, which removes the convex portions and breaks the narrow connections to smooth the outline portion of the image. However, the shape and size of the object is preserved. It is also called a removal operation because it removes protrusions and narrow connections. To make the outline part more smooth, repeat the erosion operation a certain number of times by applying the thermal operation repeatedly, and then repeat the expansion operation the same number of times.

The most basic of morphological operations, expansion and erosion, can be defined by the following equation.

로 정의되면, 물체 A와 구조요소(structuring element)B의 성분들의 모든 벡터 합을 말한다. 즉 A와 B의 성분들의 가능한 모든 평행이동의 합집합인 것이다. 따라서 물체 내부의 돌출부는 감소하고, 외부의 돌출부는 증가하는 결과를 가져오게 된다.Expansion

Is defined as the sum of all the vectors of the components of the object A and the structuring element B. That is, the union of all possible translations of the components of A and B. This results in a decrease in protrusions inside the object and an increase in outside protrusions.

로 정의되며, 팽창 연산과 반대로 평행 이동된 부분의 교집합을 말한다. 따라서 물체의 크기를 감소시키며, 외부 돌출부는 감소하고, 내부 돌출부는 증가하는 결과를 가져오게 된다. 이러한 침식연산과 팽창연산을 거쳐 구조요소(여기서는 피부 주름) 이하의 객 체나 잡음을 제거하는 것을 열기 연산이라고 한다. 본 발명은 이러한 열기 연산을 이용해 가장 의미 있는, 강한 주름 에지만을 검출하여 이를 비교하는 작업을 수행한다. 도 8은 본 발명에 따라 열기연산을 수행하기 전과 열기연산을 수행한 후의 결과를 보여준다. Anti erosion

It is defined as the intersection of the parallelized parts as opposed to the expansion operation. Therefore, the size of the object is reduced, the outer protrusion is reduced, and the inner protrusion is increased. Through this erosion and expansion operation, removing objects or noise below structural elements (skin wrinkles here) is called thermal operation. The present invention uses this open operation to detect and compare only the most significant, strong crease edges. Figure 8 shows the results before and after performing the thermal operation in accordance with the present invention.

The slopes of the respective edges detected in the two 2D skin wrinkle images (left image and right image) are then calculated and compared, so that the edges with the most approximate slope are the respective registration points. Such a method of searching for matching points using edges can work more efficiently than conventional methods using corners. 9 shows a matching operation using gradient comparison. The match points thus detected are used in the next step, the F mattress calculation step.

Subsequently, the F matrix computing step S220 is a step of calculating a fundamental matrix using an 8-point algorithm at the detected matching point.

To do this, first multiply the detected edges of two skin wrinkle images to extract only the areas where the two edges coincide. The present invention uses both end points of each edge as a matching point in the detected edge image. The purpose of this work is to detect a common edge part that is 1 by multiplying a binary image consisting of only 0 and 1 values. 10 shows an example of a product operation used in the present invention. FIG. 11 shows a result of matching points of both end points of an edge in the resultant image obtained by multiplying two edge images.

The F matrix, which is a 3 x 3 matrix, can be determined given eight matching points. If m´ is the point in the second image that corresponds to a point m in the first image, it can be expressed as

의 공식을 통해 에프 매트릭스(F Matrix)를 구할 수 있다. 이를 정리해서 두 점과 f는 아래와 같이 나타낼 수 있다. And the following

F matrix can be found by In summary, two points and f can be expressed as

The last term, 1, can be omitted so that it is summarized as follows.

A is a matrix created using a pair of eight matching points. At this time, the present invention used an edge rather than a corner. Therefore, the matching point (matching point) can be found more effectively in the skin image that is difficult to detect feature points.

Subsequently, the image rectification step S230 is an operation of warping the image plane using the F matrix value obtained above.

Warping refers to a technique for modifying a shape of a size, a length, and a thickness of an input image according to a specific rule. Used to make the distorted image look correct or to stretch or shrink one part. That is, image rectification is to faithfully represent the original image by correcting distorted or degraded image data, and from the stereo image, the depth of the image is set from the stereo image (Setreo Image). Information for obtaining a dense depth map is effectively applied to increase accuracy and reduce time in obtaining a binocular disparity map. This work generally involves working to correct geometric distortions.

The present invention requires an operation for arranging the registration points of two skin wrinkle images to be connected in a straight line. Preferably, it is necessary to align the epipolar line (Linear Epipolar Line). 12 shows an example of Epipolar geometry.

As shown, the epipolar geometry, which describes the structure of a point in a three-dimensional space in a two-dimensional image through a camera, shows a point in three-dimensional space X, a point projected on the two-dimensional image plane x, ΔMC ₁ C ₂ is called the epiplkar plan, and any point of the left image is given when the light source of the stereo camera is C ₁ , C ₂ , and the left and right input images are L and R, respectively. The straight line on which x 'corresponds to x is called the epipolar line for x and x'. E and e ', the intersection of the straight line between C ₁ and C ₂ and the epipolar line, are called epipoles.

In stereo matching, epipolar geometry exists that explains the relationship between the point in the real three-dimensional space and the point in the two-dimensional plane formed by the camera, in particular the relationship between the corresponding points of the two images. If the matching point and the matching point matching are obtained correctly, the geometric structure between the two images can be calculated.

The basic elements that make up the epipolar geometry result in the F matrix calculated earlier. The F matrix is a matrix that contains all the geometric information between two images obtained through the camera, and forms the basis of epipolar geometry.

의 관계가 형성된다. 이를 에피폴라 등식이라고 한다. 이 등식을 이용하여 좌 영상에 속한 임의의 점에 대응되는 우 영상의 한 점이 속해 있는 에피폴라 라인을 구할 수 있다. In the present invention, the 8-point algorithm is applied using the matching result of the feature points to obtain the F matrix. In FIG. 13, when a point in the three-dimensional image is a point x of the left image with respect to X and a point of the right image with x ', between the two points and the F matrix

Relationship is formed. This is called epipolar equation. Using this equation, the epipolar line to which one point of the right image corresponds to an arbitrary point belonging to the left image can be obtained.

Such image correction can be largely divided into four types as follows. First, the line segment PQ generated at the intersection of II and II ', which is a 2D original retinal plane, is calculated. Second, the vector CC 'which is a baseline is calculated. Third, the plane including the line segment PQ and parallel to CC 'is calculated. Fourth, the plane values are estimated to match the matching points in parallel. 14 shows a result of image correction of a skin wrinkle image.

Subsequently, the dense matching step S230 is a step of disparity mapping, which is important in 3D stereoscopic image (3D Reconstucton). After the Image Rectification operation, the coordinates are calculated for all the pixels. By calculating the registration point coordinates of the left and right images, we can see that the values of the x-axis coordinates are different. Computation of this difference is called disparity computation

Stereo image processing methods for obtaining disparity vectors are classified into two types. There are a correlation based method in which similarity measurement is performed on a block basis for all pixels, and a feature based method in which matching is performed only on a set of extracted feature points.

Herein, the present invention performs disparity mapping by using a feature-based method using feature points using a scale invariant feature transform (SIFT) algorithm. The SIFT algorithm is able to extract features that are invariant in size and rotation, and use them in search or recognition. In addition, SIFT generates a Gaussian image pyramid from the input image and extracts the local maximum and minimum points by Dog (Difference of Caussian). The maximum slope is extracted from the slope histogram of each feature point and quantized at eight angles around the feature point to generate a 128-dimensional feature vector using the directional information around the feature point as a 4 × 4 azimuth histogram.

In addition, the error of false matching that occurs during the matching process is eliminated and the distortion is corrected. Traditional algorithms required the user to define the disparity range manually and adjust the range until the result was calculated in order to calculate the disparity.

However, in the present invention, as shown in FIG. 15, the minimum, maximum matching on the x-axis (left, right) and the minimum and maximum matching on the y-axis (top, bottom) based on the left image among the SIFT matching pairs in the left and right images. Pairs can be determined, with the margin of error defined as α. The following formula is used to automatically specify the scan range so that the area to be matched with is calculated as a rectangle (left, right, top, bottom). In other words, the scan range is automatically specified so that the area to be matched with is calculated as a rectangle (left, right, top, bottom).

Then, Euclidean distance by SIFT vector is calculated using the following equations. 16 shows an example of Euclidean distance measurement by vector.

Setting of threshold

Distance calculation

As a result, when the coordinates and distances for each pixel of the two images are calculated and matched, disparity calculation is completed. 17 shows the result of parallax calculation. As shown, when the registration point coordinates of the left image (x, y) and the right image (x, y) is calculated, it can be seen that the values of the x-axis coordinates are different (350, 375). This is because the positions of the two cameras are parallel to the left and right sides when shooting. The present invention works through the skin wrinkle image in order to increase the accuracy by applying this automatic region setting method. This work result is shown in FIG.

The dense disparity map is generated using the calculated disparity value. In other words, it can be said that this difference is made into an image.

Subsequently, the 3D reconstruction step S250 is a process of creating a 3D model by performing triangulation. Triangulation is an operation of calculating the actual z coordinate value through coordinate values obtained in the previous step, dense matching. The definition of triangulation is as follows. The coordinates and distances of a point are determined using the properties of a triangle.If a point and two reference points are given, the angles between the base and the two other sides of the point and the two reference points are measured, respectively. After measuring the length, a series of calculations are performed using the sine law to find the coordinates and distance to the point. This distance value is the z value.

The calculation process will be described with reference to the figure shown in FIG. First, find the position of C using the law of sines and cosine.

Then, AC and BC are obtained as follows.

Then find RC as

Finally, after calculating MR, MC is calculated as follows.

Triangulation is performed in this manner, and the z coordinate value is calculated. The reason for doing this is to calculate the actual distance (depth information) between the two cameras and the object (skin wrinkles).

Thus, all coordinate values of x, y, and z were calculated. Then we work on the mesh to finalize the 3D reconstruction. The result is shown in FIG.

Finally, the 3D Wrinkle Data Estimation step (S300) is a task of comparing the 3D skin wrinkle data by parking to determine the degree of improvement of the wrinkles. The length, depth, and volume of the wrinkles in the area are measured to assess how much improvement has been achieved after the use of cosmetics or drugs.

This can be easily calculated using the x, y, z information, that is, length and depth information, after the dense matching and triangulation operations previously calculated. 21 shows the results of measuring 3D skin wrinkle data.

As a result of the measurement of the two 3D skin wrinkle data, the improvement of skin wrinkles by two parkings can be confirmed quantitatively.

1 is a block diagram showing a low-cost 3D measurement system for skin wrinkle evaluation according to the present invention,

Figure 2 is a photograph showing a three-dimensional image recording apparatus used in the present invention,

3 is a photograph showing how software is being displayed to allow two stereo images to be taken remotely;

Figure 4 is a schematic flow diagram showing a 3D measurement method for skin wrinkle evaluation according to the present invention,

5 is a photograph showing a stereo image acquisition method according to the present invention;

6 is a photograph showing a stereo image obtained according to the present invention;

7 is a photo showing before and after detecting edges using a Canny mask,

8 is a photograph showing a comparison between before and after the open operation operation,

9 is a photo showing a matching operation using the comparison of the slope of the edges,

10 is a formula showing an example of a multiplication operation used in the present invention,

11 is a photograph showing the selection of both ends of an edge as a registration point in the resultant image obtained by multiplying two edge images;

12 shows an example of epipolar geometry,

13 is a diagram for explaining image correction;

14 is a photograph showing the result of image correction;

15 is a photo showing an automatic region setting method using a SIFT matching pair,

16 shows an example of Euclidean distance measurement by vector,

17 is a photograph showing the result of parallax calculation,

18 is a photograph showing a dense matching result;

19 is a diagram for explaining a triangulation method,

20 is a photograph showing the results of 3D reconstruction,

21 is a photograph showing a comparison of 3D skin wrinkle image measurement results.

Claims (8)

delete delete delete delete Acquire Images (Acquire Images), which acquires two stereo images (2D) by two DSLR cameras installed in parallel at a predetermined interval, and match two stereo images (2D) obtained in the image acquisition step to obtain a stereoscopic image ( 3D image generation step for generating 3D) and 3D skin wrinkle image measurement step for measuring changes in skin wrinkles by comparing the length, depth or volume of wrinkles by measuring three-dimensional skin data obtained before and after using cosmetics or medicines In the stereoscopic image measurement method for the evaluation of the skin wrinkles, The 3D image generation step may include: a registration point search step of searching for registration points of two 2D wrinkle images in order to reconstruct the 2D skin wrinkle image photographed in the stereo image acquisition step; A f matrix calculation step of calculating a fundamental matrix using an eight-point algorithm at the detected matching point; An Image Rectification step of warping the image plane using the F matrix values obtained in the F matrix calculation step; After image rectification, dense matching is performed to calculate the coordinates of the matching points of the left image and the right image for all pixels, and to create a dense disparity map using the calculated disparity values. Matching step; Stereoscopic image reconstruction (3D Reconstruction) step of performing a triangulation (triangulation) using the 3D coordinate value obtained in the dense matching step to create a 3D skin wrinkle image (3D Reconstruction) comprising a three-dimensional for skin wrinkle evaluation Image measuring method. The method of claim 5, The registration point searching step is a three-dimensional image measurement method for skin wrinkle evaluation, characterized in that for using the edge (Edge) rather than the corner to find the matching point effectively. The method according to claim 5 or 6, The matching point searching step detects the edges of two skin wrinkle images using Canny Edge Detection, an edge detector, calculates the slopes of the detected edges, compares them, and has a maximum approximate slope. A three-dimensional image measuring method for skin wrinkle evaluation, characterized in that both ends of the edge as a registration point. The method of claim 7, wherein In the matching point searching step, the fine wrinkles that are not necessary to efficiently compare edges are removed using an opening operation that performs erosion and dilation, which are morphology operations. Stereoscopic measurement method for skin wrinkle evaluation.

Images