JP4305523B2 - Hair image display method and display device - Google Patents

Description

  The present invention relates to a hair image display method and a display device that display the styling state of hair in an easy-to-understand manner.

  In hair styling, there are problems such as whether the extent of spread, flow, and curl is intended, and whether there is hair deviating from the flow (so-called splash hair). Accordingly, various styling agents, hair straighteners, permanent agents and the like have been developed.

  However, although measuring devices for physical properties such as the thickness, tensile strength, moisture content, etc. of individual hairs and observation devices (Patent Document 1) for analyzing images such as cuticles, cortices, and medulla have been developed, There is no device that displays the overall styling status in an easy-to-understand manner and objectively evaluates it. For this reason, evaluation of styling agents and the like is performed exclusively by actually using them and performing sensory evaluation on the presence or absence of hairstyle collapse (Patent Document 2).

JP 2000-287953 A Japanese Patent Laid-Open No. 11-92346

  In contrast to the above-described prior art, an object of the present invention is to display hair in a direction that deviates from the flow in styling of hair in an easy-to-understand manner so that styling can be easily evaluated.

  The present inventor performs specific image processing on the hair styling image, and displays the hair deviated from the flow of the hair in a different color from the hair along the flow according to the degree of desorption. The present invention has been completed by discovering that it is possible to easily recognize the degree of tension and to easily evaluate the success or failure of styling.

  That is, the present invention selects a hair region from a hair image, performs edge detection on the hair region to calculate an edge image, performs a representative process on the edge image to calculate a representative value image, the edge image and the representative value image A hair image display method for displaying a direction difference image by calculating a color difference or a gray image of the direction difference of each pixel is calculated.

Further, the present invention provides a hair image acquisition means as an apparatus for carrying out the above-described hair image display method,
Means for selecting a hair area from a hair image, means for performing edge detection on the hair area and calculating an edge image, means for performing a representative process on the edge image and calculating a representative value image, a direction difference between the edge image and the representative value image There is provided a hair image display device comprising: a means for calculating the direction difference; a means for calculating the direction difference image by converting the direction difference of each pixel into a color image or a gray image; and a display means for displaying the direction difference image.

  According to the hair image display method of the present invention, the hair in the direction deviating from the flow in the hair styling is displayed in a color or darkness different from the hair along the flow according to the degree of the detachment by the direction difference image. Therefore, it is possible to easily recognize the unity of the entire styling, and it is also possible to easily add up the degree and distribution of the hair that has deviated from the flow. Therefore, it is possible to evaluate hair styling, and it is also useful for evaluating hair cosmetics used for styling and styling methods.

  Moreover, according to the hair image display device of the present invention, the hair image display method of the present invention can be easily performed.

  Hereinafter, the present invention will be described in detail with reference to the drawings. In addition, in each figure, the same code | symbol represents the same or equivalent component.

  FIG. 1 is a flowchart of a hair image display method according to an embodiment of the present invention, and FIG. 2 is a block diagram of a hair image display device 1 that performs this hair image display method.

  The hair image display device 1 includes a digital camera 2 that obtains a hair image of a hair style to be evaluated, a personal computer 3 that stores a hair image taken by the digital camera 2, and performs arithmetic processing. It has a display 4 and a printer 5 for displaying the previous hair image (original image) and calculation results.

  As a hair image acquisition means, in addition to the digital camera 2, an image scanner 6 that captures a hair image from a photograph of the subject's hair style may be provided, and the hair image is read from an arbitrary recording medium on which the hair image is recorded. A drive may be provided, and a hair image may be captured from a communication line such as the Internet. Further, the personal computer 3 may be connected with a hair cosmetic such as a styling agent and a database 7 in which hair styling states before and after application thereof are accumulated.

The personal computer 3 includes a program that functions as the following means (a) to (e).
(a) means for selecting a hair region from a hair image;
(b) means for performing edge detection on the selected hair region and calculating an edge image;
(c) means for performing a representative process on the edge image to calculate a representative value image;
(d) means for calculating a direction difference between the edge image and the representative value image;
(e) Means for calculating a direction difference image by converting a direction difference into a color image or a gray image

  Here, the hair image of (a) means a face image including a hair region, a head image, etc., and selection of the hair region from the hair image is, for example, in the case of black hair, the brightness is higher than a specific threshold value. A dark pixel is used as a hair region, and in the case of brown hair, blond hair, etc., a pixel whose saturation is higher than a specific threshold is used as a hair region. Moreover, you may select a hair area | region combining the conditions that edge strength is larger than a specific threshold value as needed. In addition, when selecting the hair region, it is preferable to set the hair region in advance so that the hair region is located at a predetermined part on the hair image by taking a hair image by a predetermined method.

The edge image calculated by edge detection in (b) corresponds to an image showing the flow of hair (including hair deviated from the flow). The edge detection itself can be performed by a known differentiation process using the density gradient of the image. More specifically, an edge extraction filter such as Prewitt, Robinson, Kirsch or the like can be used.

  For example, in the method using the Prewitt filter, the density difference dx in the horizontal direction is

The density difference dy in the vertical direction is

Is calculated using The edge direction θ corresponding to the direction of hair flow is

The edge strength r is calculated by

Is calculated by

  More specifically, by performing edge detection using an edge extraction filter, the horizontal edge image of FIG. 3B and the vertical edge of FIG. An image is obtained, and the edge direction is calculated from the horizontal edge image and the vertical edge image by the above equation (3) to form the edge direction image of FIG. Further, the edge strength is calculated, and the edge strength image of FIG.

  The representative processing of (c) is generally used when noise is removed by image processing, etc., but in the present invention, in order to remove the influence of hair deviating from the flow from the edge image showing the hair flow described above. Do.

  The representative process can be performed using a median filter, an average filter, a pruning average filter, or the like, but in the present invention, it is preferable to use a median filter. That is, the representative process using the median filter is a process of replacing the pixel value of a certain pixel with the median of the pixel values of the surrounding window area, and the process using the average filter is a process of replacing the pixel value of a certain pixel. This is a process of replacing the average value of the pixel values in the surrounding window area. Therefore, for example, as shown in FIG. 4A, there are a large number of pixels in the θ1 direction and one pixel in the θ2 direction in the surrounding window region, and the edge intensity of the pixels in the θ2 direction is the pixel in the θ1 direction. When representative processing is performed using an average value filter when it is extremely strong compared to edge strength, the edge direction θav after representative processing is the direction of a large number of pixels as shown in FIG. The edge direction θav after the representative treatment does not correspond to the hair flow. On the other hand, when the median filter is used, as shown in FIG. 4C, the edge direction θo after the representation corresponds to the hair flow. The average value here was calculated by obtaining the principal axis of the edge included in the surrounding window region. The median value was calculated by obtaining an edge that minimizes the sum of distances to other edges included in the surrounding window region.

  When performing the representative process, if the size of the filter (peripheral window area) is too small, the edge direction after the representative process becomes difficult to correspond to the hair flow. Will contribute to the edge direction. For example, taking the median value of the range of the neighborhood radius of 2 pixels with respect to the edge images (b) and (c) of FIG. 3 described above, the representative value images of FIGS. 5 (a) and 5 (b) are obtained. Taking the median of the range of the neighborhood radius of 5 pixels, the representative value images shown in FIGS. 5C and 5D are obtained. The size of the preferred filter varies depending on the hair style. In the case of a curled hair style with respect to a straight hair style, the edge direction after the representative treatment cannot easily correspond to the hair flow unless the filter is made small. More specifically, the size of the filter is preferably set so that the radius of the surrounding window region is not less than three times the hair diameter in the case of the straight hair style, and further, in the case of the curled style, the size of the surrounding window is set. It is preferable that the radius of the region is 0.5 times or less the curl radius of curvature.

  The direction difference between the edge image and the representative value image in (d) corresponds to the degree of hair detachment from the hair flow. As this direction difference, a Mahalanobis general distance, a Euclidean distance, or the like can be calculated. Among them, it is preferable to calculate the Mahalanobis general distance.

  Mahalanobis general distance is a measure of distance calculated taking into account correlations and variations of multidimensional variables. Specifically, the Mahalanobis general distance between the edge image and the representative value image can be obtained by the following equation.

(Where, v1: length of projection in the direction of the representative value of the edge for which the distance is obtained v2: length of projection in the direction orthogonal to the representative value σ1: direction of the representative value of the edge included in the surrounding window region Standard deviation of projection length to σ2: standard deviation of projection length in the direction perpendicular to the representative value)

  The Mahalanobis general distance means that the smaller the value is, the closer the object is to the population, that is, the hair of interest is along the flow of the hair. This means that the hair of interest is in a direction away from the hair flow.

  After calculating the Mahalanobis pan-distance, noise removal processing using a median filter, an average filter, or the like may be further performed on the direction difference image. This makes it possible to make the hair that is out of the flow of the hair stand out more.

  The process of converting the direction difference of (e) into a color image or grayscale image may be performed by, for example, associating the hue angle with the numerical value of the direction difference or the grayscale level. Thereby, the direction difference can be visualized.

  For example, the direction difference image of FIG. 6 can be obtained by converting the Mahalanobis general distance between the edge image of FIG. 3 and the representative value image of FIGS. 5A and 5B to a grayscale image with 256 gradations. Further, by performing noise removal processing using a median filter having a neighborhood radius of 2 pixels, a direction difference image (noise removal) in FIG. 7 can be obtained.

  In the present invention, in addition to displaying the direction difference as the direction difference image as described above, the direction difference and the appearance frequency thereof may be represented in a relational diagram such as a histogram as shown in FIG. This makes it easy to statistically analyze the degree of hair detachment, the frequency of appearance, and the like that deviate from the hair flow.

  Furthermore, this analysis result and the appearance of the above-described direction difference image may be associated with hair cosmetics and styling methods stored in the database 7.

  Taking a hair image (original image) of black hair with a digital camera (diffuse illumination, camera image sensor: about 4 million pixels), about the hair region P (256 × 256 pixels) shown in FIG. Edge detection is performed by the Prewitt filter, and a representative value image is calculated from the obtained edge image by a median filter (filter size: radius 20 pixels, equivalent to about 7 times the hair diameter). 9B is obtained by calculating the Mahalanobis general distance to the image, obtaining a grayscale image, obtaining a direction difference image, and further removing noise using a median filter (filter size: radius 1 pixel). A direction difference image was obtained. In this direction difference image, the hair extending in the left-right direction was able to stand out and displayed with respect to the hair flowing substantially from top to bottom.

  The hair image display method or display device of the present invention is useful for evaluation of styling based on the degree of hair detachment from the flow in hair styling, the distribution, etc., the hair cosmetics used for styling, and the styling method itself. In addition, it is useful for the development of hair cosmetics and styling methods, as well as advertising sales.

It is a flowchart of the hair image display method. It is a block diagram of a hair image display device. An original image and an edge image. It is explanatory drawing of the edge direction in a representative process. It is the smoothed image for every magnitude | size of a median filter. It is a direction difference image. It is a direction difference image after noise removal. It is a histogram of a direction difference and its appearance frequency. It is the original image and direction difference image in an Example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Hair image display apparatus 2 Digital camera 3 Personal computer 4 Display 5 Printer 6 Image scanner 7 Database

Claims (6)

Select the hair area from the hair image, perform edge detection on the hair area, calculate the edge image,
Hair that displays a direction difference image by performing a representative process on the edge image to calculate a representative value image, calculating a direction difference between the edge image and the representative value image, and converting the direction difference of each pixel into a color image or a gray image. Image display method.   The hair image display method according to claim 1, wherein a median filter is used for representative processing.   The hair image display method according to claim 1, wherein the Mahalanobis general distance is calculated as the direction difference.   Hair image acquisition means, means for selecting a hair area from the hair image, means for performing edge detection on the hair area and calculating the edge image, means for performing representative processing on the edge image and calculating a representative value image, edge image and representative value A hair image display device comprising: means for calculating a direction difference from an image; means for calculating a direction difference image by converting the direction difference of each pixel into a color image or a gray image; and a display means for displaying the direction difference image.   The hair image display device according to claim 4, wherein a median filter is used for the representative processing.   The hair image display device according to claim 4 or 5, wherein the Mahalanobis general distance is calculated as the direction difference.