KR20060131354A - Colorimetric analysis of digital images obtained using digital camera - Google Patents

Abstract

A method for color-metrically analyzing a digital image in a digital camera is provided to objectively diagnose and observe the state of the skin. An LED(Light Emitting Diode) illumination device specially manufactured is used as the illumination when photographing. When performing the real photographing, the external light source is minimized and the illumination, a digital camera, the arrangement and distance of an object are constantly maintained all the time. The white balance of the digital camera is previously fixed and a flash is not used. The photographing is performed by the manual mode. The digital image obtained from the digital camera is stored in an sRGB color space.

Description

Colorimetric analysis of digital images obtained using digital camera}

The color system or color space is an essential element for measuring, understanding, and reproducing color, and much research has been conducted on the change of skin color using the color system or color space. Dermatologically, it is CIELAB, or CIE L * a * b *, which is relatively suitable in the color system, and creates perceptually uniform color space by linearizing the perception of color difference. Therefore, the distance between the two points in the color space corresponds to the difference of the color felt by the eyes, and can be expressed quantitatively. It is represented by three coordinates, L *, a *, and b *, where L * represents the axis of light or dark, a * represents the axis from red to green, and b * represents the axis from yellow to blue. In the past, a spectrophotometer has been used to measure color space coordinates of the skin, but the device is very expensive, and it is possible to measure it by touching the area, depending on the shape or location of the area to be measured. In many cases, it is difficult to measure, and although there is a great need for color measurement, it is difficult to be widely used in clinical practice.

Currently, digital cameras are widely used in clinical fields such as dermatology and replace film cameras. By using digital cameras that are widely distributed to most dermatologists without the need for expensive equipment, it is possible to perform dermatologically important colorimetric analysis to objectively assess the patient's skin condition and to easily obtain academically useful data. I hope to get it.

     In order to achieve the above task, the image of the color chart is obtained with a high resolution digital camera capable of a certain degree of closeness, but the color chart uses a color chart having dozens of color blocks. Each color segment of the color chart is measured by L *, a *, b * values using a spectrophotometer which is used as a standard method for measuring color, and is used as a reference value.

When shooting, the internal and external flashes should not be used, and tungsten, trichromatic fluorescent lamps, etc., which are well-reproducible and well-reproducible, may be used. Reproducible images are essential for the analysis of images, but in the case of flash photography, the results of the images are greatly influenced by the simple reflections occurring on the surface due to the characteristics of light, and there are many differences in colors even under the slight changes in the shooting conditions. give. In the case of taking dozens of the same close-up photo using the flash, the authors analyzed the characteristics of the flash photography, which was easily bright or dark, with insufficient reproducibility and consistency. In addition, dermatologically, flash photography obscures subsurface features due to the prominent reflection of the skin surface, making it difficult to identify various skin conditions such as pigmentation, inflammation, and vasodilation.

The actual shooting is as follows. To avoid being influenced by the light from the surrounding environment, minimize the external light source to shoot in a dark room, and keep the lighting, digital camera, subject placement, and distance at all times constant. The white balance automatically made by the digital camera is set to be arbitrarily set in a preset state, no exposure compensation is set, and the flash is not set to fire. Among other camera modes, the camera takes a manual or manual mode in which distortion by the processor inside the digital camera is minimized. When taking close-up shots of Korean skin in the same auto mode using the above lighting fixtures and digital cameras, the camera's settings, such as aperture, shutter speed, and sensitivity, are common.

Most digital images obtained from digital cameras are stored in the color space of sRGB. The conversion of sRGB coordinates to CIE L * a * b * is known as follows. That is, a floating-point non-linear sR'G'B '(8-bit integer sRGB) of 8-bit integers, ranging from 0 to 255, is given by the formula [1]. Switch to). The nonlinear sR'G'B 'is calculated as a linear RGB value using formula [2], and the RGB value can be converted to the tristimulus values of the CIEXYZ system, ie X, Y, and Z, using the following formula [3]: . Then convert to CIE L * a * b * coordinates using formula [4].

R ' _sRGB = R _8bit /255.0

G ' _sRGB = G _8bit /255.0 [1]

B ' _sRGB = B _8bit /255.0

R = R ' _sRGB /12.92

G = G ' _sRGB /12.92 If R' _sRGB , G ' _sRGB , B' _sRGB ≤0.04045

B = B ' _sRGB /12.92

[2]

R = ((R ' _sRGB +0.055) /1.055) ^2.4

G = ((G ' _sRGB +0.055) /1.055) ^2.4 If R' _sRGB , G ' _sRGB , B' _sRGB > 0.04045

B = ((B ' _sRGB +0.055) /1.055) ^2.4

[3]

[3]

L * = 116 (Y / Y _n ) ^1/3 -16, if (Y / Yn) ＞ 0.01

L * = 903 (Y / Y _n ), otherweise [4]

a * = 500 [(X / X _n ) ¹ / ^3- (Y / Y _n ) ^1/3 ]

b * = 200 [(Y / Y _n ) ¹ / ^3- (Z / Z _n ) ^1/3 ]

X _n , Y _n and Z _n : X, Y and Z tristumulus of reference white

However, as described above, the CIE L * a * b * value calculated in the sRGB of the image is different from the actual CIE L * a * b * value, so the error can be corrected to be closer to the actual CIE L * a * b * value. It is necessary to have a process.

First, photograph each and every section of the color chart in the manner and environment described above. The L *, a *, b * reference values of each color segment measured by the spectrophotometer, and the L *, a *, b * results calculated numerically using the above formulas in the digital image taken are statistically shown below. Compare with. In order to check the correlation between the two groups of data, Pearson correlation coefficients were obtained and correlated to confirm the linear relationship, and then the regression analysis using the simple linear regression model can be used to explain the following relationship. Find the formula, or estimated estimated regression line. In this case, it is possible to determine how high the explanatory power of the estimated session line is through the coefficient of determination (R ² ).

[5]

[5]

: estimated value of dependent variable,

: explanatory variable

estimated value of dependent variable,

explanatory variable

,

: least squares estimator of regression coefficien

,

least squares estimator of regression coefficien

According to the authors' results, when the digital images were obtained and analyzed as above, the reference values and the calculated values were linearly related to L *, a *, and b *, and the estimated explanatory line with high explanatory power was obtained. This allowed us to calculate the actual CIE L * a * b * values for specific sites. In other words, digital images of the skin taken with specific reproducible lighting conditions, digital cameras, camera settings, shooting distances, and methods can be used to derive CIE L * a * b * practically using the above formulas [1] to [5]. It can be calculated as a value very close to the value. It does not require any special expensive equipment, so it can be put directly into the clinical field and used for skin condition measurement and follow-up.

Unlike expensive spectrophotometers that are used for colorimetric analysis, they require less cost, do not have to touch the area to be measured, and can select and analyze only specific areas of interest. The site can also be easily measured. In addition to providing objective and analytical numerical data, a clinical picture of a patient that can be evaluated by the clinician's naked eye is obtained at the same time so that it can be used for later data validation or clinical evaluation.

Claims (2)

Digital images of multiple color plots of a color chart under reproducible, consistent lighting are obtained with sRGB files, the CIE L * a * b * coordinates are calculated for each color plot, and then a simple linear regression model, which is a statistical method, is generated. The analysis of the sessions was performed to obtain a formula that can be calibrated close to the actual CIE L * a * b * coordinates.        Using the above shooting method and statistically obtained formula, the skin color can be measured and described as an objective color space figure.