KR20100104048A - Method and apparatus for classifying a sky image using histogram analysis - Google Patents

Abstract

PURPOSE: A method and a device for classifying a sky image using histogram analysis are provided to analyze the characteristic of a histogram image and classify an input image based on analysis result, thereby reducing the complexity and processing time of calculation for analyzing histogram characteristic and exactly classifying a sky image. CONSTITUTION: Converting an input RGB image into a luminance/color difference image and using the converted color difference image to remove a cloud component included in the input RGB image. Converting an image which removed the cloud component into a histogram image of gray level and analyzing the characteristic of converted histogram image. A YCC converting unit(110) converts an input RGB image into a luminance/color difference image. A cloud removing unit uses the converted color difference image and removes the cloud component, a cloud component included in the RGB image. A histogram analyzing unit(130) converts the image, an image that removed the cloud component, into a histogram image of a gray level and analyzes the characteristic of the converted histogram image.

Description

Method and apparatus for classifying a sky image using histogram analysis

The present invention relates to image processing, and more particularly, to a sky image classification method and apparatus using a histogram.

The image histogram shows the distribution of contrast values for the pixels in the image, representing the range and value of light and dark pixels when they are distributed. This graph is called a histogram graph, and the range of contrast values in the 256 gray level image ranges from 0 to 255, and the frequency of each intensity level is investigated and expressed as the height of the graph. Such a histogram has a lot of information of an image and is variously used for image processing. In addition, it is possible to analyze the histogram to determine the characteristics of the image.

Korean Patent Laid-Open Publication No. 2001-0055492 relates to an image retrieval system and method using an image histogram. The image retrieval is obtained by obtaining the center point and the variance information from the image as well as the quantity information of the colors of the bins of each histogram. Disclosed is a technique used for mapping information.

However, there is a problem in that an image histogram has a large amount of calculation for obtaining a variance value and the like, and a processing time is long because a feature vector is extracted.

Meanwhile, with the development of a digital photographing apparatus including a digital camera or a camcorder, a function of recognizing a scene to be photographed and setting appropriate photographing control values is added. In particular, by distinguishing the color of the sky, it can be recognized as a sunset mode, sunny sky mode, cloud mode, etc. can guide the shooting accordingly. In addition, by detecting only a part of the sky from the captured sky image, the saturation of the sky may be adjusted to correct the image with a clearer and clearer sky. To this end, sky detection or sky recognition is required.

The present invention is to solve the above-mentioned problems of the prior art, and by effectively removing the cloud component for the classification of the sky image, by analyzing the characteristics of the image histogram can simply classify whether the input image is a sky image, An object of the present invention is to provide a sky image classification method and apparatus.

In order to achieve the above technical problem, the sky image classification method according to an embodiment of the present invention converts an input RGB image into a luminance / color difference image, and uses the converted color difference image to include cloud components included in the input RGB image. Removing; Converting the image from which the cloud component is removed into a histogram image having a gray level, and analyzing characteristics of the converted histogram image; And classifying the input RGB image into a sky image based on the analysis result.

The characteristic of the histogram image may include at least one of a frequency, a number of peaks, a peak distance, and a peak width of the histogram image.

Preferably, the analyzing step further comprises the step of setting a frequency indicating the number of gray levels of the histogram image to a predetermined value,

The classifying step may include classifying the input RGB image into a sky image according to the presence or absence of peaks representing the set frequency or more.

Preferably, the analyzing step comprises setting a first threshold value for the width of the peak representing the number of gray levels passing the set frequency, and determining whether the width of the peak is less than or equal to the first threshold value. More,

In the classification step, when the width of the peak is less than or equal to the first threshold value, the input RGB image may be classified into a sky image.

The analyzing may further include normalizing the converted histogram image.

Preferably, the cloud removing step is characterized in that to remove the cloud component by using the Cb value and the Cr value in the color difference image.

According to another aspect of the present invention, there is provided a sky image classification apparatus, including: a YCC converter configured to convert an input RGB image into a luminance / color difference image; A cloud removing unit which removes cloud components included in the input RGB image by using the converted color difference image; A histogram analyzer for converting the image from which the cloud component is removed to a histogram image having a gray level, and analyzing characteristics of the converted histogram image; And an image classification unit classifying the input RGB image into a sky image based on the analysis result.

The characteristic of the histogram image may include at least one of a frequency, a number of peaks, a peak distance, and a peak width of the histogram image.

Preferably, the histogram analyzer further includes a frequency setting unit that sets a frequency indicating the number of gray levels of the histogram image to a predetermined value.

The image classifying unit may classify the input RGB image into a sky image according to the presence or absence of peaks indicating the set frequency or more.

Preferably, the histogram analyzer sets a first threshold value for the width of the peak representing the number of gray levels passing the set frequency, and determines whether the width of the peak is less than or equal to the first threshold value. Further comprising a width determining unit,

The image classifying unit may classify the input RGB image into an empty image when the width of the peak is less than or equal to the first threshold value.

Preferably, the histogram analyzer further comprises a normalizer for normalizing the converted histogram image.

Preferably, the cloud removing unit removes the cloud component by using the Cb value and the Cr value in the color difference image.

And a recording medium on which a program for executing the sky image classification method on a computer for recording the technical problem is achieved.

The sky image classification method according to an embodiment of the present invention removes the cloud component from the input image, analyzes the characteristics of the histogram image, and classifies the input image based on the analysis result, thereby calculating complexity for the histogram characteristic analysis. And processing time can be reduced, and sky image classification can be made correctly.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, only parts necessary for understanding the operation according to the present invention will be described, and descriptions of other parts may be omitted so as not to distract from the gist of the present invention.

In addition, terms and words used in the following description and claims should not be construed to be limited to ordinary or dictionary meanings, but are to be construed in a manner consistent with the technical idea of the present invention As well as the concept.

1 is a schematic block diagram of a sky image classification apparatus 100 according to an exemplary embodiment.

Referring to FIG. 1, the sky image classification apparatus 100 includes a YCC converter 110, a cloud remover 120, a histogram analyzer 130, and an image classifier 140.

The Y / C converter 110 converts the RGB input image into a luminance / chromatic color space and outputs the luminance (Y) and chroma signal (C). Here, the RGB input image is a sky image including a cloud component.

Here, the conversion from the RGB color space to the YCbCr color space uses the following equation (1).

Y = 0.29900R + 0.58700G + 0.11400B

Cb = -0.16874R-0.33126G + 0.50000B

Cr = 0.50000R-0.41869G-0.08131B

Here, the luminance Y is a component expressing the light and dark degree, and the color difference C is a component expressing the color information. The two color difference components have Cb and Cr values, where Cb is the difference between the blue (B) component and the reference value, and Cr is the difference between the red (R) component and the reference value. In addition, although the preferred embodiment of the present invention describes the conversion of an RGB image to a YCbCr signal, it is applicable not only to YCbCr conversion coordinates but also to other color space coordinate systems such as YUV, Lab, and YCC. Of course it can.

The cloud remover 120 removes the cloud component by using the Cb value and the Cr value in the color difference image output from the YCC converter 110. Therefore, in order to accurately detect the sky image, it is possible to effectively remove clouds separated from the sky components.

For each image, the cloud component is removed using the minimum and maximum values of Cb and Cr. That is, it is determined that the component within the range shown in Equation 2 is a cloud component, and an image from which the cloud component is removed from the original image is obtained.

Min <Cb <Max

Min <Cr <Max

Here, each of the minimum and maximum values is a value that can be arbitrarily determined and can be changed depending on the characteristics of the sky image, that is, how many cloud components are included.

The histogram analyzer 110 receives a sky image from which cloud components have been removed and analyzes characteristics of the histogram image. Here, the characteristic of the histogram image includes the frequency, number of peaks, peak distance, and peak width of the histogram image. In general, a histogram is a distribution of contrast values for pixels in an image. The histogram represents the range and value of light and dark pixels when they are distributed. The histogram is called a histogram graph. The range of the contrast value in the level image has a value of 0 to 255, and each contrast value, that is, the frequency of the level is represented by the height of the graph. The histogram has a lot of information of the image and is used for various image processing. Also, in a preferred embodiment of the present invention, the histogram may be analyzed to determine characteristics of the image. The detailed configuration of the histogram analyzer 110 will be described later with reference to FIG. 2.

The image classifier 140 classifies the input image into the sky image based on the analysis result of the histogram analyzer 130. The image classifying unit 140 classifies whether the input image is a sky image according to the characteristics of the histogram image described above. That is, the input image is classified with information about the number of peaks, the width of the peak, and the change of the peak of the histogram image.

2 is a schematic block diagram of the histogram analyzer 130 illustrated in FIG. 1.

Referring to FIG. 2, the histogram analysis unit 130 may include a normalization unit 131, a frequency setting unit 132, a peak number determination unit 133, a peak distance determination unit 134, and a peak width determination unit 135. It is configured to include.

The normalizer 131 normalizes the histogram image. Here, normalization is used to remap these values within a desired range when the number of pixels (frequency) corresponding to the Y value transformed in the histogram, that is, the corresponding contrast value in the image, that is, the gray level value, exceeds a certain range value. Means to readjust.

The frequency setting unit 132 sets a frequency indicating the number of gray levels of the histogram image to a predetermined value. Here, the predetermined value is a value that can be arbitrarily set according to the type of the image, that is, the sky image, and the frequency means the number of pixels belonging to the gray level.

6 is a diagram for describing a histogram analysis method according to another exemplary embodiment.

Referring to FIG. 6, the histogram image of FIG. 3D is illustrated. The X axis is gray level and the Y axis is normalized frequency. Here, the frequency is set to 160. The frequency setting unit 132 may set the frequency differently according to the type of images to be classified.

The peak number determination unit 133 determines the number of peaks representing the frequency or more set by the frequency setting unit 132. Referring again to FIG. 6, the number between peaks of frequency 100 or more is zero.

The peak distance determination unit 134 sets a first threshold value for the distance value between the peaks when the number between the peaks is 1 or more, and determines whether the distance value is greater than or equal to the first threshold value. Referring back to FIG. 6, since the number between peaks is zero, the distance value between the peaks is not determined.

The peak width determining unit 135 sets a second threshold value for the width of the peak indicating the number of gray levels passing the frequency set by the frequency setting unit 131, and determines whether the width of the peak is less than or equal to the second threshold value. To judge. Referring back to FIG. 6, the width of the peak passing over frequency 160 is 10. Then, it is determined whether the width 10 of the peak is less than or equal to the second threshold value. Here, the second threshold value is a value that can be arbitrarily set.

The image classifying unit 140 includes the presence or absence of peaks indicating the frequency or more set by the frequency setting unit 132, the number of peaks determined by the peak number determination unit 133, and the distance value determined by the peak distance determination unit 134. The input image may be classified as an empty image according to whether the sum of the peak widths determined by the peak width determination unit 135 is equal to or greater than the first threshold value or less.

3 to 5 illustrate histograms of various sky images.

3A is an RGB image of a sky image including a cloud, FIG. 3B is an image from which sky components are removed using a Cb value and a Cr value, and FIG. 3C is a gray image of the image of FIG. 3B. 3D is a histogram image in which the image of FIG. 3C is displayed as a histogram image having 256 levels. Here, the histogram image is a histogram image obtained by normalizing a frequency, that is, the number of pixels of a specific level. In FIG. 3D, the frequencies shown in gray levels 0 to 5 are black parts from which the cloud component is removed in FIG. 3B, and thus are excluded from the histogram analysis.

4A is an RGB image of a sky image including a cloud, and the distribution of the cloud is different from that of FIG. 3A. FIG. 4B is an image from which sky components are removed using a Cb value and a Cr value, and FIG. 4C is a gray image of the image of FIG. 4B. 4D is a histogram image in which the image of FIG. 4C is displayed as a histogram image having 256 levels. Here, the histogram image is a histogram image obtained by normalizing a frequency, that is, the number of pixels of a specific level. In FIG. 4D, the frequencies shown in the gray levels 0 to 5 are black parts from which the cloud component is removed in FIG. 4B, and thus are excluded from the histogram analysis.

FIG. 5A is an RGB image of a sky image including clouds, FIG. 5B is an image obtained by removing sky components using Cb and Cr values, and FIG. 5C is a gray image of the image of FIG. 5B. 5D is a histogram image in which the image of FIG. 5C is displayed as a histogram image having 256 levels. Here, the histogram image is a histogram image obtained by normalizing a frequency, that is, the number of pixels of a specific level. In FIG. 5D, the frequencies shown in the gray levels 0 to 5 are black parts from which the cloud component is removed in FIG. 5B, and thus are excluded from the histogram analysis.

6 to 8 are diagrams for describing histogram analysis of sky images illustrated in FIGS. 3 to 5, respectively.

The basic conditions for analyzing the histograms of the various sky images shown in FIGS. 3D, 4D, and 5D, that is, the frequency, the presence or absence of peaks, the number of peaks, the threshold for distance values between peaks, and the set frequency The threshold for the width of the peak is as follows. The frequency is set to 160, the number between peaks is 0, and the threshold for the distance value between the peaks is 30.

Referring back to FIG. 6, in the histogram image illustrated in FIG. 3D, a peak value exists between gray levels 30 to 60, and one peak passes through a frequency 160. Therefore, since the number between peaks is 0, the distance value between peaks is not determined. Since the number of gray levels of the histogram passing through the frequency 160 is 10, the number of gray levels is satisfied. Therefore, the histogram image may be classified as a sky image.

Referring back to FIG. 7, in the histogram image illustrated in FIG. 4D, a peak value exists between gray levels 30 to 60, and one peak passes through a frequency 160. Therefore, since the number between peaks is 0, the distance value between peaks is not determined. Since the number of gray levels of the histogram passing through the frequency 160 is 14, the condition below 30 is satisfied, and therefore, the histogram image may be classified as a sky image.

Referring back to FIG. 8, in the histogram image illustrated in FIG. 5D, a peak value exists between gray levels 30 to 60, and one peak passes through a frequency 160. Therefore, since the number between peaks is 0, the distance value between peaks is not determined. Since the number of gray levels of the histogram passing through the frequency 160 is 22, it satisfies a condition of 30 or less, and thus the histogram image may be classified as a sky image.

9 is a flowchart illustrating an image classification method according to another exemplary embodiment.

9, in operation 900, an RGB sky image is input.

In step 902, the RGB image is YCC-converted to calculate Cb and Cr values. In step 904, the cloud component included in the sky image is removed using Cb and Cr values. Therefore, the accuracy of sky image detection can be improved by removing unnecessary clouds to analyze the sky image.

In step 906, the histogram image is normalized to have a frequency within a certain range. In step 908, a specific frequency of the histogram is set. Here, the frequency may be set differently according to the type of the sky image, for example, the degree of inclusion of the cloud component.

In step 910, the number of peaks having peaks above a certain frequency is determined. In step 912, when the number between peaks is 1 or more, a specific threshold value for the distance between the corresponding peaks is set, and it is determined whether the distance value is within the range of the threshold value.

In step 914, it is determined whether the sum of the peak width or the width of the peaks passing the frequency set in step 908 is equal to or less than a specific threshold value.

In operation 916, the input image is classified into a sky image based on the determination result of the number of peaks, the peak distance, and the peak width determined in the operations 910 to 914.

The present invention can be applied to various image processing devices such as digital cameras, mobile phones equipped with cameras, digital camcorders, and the like.

Meanwhile, the present invention can be embodied as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored.

Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disks, optical data storage devices, and the like, which may also be implemented in the form of carrier waves (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion. In addition, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers skilled in the art to which the present invention belongs.

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will understand that the present invention can be embodied in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown not in the above description but in the claims, and all differences within the scope should be construed as being included in the present invention.

1 is a schematic block diagram of a sky image classification apparatus 100 according to an exemplary embodiment.

2 is a schematic block diagram of the histogram analyzer 130 illustrated in FIG. 1.

3 to 5 illustrate histograms of various sky images.

6 to 8 are diagrams for describing histogram analysis of sky images illustrated in FIGS. 3 to 5, respectively.

9 is a flowchart illustrating a sky image classification method according to another exemplary embodiment.

<Explanation of symbols for main parts of the drawings>

100: sky video classification device 110: YCC conversion unit

120: cloud removing unit 130: histogram analysis unit

140: image classification unit 131: normalization unit

132: frequency setting unit 133: number of peak determination unit

134: peak distance determination unit 135: peak width determination unit

Claims (13)

Converting an input RGB image into a luminance / color difference image and removing cloud components included in the input RGB image using the converted color difference image; Converting the image from which the cloud component is removed into a histogram image having a gray level, and analyzing characteristics of the converted histogram image; And And classifying the input RGB image into a sky image based on the analysis result. The method of claim 1, The characteristic of the histogram image, Sky image classification method using a histogram analysis, characterized in that it comprises at least one or more of the frequency, number of peaks, peak distance, and peak width of the histogram image. The method of claim 2, The analyzing step, The method may further include setting a frequency indicating a number of gray levels of the histogram image to a predetermined value. The classification step, The sky image classification method using a histogram analysis, characterized in that the input RGB image is classified into a sky image according to the presence or absence of peaks indicating the frequency above the set frequency. The method of claim 3, wherein The analyzing step, Setting a first threshold value for the width of the peak representing the number of gray levels passing the set frequency, and determining whether the width of the peak is less than or equal to the first threshold value, The classification step, The sky image classification method using histogram analysis, wherein the input RGB image is classified into a sky image when the width of the peak is less than or equal to the first threshold value. The method of claim 2, The analyzing step, Normalizing the transformed histogram image sky image classification method using histogram analysis, characterized in that it further comprises. The method of claim 2, The sky image classification method using the histogram analysis, characterized in that the cloud component is removed using the Cb value and the Cr value in the chrominance image. A recording medium having recorded thereon a program for executing a method according to any one of claims 1 to 6 on a computer. A YCC converting unit converting an input RGB image into a luminance / color difference image; A cloud removing unit which removes cloud components included in the input RGB image by using the converted color difference image; A histogram analyzer for converting the image from which the cloud component is removed to a histogram image having a gray level, and analyzing characteristics of the converted histogram image; And The sky image classification apparatus using a histogram analysis including an image classifier which classifies the input RGB image into a sky image based on the analysis result. The method of claim 8, The characteristic of the histogram image, Sky image classification apparatus using a histogram analysis, characterized in that at least one of the frequency, number of peaks, peak distance, and peak width of the histogram image. The method of claim 9, The histogram analysis unit, And a frequency setting unit for setting a frequency indicating a number of gray levels of the histogram image to a predetermined value. The image classification is, The sky image classification apparatus using a histogram analysis, characterized in that the input RGB image is classified into a sky image in accordance with the presence or absence of peaks indicating the set frequency or more. The method of claim 10, The histogram analysis unit, And a peak width determining unit that sets a first threshold value for the width of the peak representing the number of gray levels passing the set frequency, and determines whether the width of the peak is less than or equal to the first threshold value. The image classification unit, The sky image classification apparatus using histogram analysis, wherein the input RGB image is classified into a sky image when the width of the peak is less than or equal to the first threshold value. The method of claim 9, The histogram analysis unit, The sky image classification apparatus using the histogram analysis, further comprising a normalization unit for normalizing the converted histogram image. The method of claim 9, The cloud removing unit, The sky image classification apparatus using histogram analysis, characterized in that the cloud component is removed by using the Cb value and the Cr value in the chrominance image.

Images