KR101564394B1 - Method and recording medium for day and night recognition of input image - Google Patents

Abstract

The present invention relates to a daytime and nighttime recognition method of an input image and a recording medium thereof, which can accurately identify daytime and nighttime conditions in an image input from a single camera regardless of various roads and surrounding environments on which a vehicle runs.
The method of recognizing day and night of an input image according to the present invention and a recording medium thereof are a method of discriminating day and night of an input image using frequency of occurrence of pixels corresponding to brightness values within a predetermined brightness brightness value range to provide.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for recognizing a day and night of an input image,

The present invention relates to a daytime and nighttime recognition method of an input image and a recording medium thereof, and more particularly, to an image recognition system using an image sensor, The present invention relates to a day and night recognition method of an input image, and a recording medium thereof, which enable judging whether or not an object is recognized, thereby providing judgment information for object recognition more accurately.

A camera-based driving assistance system, such as a Forward Collision Warning system and a Lane Departure Warning system, operates based on the image analysis obtained from the camera and displays the lane and vehicle appearance It is necessary to develop a system capable of recognizing the daytime and nighttime automatically in order to adaptively recognize the object in such a situation.

A conventional method that can be used to recognize daytime and nighttime conditions in an image is to analyze statistical average brightness information for an upper sky region of an image using Equation (10).

&Quot; (10) "

The upper sky area of the image has a fairly high brightness level in the daytime state and a very low brightness level in the nighttime state. Therefore, when the average brightness value m is equal to or greater than a predetermined threshold value, the daytime and nighttime states can be easily recognized by judging to be the daytime and when the average brightness value m is less than the predetermined threshold value, the nighttime.

However, actual daytime and nighttime road driving environments do not necessarily represent ideal scenes, and various patterns of lightness and brightness appearing in the upper sky region may vary depending on various roads and surrounding environments on which the vehicle travels.

For example, when the upper sky area is dark due to the overpass during the interstate highway, when the road is running in the city during the daytime, the upper sky area is dark due to the roadside trees, when the interstate highway is entered, Since the signboard is bright when traveling on the road, when the upper sky area is relatively bright, and when the traffic light and the street light are blurred during the nighttime driving on the country road, the upper sky area is relatively bright. There is a limitation that it is impossible to accurately recognize the daytime state and the nighttime state by using the average brightness information of Equation (10).

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above-mentioned problems, and it is an object of the present invention to provide an image display apparatus and a display method thereof that can accurately identify daytime and nighttime conditions in an image input from a single camera, A day and night recognition method of an input image, and a recording medium therefor.

According to an aspect of the present invention, there is provided a method of recognizing day and night of an input image, the method including: a first step of receiving an image including a sky area and a terrestrial area; A second step of setting an area including at least the sky area in the input image as a first area of interest; Calculating a weekly rate of the first ROI by using the frequency of occurrence of pixels corresponding to the respective BRI values within the first brightness level value range, and calculating a weekly ratio of the pixels corresponding to the brightness values within the second brightness level range, A third step of calculating a nighttime rate of the first ROI using the frequency of occurrence of the first ROI; A fourth step of comparing the calculated weekly rate of the first ROI with a predetermined first threshold; If the daytime rate of the first area of interest is greater than the first threshold value as a result of the comparison in the fourth step, recognizes the input image as a daytime state, and if the ratio is smaller than the first threshold value, A fifth step of comparing the threshold value with a threshold value; And a sixth step of recognizing the input image as a night scene when the nighttime ratio of the first ROI is greater than the second threshold as a result of the comparison of the fifth step.

According to the day and night recognition method of the input image according to the present invention, various optional steps are adaptively applied in consideration of various road driving environments, so that not only general daytime and nighttime environments but also daytime and nighttime conditions are difficult to recognize It is now possible to recognize normal.

1 is a block flow diagram illustrating a process flow of a day and night recognition method of an input image according to the present invention.
2 is an example of a sky area and a ground area divided according to the present invention;
3 is an example for explaining an input image size reduction step using the average interpolation method of the present invention.
4 is an example showing a histogram (right lower side) for the daytime state (left side) and the first ROI (upper right side) recognized according to the present invention.
5 is an example showing a histogram (lower right) for the nighttime state (left side) and the first ROI (upper right side) recognized according to the present invention.
6 is an example showing a histogram of the first state of interest and the daytime state at the time of the rainy day running recognized according to the present invention.
FIG. 7 is an example showing a histogram of a first state of interest and a daytime state at the time of entering a tunnel recognized according to the present invention;
FIG. 8 is an example showing a histogram of a first state of interest and a daytime state at the time of appearance of a leg with a high leg recognized according to the present invention.
9 is an example showing a histogram (left side) of a daytime general road area and a histogram (right side) of a shadow road area.
10 is an example showing a histogram (left side) of a nighttime vehicle road region and a histogram (right side) of a non-vehicle road region.
11 is an example showing a histogram (upper right side) for the daytime state (left side) and the second attention area (lower right side) recognized according to the present invention.
12 is an example showing a histogram (upper right side) for the nighttime state (left side) and the second attention area (lower right side) recognized according to the present invention.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

Furthermore, although the terms first, second, etc. may be used herein to describe various components, the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

In the following, preferred embodiments, advantages and features of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block flow diagram illustrating a process flow of a day and night recognition method of an input image according to the present invention.

Referring to FIG. 1, a day and night recognition method of an input image according to the present invention includes an image input step (step 1), a step of setting a sky area as a first ROI (step 2) (Step 3) of calculating a daytime rate and a nighttime rate of the first ROI, comparing the weekly rate of the first ROI with the first threshold (step 4), and comparing the result of step 4 with the first threshold, (Step 5) of recognizing the input image as a daytime or comparing a nighttime ratio of the first ROI with a second threshold value, and a night recognition step according to the comparison result of step 5 (step 6) .

Step 2 of the present invention may further comprise setting the terrestrial region as a second ROI, wherein the step 6 calculates a difference value between the weekly rate and the nighttime rate of the first ROI according to the comparison result of Step 5 .

In the above case, the method of recognizing day and night of the input image according to the present invention may further comprise a step (step 7) of comparing the difference value of the sixth step with a predetermined third threshold value, Calculating the weekly and nighttime ratio of the region, or calculating the degree of the disorder of the first region of interest (Step 8).

(Step 8b) of comparing the calculated disorder and the fourth threshold value in step 8, and step 9b of judging the daytime recognition or the nighttime recognition according to the comparison result of step 8b, As shown in FIG.

If the weekly rate and the nighttime rate of the second area of interest are calculated in step 8, the step of comparing the weekly rate and the nighttime rate is performed (step 8a). In step 8a, ).

Further, preferably, between step 1 and step 2, a step of reducing the input image of step 1 (step 1-1) may be further included.

Hereinafter, preferred embodiments according to the respective steps will be described in more detail.

<Step 1>

Step 1 of the present invention is a step in which the camera installed in the vehicle of the present invention picks up the foreground of the front and receives the captured forward image. Therefore, the input image is displayed together with a ground area including a road, a building on the road or a car, and a sky area corresponding to the sky.

<Step 2>

Step 2 of the present invention is configured to divide the sky area and the ground area represented in the input image 10 and set the sky area as the first area of interest 11.

2 is an example of a sky area and a ground area divided according to step 2 of the present invention. Referring to FIG. 2, a sky area and a ground area are respectively extracted from the input image 10 using the ordinate values of the segment line 20 between the predefined sky area and the ground area of step 1, 1 region of interest (11) and a second region of interest (13), and then the image is divided.

Here, the dividing line 20 may be constituted by a longitudinal line having a corresponding ordinate value at a position equal to or lower than the missing line. Thus, only the sky area may be included in the first area of interest 11, or a part of the top area of the terrestrial area may be included.

That is, an area including at least the entire sky area in the input image is set as the first area of interest 11, and an area excluding the first area of interest 11 in the input image is set as the second area of interest 13 .

For reference, the disappearance line means a line corresponding to a point (vanishing point) that meets with the horizon among a plurality of lines dividing in the horizontal direction along the vertical axis of the entire area of interest.

The ordinate of the line is lost, so when the value y _P Y _W of the world coordinate system to converge to the infinity (∞) values can be calculated by Equation 1 as below.

According to a preferred embodiment, the ordinate value for the dividing line 20 is set to the ordinate of a position slightly lower than the vanishing line as in the example of the dividing line 20 of Fig. 2 in consideration of the uphill road and the downhill road environment It is good to set it to have value.

Preferably, the step 1 further includes a step of reducing the input image between the step 1 and the step 2.

3 is an example for explaining an input image size reduction step using the average interpolation method of the present invention. For reference, an example of FIG. 3 shows an image reduced to 1/2 size by using an input image (intensity) and an average interpolation method.

Referring to FIG. 3, rapid image processing is required for real-time recognition of input images such as lane recognition. Therefore, in order to speed up the processing speed of the algorithm, it is preferable to reduce the total size of the input image acquired from the camera by 1 / N (preferably 1/2) times.

At this time, the brightness value of the pixel newly determined at the time of image reduction can be configured to use Average Interpolation, which has a fast processing speed as shown in FIG. 3 and noise reduction effect due to smoothing phenomenon.

The brightness value (P) of the newly determined pixel after the reduction of the input image can be calculated as an average brightness value in a block of 2x2 size. That is, assuming that the brightness values of each pixel in a block of 2x2 size are "25, 30, 15, 50" as shown in FIG. 3, 25, 30, 15, 50) / 4 ") becomes the brightness value (P) of the newly determined pixel after the reduction of the input image.

If the method further includes a step of reducing the input image as described above, step 2 sets the first ROI and the second ROI with the reduced image.

<Step 3>

In the step 3 of the present invention, the weekly rate of the first ROI is calculated using the frequency of occurrence of the pixels corresponding to the BRI values within the first brightness level, and in the second BR lightness value range, And calculates the nighttime rate of the first ROI using the frequency of occurrence of the pixel corresponding to the brightness value.

Here, the first brightness level value range may be a first brightness level value; And a second brightness level value smaller than the first brightness level value.

The second brightness level value range may be a third brightness level value; And a fourth brightness level value smaller than the third brightness level value.

The second brightness level value may be greater than the fourth brightness level value, and the third brightness level value may be less than the first brightness level value.

Hereinafter, with reference to Figs. 4 and 5, the preferred embodiments of steps 3 to 6a will be described in detail.

FIG. 4 illustrates an example of a histogram of a first state of interest and a state of a week recognized according to the present invention, and FIG. 5 illustrates an example of a histogram of a night state and a first state of interest recognized according to the present invention.

As described above, Step 3 is a step of calculating the ratio of the daytime to the daytime for the first ROI divided through Step 2. At this time, the statistical characteristic information of the BRI is calculated And the like.

To do this, the histogram of the image is calculated, which can be expressed by the following equation (2).

The expression range of 'g' in Equation (2) may range from '0' to '255', which is a range of brightness brightness. Since the result of the above function may be small or large according to the size of the image, normalization is performed by dividing 'n _g ' by the total number of pixels of the image, and a normalized histogram is calculated as shown in Equation 3 below.

The normalized histogram of Equation (3) can be expressed as Probability because the corresponding pixel shows the frequency occurring within the expression range of intensity brightness as shown in Equation (4) below, and the function value for the entire intensity luminance value is added Case 1 is satisfied.

The sky_daytime_ratio of the first area of interest (i.e., the sky area) can be calculated using Equation (5), and the night ratio sky_nighttime_ratio of the first area of interest (i.e., sky area) Can be calculated using Equation (6).

Here, T _DSMin is the minimum brightness value (second brightness brightness value) used in calculating the weekly ratio of the first ROI,

T _DSMax : the maximum brightness value (first brightness brightness value) used in calculating the weekly ratio of the first ROI,

Here, T _NSMin : the minimum brightness value (fourth brightness brightness value) used in calculating the nighttime ratio of the first ROI,

T _NSMax : maximum brightness value (third brightness brightness value) used in calculating the nighttime ratio of the first ROI

In Equation (5), the range of the minimum brightness value (T _DSMin , second brightness brightness value) and maximum brightness value (T _DSMax , first brightness brightness value) used in the calculation of the weekly rate of the first ROI Brightness intensity value range) is set to a high range of values in consideration of the characteristics of the brightness intensity value of the bright sky in the daytime sky.

On the other hand, the range of the minimum brightness value (T _NSMin , the fourth brightness brightness value) and the maximum brightness value (T _NSMax , the third brightness brightness value) used in calculating the nighttime ratio of the first ROI in Equation 2 contrast level brightness value) is set to a low range value in consideration of the characteristic of the dark intensity brightness value of night sky.

&Lt; Step 4 to Step 6 >

Step 4 and Step 5 of the present invention compare the weekly rate of the first region of interest calculated in Step 3 with a predetermined first threshold value and finally determine the input image as a daytime state according to the comparison result, 1 nighttime ratio of the area of interest to a predetermined second threshold value.

Specifically, if the weekly rate of the first ROI calculated in the step 3 satisfies the condition 1 "week rate> first threshold ", it is determined that the final recognition status is the week, and otherwise, Is less than or equal to the first threshold value), it is determined whether or not the nighttime rate of the first ROI calculated in Step 3 satisfies Condition 2 "nighttime rate > second threshold ".

Here, the first threshold value of the condition 1 and the second threshold value of the condition 2 are thresholds for the probability between '0' and '1' satisfying the characteristic of the equation (4) It is preferable that the first threshold value is set to a relatively low value and the second threshold value is set to a relatively high value.

The step 6 of the present invention is configured to recognize the input image as a night scene when the nighttime ratio of the first ROI is greater than the second threshold as a result of the comparison according to the condition 2 of the step 5.

On the other hand, when both of the conditions 1 and 2 are not satisfied, the brightness distribution of the histogram is displayed in a range that is too bright or too dark. In this case, it is determined that the recognition is ambiguous and the detailed step for more detailed inspection is entered do.

That is, if it is determined that the daytime and nighttime conditions are not determined through the above-described process, that is, if the nighttime rate of the first ROI is less than or equal to the second threshold value, And calculate a difference value between the calculated weekly rate and the nighttime rate of the first ROI.

Specifically, when the determination of the daytime and nighttime state of the input image is ambiguous, the absolute value of the difference between the daytime rate (sky_daytime_ratio) and the nighttime rate (sky_nighttime_ratio) of the first ROI calculated through Step 3 is calculated do.

&Lt; Step 7 and Step 8 &

Step 7 and Step 8 of the present invention are based on the ratio difference value calculated in the detailed step of Step 6 to enter the step of calculating the degree of disorder of the first region of interest or calculate the daytime / nighttime ratio of the second region of interest .

First, when the ratio difference value (difference_ratio) calculated in step 6 satisfies the condition 3 &quot; ratio difference value <the third threshold value &quot;, it is determined that the daytime and nighttime ratio difference is small and the degree of disorder of the first ROI is calculated . Here, the third threshold value may be a value represented by a probability of '0' to '1'.

A method for calculating the degree of disorder of the first region of interest will be described in detail.

If the ratio difference value is less than the third threshold value, it is necessary to perform a more sophisticated inspection on the first ROI since the difference in the ratio between the daytime and the nighttime of the first ROI is small and it is ambiguous to distinguish between the daytime and the nighttime.

Especially, in the daytime situation, it is hard to recognize because there are scenes in which the distribution of brightness variation is varied rather than the nighttime situation. In order to enable robust recognition even in such a situation, the randomness degree, which is statistical characteristic information, is calculated for the corresponding sky area (i.e., the first ROI) using the following Equation (7).

Here, e (z)

z: random variable representing brightness

p (z _i ): corresponding histogram probability, i = [0, 1, 2, ... , 255]

Thereafter, the method further includes the step 8b of comparing whether the randomness value calculated by the equation (7) satisfies the condition 4 &quot; disorder value &quot; fourth threshold value &quot;, and if the comparison result in the step 8b satisfies the condition 4 It is determined that the first ROI is low and the final recognition state is determined as the week (step 9b).

If the comparison result in step 8b does not satisfy the condition 4 (i.e., if the disorder value is greater than or equal to the fourth threshold value), it is determined that the disorder degree is high and the final recognition state is judged as night (step 9b) do.

By applying the above-described method, it can be seen that normal scenes can be recognized even in scenes that are difficult to recognize in the daytime situation as shown in FIGS. 6 to 8, the upper portion of the right photograph shows the first region of interest, and the lower portion of the right photograph shows the histogram of the first region of interest.

Specifically, FIG. 6 illustrates a situation in which the wiper operates due to rain during highway driving in the daytime, and the result of recognizing the state of the weekday normally even when the sky area (first ROI) is dark. FIG. 7 shows the result of recognizing the daytime state normally even when the first area of interest is dark due to the mountains on the tunnel during the daytime on the highway. FIG. 8 shows a state in which there is a high-priced leg when traveling on a highway, and a result of recognizing the state of the weekday normally even when the first area of interest is dark.

If the ratio difference value (difference_ratio) calculated in step 6 does not satisfy the condition 3 &quot; ratio difference value <third threshold value &quot; (i.e., the ratio difference value is greater than or equal to the third threshold value) And calculate the daytime rate and the nighttime rate of the second area of interest, which can use the statistical property information of the intensity of brightness such as the daytime rate calculation step of the first area of interest described above.

Here, the weekly ratio of the second ROI is calculated using the frequency of occurrence of the pixels corresponding to the brightness values within the first 'lightness brightness value range.

And the nighttime ratio of the second ROI is calculated using the frequency of occurrence of the pixels corresponding to the brightness values within the second range of brightness values.

Here, the first 'lightness brightness value range is defined as a first' lightness brightness value; And a second intensity of lightness value less than the first intensity of lightness value.

The second 'brightness level brightness value range is a third brightness level brightness value; And a fourth brightness level brightness value smaller than the third brightness level brightness value.

In addition, the second 'lightness brightness value may be set to be larger than the fourth' lightness brightness value, and the third 'lightness brightness value may be less than the maximum' first 'lightness brightness value.

More specifically, the probability distribution of the normalized histogram is calculated using Equations (2) to (4).

Based on this, the week-to-day ratio (ground_daytime_ratio) of the second ROI can be calculated by the following equation (8).

Here, T _DGMin is the minimum brightness value (second 'brightness level brightness value) used in calculating the weekly rate of the second ROI,

T _DGMax : the maximum brightness value (the first 'brightness intensity value') used in calculating the weekly rate of the second ROI,

On the other hand, the range of the minimum (T _DGMin ) and the maximum brightness value (T _DGMax ) used in calculating the weekly ratio of the second ROI (i.e., the first ' It is preferable to set the values of the range excluding the brightness range that is too bright or too dark by considering the brightness distribution of the ground road area including the shadows of the trees and the like.

FIG. 9 shows the brightness distribution of a land area including a shadow area such as a building or a tree line, which is common in the daytime. As can be seen from FIG. 9, it can be seen that the brightness distribution that forms the peak of the histogram is concentrated in the vicinity of the center.

The weekly ratio (ground_nighttime_ratio) of the second ROI can be calculated by the following equation (9).

Here, T _NGMin : the minimum brightness value (the fourth brightness intensity value) used in the calculation of the nighttime ratio of the second ROI,

T _NGMax : the maximum brightness value (the third brightness intensity value) used in the calculation of the nighttime ratio of the second ROI,

10, when the preceding vehicle and the opposing vehicle are located in the road area, the range of the dark brightness distribution is relatively wide due to the influence of the headlights and tail lights. However, as shown in the right- In the absence of a vehicle, the range of dark brightness distribution is relatively narrow. Taking this into consideration, the range of the minimum (T _NGMin , the fourth ' _darkness brightness value) and the maximum brightness value (T _NGMax , the third' _darkness brightness value) used in the calculation of the nighttime ratio of the second ROI 'Intense brightness value range) is set to allow the dark brightness distribution to be wide enough.

(Step 8a) whether the daytime and nighttime ratios of the second ROI calculated using equations (8) and (9) satisfy Condition 5 &quot; daytime rate> nighttime rate & If it is satisfied, the final recognition state of the input image is determined as a week. If not, the final recognition state is determined to be night (step 9a) if the daytime ratio is less than or equal to the night ratio.

Figs. 11 and 12 show the histograms of the finally recognized weekly and nighttime states and the corresponding terrestrial regions (i.e., the second ROI) through steps 7, 8a, and 9a, respectively. As can be seen from FIGS. 11 and 12, it can be confirmed that the daytime and nighttime conditions can be normally recognized even when the brightness distribution of the ground road area is not very bright or dark.

The method of recognizing day and night of an input image according to the present invention described and illustrated above can be realized in a program form and realized through an electric / electronic device. The term &quot; electric &lt; RTI ID = 0.0 &gt; electronic &lt; / RTI &gt; device includes both a computer and an electrical and electronic device including a central processing unit and a memory such as a smart terminal.

In the above case, the recording medium of the present invention includes: an electric / electronic device having a function of receiving an image including a sky area and a ground area; Setting an area including at least the sky area as a first area of interest in the input image; Calculating a weekly rate of the first ROI by using the frequency of occurrence of pixels corresponding to the respective BRI values within the first brightness level value range, and calculating a weekly ratio of the pixels corresponding to the brightness values within the second brightness level range, Calculating a nighttime rate of the first ROI using the frequency of occurrence of the first ROI; Comparing the calculated weekly rate of the first ROI with a predetermined first threshold; If the daytime rate of the first ROI is greater than the first threshold value, recognizing the input image as a daytime state, and comparing the nighttime rate of the first ROI with a predetermined second threshold value Function; And a program for realizing a function of recognizing the input image as a nighttime state when the nighttime rate of the first ROI is greater than the second threshold value as a result of the comparison, .

According to a preferred embodiment, the recording medium of the present invention further comprises: a function of setting a remaining region of the input image excluding the first region of interest as a second region of interest; Calculating a difference value between the calculated weekly rate and the nighttime rate of the first ROI when the comparison result indicates that the nighttime rate of the first ROI is less than the second threshold; Comparing the difference value with a predetermined third threshold value; If the difference is smaller than the third threshold value, the degree of disorder of the first region of interest is calculated. If the difference is larger than the third threshold value, the frequency of occurrence of the pixels corresponding to the brightness values within the first ' And calculating the nighttime rate of the second ROI using the frequency of occurrence of the pixels corresponding to the brightness values within the second ' And may be a medium that can be read by an electric / electronic apparatus in which a program for realizing the present invention is recorded.

Comparing the calculated degree of chaos of the first ROI with a predetermined fourth threshold when calculating the degree of chaos of the first ROI; And a program for realizing a function of recognizing the input image as a daytime state when the degree of disorder of the first ROI is greater than the fourth threshold value and recognizing the input image as a nighttime state when the comparison is less than the fourth threshold, Lt; / RTI &gt;

A function of comparing the calculated weekly rate and the nighttime rate of the second ROI when calculating the weekly rate and the nighttime rate of the second ROI; A program for realizing a function of recognizing the input image as a daytime state when the daytime rate of the second area of interest is greater than the nighttime rate of the second area of interest as a result of the comparison, Lt; / RTI &gt; readable medium.

The detailed configurations and features of the above-described functions are the same as those described and illustrated in FIG. 1 to FIG. 12, and a detailed description thereof will be omitted.

While the preferred embodiments of the present invention have been described and illustrated above using specific terms, such terms are used only for the purpose of clarifying the invention, and it is to be understood that the embodiment It will be obvious that various changes and modifications can be made without departing from the spirit and scope of the invention.

For example, the nighttime rate of the first area of interest may not be calculated in the third step, but may be calculated in the fifth step described above, and then compared with the predetermined second threshold value in the calculated nighttime rate of the first area of interest It goes without saying that the same object can be achieved.

Such modified embodiments should not be understood individually from the spirit and scope of the present invention, but should be regarded as being within the scope of the claims of the present invention.

10: input image
11: First area of interest
13: second region of interest
20: Splitting line

Claims (16)

delete A first step of receiving an image including a sky area and a ground area;
A second step of setting at least an area including the sky area in the input image as a first ROI;
Calculating a weekly rate of the first ROI by using the frequency of occurrence of pixels corresponding to the respective BRI values within the first brightness level value range, and calculating a weekly ratio of the pixels corresponding to the brightness values within the second brightness level range, A third step of calculating a nighttime rate of the first ROI using the frequency of occurrence of the first ROI;
A fourth step of comparing the calculated weekly rate of the first ROI with a predetermined first threshold value;
If the daytime rate of the first ROI is greater than the first threshold value as a result of the comparison of the fourth ROI, recognizes the input image as a daytime state, and if the input image is smaller than the first threshold, 2 &gt; threshold value; And
And a sixth step of recognizing the input image as a night scene when the nighttime ratio of the first ROI is greater than the second threshold as a result of the fifth step,
The first brightness level value range is a first brightness level value; And a second brightness level value smaller than the first brightness level value,
The second brightness brightness value range is a third brightness brightness value; And a fourth brightness level value smaller than the third brightness level value,
Wherein the second intense brightness value is greater than the fourth intense brightness value and the third intense brightness value is less than the first intense brightness value.
3. The method of claim 2,
Wherein the daytime rate and the nighttime rate of the first area of interest, the first threshold value, and the second threshold value are values represented by a probability of '0' to '1'. Way.
A first step of receiving an image including a sky area and a ground area;
A second step of setting an area including at least the sky area as a first area of interest in the input image and setting an area other than the first area of interest as a second area of interest;
Calculating a weekly rate of the first ROI by using the frequency of occurrence of pixels corresponding to the respective BRI values within the first brightness level value range, and calculating a weekly ratio of the pixels corresponding to the brightness values within the second brightness level range, A third step of calculating a nighttime rate of the first ROI using the frequency of occurrence of the first ROI;
A fourth step of comparing the calculated weekly rate of the first ROI with a predetermined first threshold value;
If the daytime rate of the first ROI is greater than the first threshold value as a result of the comparison of the fourth ROI, recognizes the input image as a daytime state, and if the input image is smaller than the first threshold, 2 &gt; threshold value;
If the nighttime ratio of the first ROI is greater than the second threshold value as a result of the comparison of the fifth ROI, A sixth step of calculating a difference value between the calculated weekly rate and the nighttime rate of the first ROI;
A seventh step of comparing the difference value of the sixth step with a predetermined third threshold value; And
If the difference is smaller than the third threshold, the degree of disorder of the first ROI is calculated. If the difference is greater than the third threshold, Calculates a daytime rate of the second ROI using the frequency of occurrence of the first ROI, and calculates a nighttime rate of the second ROI using the frequency of occurrence of pixels corresponding to the brightness values of the second ROI within the second ' The method comprising the steps of: (a) inputting an input image;
5. The method of claim 4,
If the randomness degree of the first ROI is calculated in the eighth step,
8b) comparing the calculated degree of chaos of the first ROI with a predetermined fourth threshold; And
Further comprising the step of recognizing the input image as a daytime state when the degree of abnormality of the first ROI is greater than the fourth threshold value and recognizing the input image as a nighttime state when the degree of abnormality of the first ROI is greater than the fourth threshold value as a result of the comparison of step 8b A method for recognizing day and night of an input image.
5. The method of claim 4,
In the eighth step, when calculating the weekly rate and the nighttime rate of the second ROI,
Comparing the calculated weekly rate and the nightly rate of the second ROI; And
If it is determined in step 8a that the daytime rate of the second area of interest is greater than the nighttime rate of the second area of interest, recognizing the input image as a daytime state; And the day and night recognition method of the input image.
5. The method of claim 4,
Between the first step and the second step,
Further comprising reducing the total size of the input image in the first step to 1 / N.
8. The method of claim 7,
Wherein the reduction of the input image is performed using an average interpolation method.
3. The method of claim 2,
Wherein the nighttime rate of the first ROI is calculated by comparing the nighttime rate of the first ROI calculated in the fifth step with a predetermined second threshold value A method for recognizing day and night of an input image.
5. The method of claim 4,
Wherein the first ROI and the second ROI are divided by a dividing line having a vertical coordinate value lower than a vanishing line of the input image,
Wherein the first area of interest includes a top portion of the land area.
3. The method of claim 2,
Wherein the fifth step is configured to compare the calculated nighttime rate of the first ROI with a predetermined second threshold even if the weekly rate of the first ROI is equal to the first threshold, The day and night recognition method of the input image.
5. The method of claim 4,
The sixth step is configured to calculate a difference value between the calculated weekly rate and the nighttime rate of the first ROI even if the calculated nighttime rate of the first ROI is equal to the second threshold,
Wherein the eighth step is configured to calculate a daytime rate and a nighttime rate of the second ROI even when the difference is equal to a third threshold value.
delete In electrical and electronic devices,
Receiving a video including a sky area and a ground area;
Setting an area including at least the sky area as a first area of interest in the input image and setting an area other than the first area of interest as a second area of interest;
Calculating a weekly rate of the first ROI by using the frequency of occurrence of pixels corresponding to the respective BRI values within the first brightness level value range, and calculating a weekly ratio of the pixels corresponding to the brightness values within the second brightness level range, Calculating a nighttime rate of the first ROI using the frequency of occurrence of the first ROI;
A function of comparing the calculated weekly rate of the first ROI with a predetermined first threshold value (hereinafter referred to as 'first comparison');
Wherein the input image is recognized as a daytime state when the weekly rate of the first ROI is greater than the first threshold value and the nighttime rate of the first ROI is smaller than a second predetermined threshold (Hereinafter, referred to as &quot; second comparison &quot;);
Recognizing the input image as a night scene when the nighttime ratio of the first ROI is greater than the second threshold as a result of the second comparison;
A function of calculating a difference value between the calculated weekly rate and the nighttime rate of the first ROI when the nighttime rate of the first ROI is smaller than the second threshold as a result of the second comparison;
Comparing the difference value with a predetermined third threshold value (hereinafter, referred to as 'third comparison'); And
As a result of the third comparison, when the difference is smaller than the third threshold value, the degree of disorder of the first ROI is calculated. If the difference is larger than the third threshold, the pixels corresponding to the brightness values within the first ' Calculates a daytime rate of the second ROI using the frequency of occurrence, and calculates a nighttime rate of the second ROI using the frequency of occurrence of pixels corresponding to the brightness values of the respective intensities within the second ' A program for realizing the functions of the present invention.
15. The method of claim 14,
Comparing the calculated degree of disorder of the first ROI with a predetermined fourth threshold (hereinafter referred to as 'fourth comparison') when calculating the degree of disorder of the first ROI; And
Recording a program for further realizing a function of recognizing the input image as a daytime state when the degree of anarchy of the first ROI is greater than the fourth threshold value and recognizing the input image as a nighttime state when the degree of abnormality of the first ROI is larger than the fourth threshold value as a result of the fourth comparison A medium readable by an electric or electronic device.
15. The method of claim 14,
A function of comparing the calculated weekly rate and the nighttime rate of the calculated second interest area (hereinafter, referred to as 'fifth comparison') when calculating the weekly rate and the night rate of the second interest area; And
A program for further realizing a function of recognizing the input image as a daytime state when the weekly rate of the second ROI is greater than the nighttime rate of the second ROI as a result of the fifth comparison, In the electronic device.

Images