JPH0969160A - Picture brightness discriminating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To surely discriminate a background part and a character part even when their discrimination is difficult by inputting operation results outputted from 1st and 2nd inner product arithmetic units, mutually comparing the sizes of the inputs and judging brightness. SOLUTION: This picture brightness discriminating device is provided with an OR/NOT arithmetic unit 4 for inputting a bright part binary picture (B picture) 12 and a dark part binary picture (D picture) 13 respectively from a local bright part binarizing device 2 and a local dark part binarizing part 3, an integration/histogram calculating device 5a for inputting the B picture 12 from the device 2 and an integration/histogram calculating part 5b for inputting the D picture 13 from the device 3. And also, an inner produce arithmetic unit 6a for inputting a B area histogram 15 and a Z area histogram 17 respectively from the integration/histogram calculating devices 5a, 5c, an inner product arithmetic unit 6b for inputting a D area histogram 16 and the histogram 17 respectively from the devices 5b, 5c and a brightness judging device 7 for inputting the output of the units 6a, 6b and judging brightness, are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image brightness discrimination device applied to an image / pattern recognition device or the like.

[0002]

2. Description of the Related Art When reading license plates or container numbers of domestic automobiles, there are various relationships between the character color and the background color. For example, the background color (ground color) is white, while the character color is green and black. There are various types such as yellow, white for green, and black for yellow.

As described above, there are various combinations of the background color and the character color. In the character extraction and character recognition performed on such a grayscale (black and white) image, first,
It is necessary to determine whether the character is locally on the bright side or locally on the dark side, and a bright / dark determining device for the character color is necessary.

As shown in FIG. 4, a conventional image brightness discrimination device includes a data input device 1, a local light area binarization device 2, a local dark area binarization device 3, pixel number counting devices 8a and 8b, and a lightness / darkness judgment device. Device 7.

In the above, the local bright part binarizing device 2 and the local dark part binarizing device 3 to which the image data is inputted from the data input device 1 are, for example, when the character is green and the ground color is white, FIG.
Pixels of the local bright area binary image (B image) 12 and the local dark area binary image (D image) 13 are set as the pixels of the black portion 20 shown in FIGS. Output to.

The pixel number counting devices 8a and 8b that have input these images 12 and 13 count the number of pixels of "1" for each of the images 12 and 13 and output them to the light / dark judging device 7, and the light / dark judging device 7 The number of pixels of "1" output from each of the pixel number counting devices 8a and 8b is compared, and one of bright and dark, which has a larger number of pixels of "1", is determined to be the background color.

[0007]

In the conventional image brightness discrimination device, as described above, the number of pixels of "1" output by the local light area binarizing device and the local dark area binarizing device are compared to each other. It was determined whether the color was light or dark.

However, the number of pixels of "1" may be smaller in the background portion depending on the situation of the background portion around the character portion, and in the case of a license plate or container number of a foreign automobile, the background portion is a character. There are cases in which the number of characters is larger than the number of characters and the number of characters is greater than the number of characters in the background. In such a case, it is difficult to determine the characters. The present invention seeks to solve the above problems.

[0009]

According to a first aspect of the present invention, there is provided a data input device for capturing a target image, and a target light image is input from the data input device to input a local bright portion binary value. A local bright area binarization device that outputs an image (hereinafter referred to as a B image) and a local dark area binarization device that outputs a local dark area binary image (hereinafter referred to as a D image), and the above B image and D image. A logical sum / negative operation device for inputting and outputting a logical sum / negative binary image (hereinafter referred to as a Z image), a local bright portion binary area histogram (hereinafter referred to as a B area) by inputting the target image and the B image A first integration / histogram calculation device that outputs a histogram), and a second integration / histogram that inputs the target image and the D image and outputs a local dark area binary area histogram (hereinafter, referred to as a D area histogram) Calculator, above A third integrating / histogram calculating device for inputting an elephant image and a Z image and outputting a logical sum / negative binary region histogram (hereinafter, referred to as a Z region histogram), and inputting the B region histogram and the Z region histogram A first inner product computing device that outputs the computation result, a second inner product computing device that inputs the D region histogram and the Z region histogram and outputs the computation result, and the first and second inner product computing devices It is characterized in that a brightness determination device for determining the brightness is provided by inputting each output calculation result and comparing the magnitudes thereof.

In the above description, the local bright area binarization apparatus and the local dark area binarization apparatus, which have input the target image from the data input apparatus, respectively extract the locally bright portion of the image to obtain 2
A binarized B image and a binarized D image by extracting a locally dark portion of the image are output.

The B image and the D image are input to a logical sum / negation arithmetic unit, which binarizes a portion which is neither the B image nor the D image, and the logical sum / negation binary of the B image and the D image. A Z image, which is an image, is output.

The B image is input to the first integration / histogram calculation device together with the target image, and the calculation device calculates and outputs the B area histogram. Similarly, the second integration / histogram calculation device inputs the D image and the target image, calculates and outputs the D region histogram, and the third integration / histogram calculation device inputs the Z image and the target image. Then, the Z area histogram is calculated and output.

The B area histogram and the Z area histogram are input to a first inner product calculating device, which regards each as a vector and calculates and outputs the inner product,
The D area histogram and the Z area histogram are input to the second inner product computing device, and the same computing device similarly obtains and outputs the inner product.

The inner product calculation results output by the respective inner product calculation devices are input to the image brightness determination device, and the image brightness determination device determines the extraction part (character part) and the background part respectively in the following manner.

When (B area histogram)-(Z area histogram)> (D area histogram)-(Z area histogram), the local bright area (B area) is extracted as the background area and the local dark area (D area) is extracted. Part (B area histogram), (Z area histogram) <
(D area histogram) When (Z area histogram), the local bright area (B area) is the extraction area, and the local dark area (D area) is the background area (B area histogram) (Z area histogram) =
(D area histogram) When (Z area histogram), the brightness is unknown.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION An image brightness discrimination device according to an embodiment of the present invention will be described with reference to FIGS.

The apparatus according to the present embodiment shown in FIG. 1 includes a data input device 1 for taking in a target image 11 and a local bright area binarizing device 2 for inputting the target image 11 from the input device 1.
And the local dark space binarization device 3, and the respective binarization devices 2 and 3
A binary OR image / Binary image (B image) 12 and a dark binary image (D image) 13 are input. A logical sum / negation operation unit 4; An integration / histogram calculation device 5a for inputting the B image 12 from the device 2, an integration / histogram calculation device 5b for inputting the target image 11 from the data input device 1 and a D image 13 from the local dark area binarization device 3; Integration / histogram calculation device 5 for inputting the target image 11 from the data input device 1 and Z image 14 from the logical sum / negation operation device 4.
c, the B area histogram 15 and the Z area histogram 1 from the integration / histogram calculation devices 5a and 5c, respectively.
The inner product calculating device 6a for inputting 7 and the inner product calculating device 6b for inputting the D region histogram 16 and the Z region histogram 17 from the integrating / histogram calculating devices 5b, 5c, and the outputs of the inner product calculating devices 6a, 6b, respectively. And a brightness determination device 7 for outputting a determination result.

Next, regarding the operation of the apparatus according to the present embodiment, a case where a license plate of a domestic automobile whose characters are green and whose ground color is white is used as a discrimination target will be described with reference to FIG.
And FIG.

In the above, the local bright area binarization device 2 and the local dark area 2 in which the target image 11 is input from the data input device 1 are input.
Each of the digitizing devices 3 has a target image 1 similar to the conventional device.
1 is binarized, and the local bright part binary image (B image) 12 in which the pixels of the bright part (the part painted black in FIG. 2) around the character are set to “1” and the dark part representing the character (FIG. 2). The local dark area binary image (D image) 13 in which the pixels in the black area in FIG.

The B image 12 and the D image 13 output from the local bright area binarizing device 2 and the local dark area binarizing device 3 are logically ORed.
It is input to the negative operation unit 4, and the same operation unit 4 inputs the B image 12
However, the binary image of the part that is not the D image 13, that is, the part that is neither "1" in the B image 12 nor the D image 13 is "1".
The binary image is output as the Z image 14.

The B image 12 is input to the integration / histogram calculation device 5a together with the target image 11 of the luminance histogram shown in FIG.
The B area histogram 15 indicating the luminance distribution of the target image 11 for the portion where the pixel of “1” is “1” is output. This histogram 15 is shown in FIG. 3 (b).

Similarly, the D image 13 is also input to the integration / histogram calculation device 5b together with the target image 11, and the calculation device 5b outputs the luminance distribution of the target image 11 for the portion where the pixel of the D image 13 is "1". A D-region histogram 16 indicating is output. This histogram 16 is shown in FIG.
It is shown in (c).

The same applies to the Z image 14,
Integration / histogram calculation device 5c together with the target image 11
And the calculation device 5c outputs a Z region histogram 17 indicating the luminance distribution of the target image 11 for the portion where the pixel of the Z image 14 is “1”. This histogram 17 is shown in FIG. 3 (d).

The above integration / histogram calculation devices 5a, 5
The B area histogram 15 and the Z area histogram 17 output from c are input to the inner product calculating device 6a, each of which is regarded as a vector to calculate the inner product, and also output from the integrating / histogram calculating devices 5b and 5c. The D-region histogram 16 and the Z-region histogram 17 are input to the inner product calculating device 6b, each of which is regarded as a vector, and the inner product thereof is obtained.
Is input to

The light / dark judging device 7 which inputs the respective inner products obtained by the inner product calculating devices 6a and 6b makes the following judgment. That is, (B area histogram) / (Z area histogram)>
(D area histogram) When (Z area histogram), the local bright area (B area) is the ground color area, and the local dark area (D area) is the character area (B area histogram). (Z area histogram) <
(D area histogram) When (Z area histogram), the local bright area (B area) is the character area, and the local dark area (D area) is the ground color area (B area histogram) (Z area histogram) =
(D area histogram) When (Z area histogram), it is determined that the brightness is unknown.

The processes performed by the inner product computing devices 6a and 6b and the brightness determination device 7 will be specifically described below by using assumed numbers.

Each integration / histogram calculation device 5
Regarding the luminance histograms output by a, 5b, and 5c,
Assuming that the number of gradations is 8 (8 gradations of 0 to 7), the brightness distribution of the B area histogram 15, the D area histogram 16, and the Z area histogram 17 can be divided into brightness areas of 0 to 7, and this is a vector. It can be expressed as follows. The numbers in parentheses are from 0 to 7 from left to right.
Shows the brightness of.

Vector of B region histogram Vb =
[0,0,0,0,0,10,20,10] D region histogram vector Vd = [0, 5,10, 5,
0, 0, 0, 0] Z domain histogram vector Vz = [0, 0, 0, 0,
0, 4, 7, 4] The arithmetic operations performed by the inner product arithmetic units 6a and 6b are Vb × Vz and Vd × Vz, and the respective arithmetic results are 220 and 0, which are input to the light / dark determination device 7.

Brightness determination device 7 to which the above calculation result is input
Since Vb × Vz> Vd × Vz, it is determined that the local bright portion is the ground color.

In the present embodiment, the brightness / darkness judgment device 7 inputs the calculation results of the inner product calculation devices 6a and 6b to make a judgment. For each of the brightness histograms, the B region brightness average values Ab and D regions are calculated. Brightness average value Ad,
It is also possible to obtain the Z-region luminance average value Az and input it, and the lightness / darkness determination device 7 may perform the determination under the following conditions.

If Az <(Ab + Ad) / 2, the local dark part is the ground color part. If Az> (Ab + Ad) / 2, the local bright part is the ground color part. Az = (Ab + Ad) / 2 is unknown.

[0032]

The image brightness discrimination device of the present invention comprises a local bright area binarization device for inputting a target image from a data input device and outputting a B image, and a local dark area binarization device for outputting a D image.
A logical sum / negation operation device that inputs the B image and the D image and outputs the Z image, and inputs the target image, inputs the B image, outputs the B area histogram, and inputs the D image, the D area histogram To output the Z image and output the Z region histogram, the first, second and third integration / histogram calculation devices respectively input the Z region histogram and input the B region histogram to obtain the calculation result. First and second inner product arithmetic devices for outputting and outputting a calculation result of the D region histogram, and first and second
Since the determination of the Z area is performed by providing the brightness determination device for inputting the calculation result of the inner product calculation device, the determination can be made with high accuracy even when it has been difficult to extract the character portion in the related art. It will be possible.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of an image brightness / darkness determining device according to an embodiment of the present invention.

FIG. 2 is an operation explanatory view according to the embodiment.

FIG. 3 is an explanatory diagram of a luminance histogram according to the above embodiment, where (a) is a target image, (b) is a B area, and (c) is a D area.
Area, (d) is a Z area luminance histogram.

FIG. 4 is an explanatory view of a conventional device.

FIG. 5 is a diagram showing a value output from a binarizing device according to the above conventional device.
It is explanatory drawing of a value image, (a) is a B image, (b) is a D image.

[Explanation of symbols]

 1 Data Input Device 2 Local Bright Area Binarization Device 3 Local Dark Area Binarization Device 4 Logical Sum / Negation Operation Device 5a, 5b, 5c Integration / Histogram Calculation Device 6a, 6b Inner Product Operation Device 7 Bright / Dark Judgment Device 11 Target Image 12 B image 13 D image 14 Z image 15 B area histogram 16 D area histogram 17 Z area histogram

Claims (1)

[Claims] 1. A data input device for capturing a target image, and a local bright region binarization device for inputting the target image from the data input device and outputting a local bright region binary image, and a local dark region binary image. A local dark area binarization device, a logical sum / negative operation device for inputting the local bright area binary image and the local dark area binary image and outputting a logical sum / negative binary image, the target image and the local bright area 2 A first integration / histogram calculation device that inputs a value image and outputs a local bright area binary region histogram, and a second integration and histogram calculation device that inputs the target image and the local dark area binary image and outputs a local dark area binary region histogram. An integrating / histogram calculating device, a third integrating / histogram calculating device for inputting the target image and the logical sum / negative binary image and outputting a logical sum / negative binary region histogram, and the local bright part binary region histogram Logical sum A first inner product computing device for inputting a negative binary region histogram and outputting the calculation result; a first inner product computing device for inputting the local dark space binary region histogram and the logical sum / negative binary region histogram, and outputting the computation result 2. An image characterized by including an inner product arithmetic unit 2 and a light / dark judging unit for inputting the respective calculation results output from the first and second inner product arithmetic units and comparing the magnitudes thereof to judge the lightness / darkness. Bright / dark discriminator.