JP2002354248A - Method, device and program for binarizing image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To binarize a multilevel image with a high quality. SOLUTION: A digital camera 100 is provided with a CCD 101 which inputs the luminance value of each of pixels constituting a multi-valued image, an average luminance value calculator 120 which calculates a local average luminance value including a pixel to be binarized on the basis of the luminance value inputted from the CCD 101, a binarization threshold setting circuit 122 which sets a binarization threshold to a value lower than the local average luminance value in accordance with the local average luminance value calculated by the average luminance value calculator 120, and a binarizing device 123 which binarizes the pixel to be binarized by the binarization threshold set by the binarization threshold setting circuit 122.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an image binarizing device,
In particular, the present invention relates to an image binarization method and an image binarization program, and more particularly to an image binarization method that removes the influence of contrast reduction and uneven brightness due to surface reflection from a multivalued image input by an image input device or the like having an irregular light source. The present invention relates to a binarization device, an image binarization method, and an image binarization program.

[0002]

2. Description of the Related Art Conventionally, when electronically storing a document,
A scanner or a scanner unit of a copier or a facsimile machine is exclusively used as an image input device. Such a scanner (scanner unit) includes a light source in the apparatus, and reflects reflected light of a document (document) of light emitted from the light source into a CC.
It was read by D or the like. Further, the read image is binarized and stored as needed. As for the image read in this way, since the light source and the optical system were constant, the occurrence of luminance unevenness and shadow was constant, and correction could be easily performed. Therefore, a high-quality image can be output, and the digital image can be binarized with high quality and easily.

On the other hand, in recent years, video cameras and digital cameras have been developed, and demands for character recognition of images input from such devices have been increasing. In particular, digital cameras have remarkably increased pixels and miniaturized, and have begun to be used for various purposes as portable information collection tools. For example, text information such as documents, signboards, and advertisements can be obtained as a binary image because sufficient information can be obtained with a binary image and the storage capacity required for storage is smaller than that of a multi-valued image. is there. Further, the binarized image can be transmitted by facsimile or can be subjected to character recognition for reuse.

As a conventional technique for appropriately binarizing a digital image, Japanese Patent Application Laid-Open No. Hei 3-237571 discloses an "image binarization threshold value calculating apparatus".
An "image binarizing device" disclosed in Japanese Patent Application Laid-Open No. 2-12591 is known.

[0005]

However, the prior art has the following problems. Images taken with a digital camera vary in the number, position, intensity, etc. of light sources, and are liable to produce shadows and uneven brightness. Moreover, since the shadow and the uneven brightness are not constant, it is not possible to apply a constant correction unlike a scanner unit of a copying machine or the like. Therefore, there is a problem that an image captured by a digital camera cannot be binarized with high quality.

[0006] In particular, when a flash is used, specular reflection occurs on the surface of the original depending on how light from the light source hits, and the contrast between the character and the background is reduced by the surface reflection. That is, the area where surface reflection occurs has higher average brightness (or lower average density) than the area where surface reflection does not occur, and so-called overexposure occurs.

Accordingly, when the binarization threshold is always the same operation regardless of the average luminance (or average density) as in the prior art, characters can be correctly binarized in the area where surface reflection occurs. There was a problem that it disappeared. In other words, there is a problem that an image having surface reflection cannot be binarized with high quality.

The present invention has been made in view of the above, and has as its object to binarize a multi-valued image with high quality.

[0009]

According to another aspect of the present invention, there is provided an image binarizing apparatus, comprising: a luminance value input means for inputting a luminance value of each pixel constituting a multi-valued image; ,
An average brightness value calculation unit that calculates a local average brightness value including a pixel to be binarized based on the brightness value input by the brightness value input unit; and a local brightness value calculated by the average brightness value calculation unit. Threshold value setting means for setting a binarization threshold value lower than the local average luminance value in accordance with the magnitude of a typical average luminance value, and a binarization threshold value set by the binarization threshold value setting means And a binarizing means for binarizing the pixel to be binarized.

That is, the invention according to claim 1 uses a binarization threshold having a different value according to the average luminance value, and employs an appropriate binarization threshold for an area where surface reflection occurs. I do.

According to a second aspect of the present invention, in the image binarization apparatus according to the first aspect, the binarization threshold value setting means is calculated by the average luminance value calculation means. The correction amount according to the local average luminance value,
When the average luminance value is in a predetermined intermediate range, a binarization threshold is set by subtracting a correction amount that is a larger value than when the average luminance value is outside the intermediate range from the local average luminance value. Features.

In other words, the invention according to claim 2 employs an appropriate binarization threshold with a correction amount that reduces the difference between the average luminance value and the binarization threshold in a luminance range in which surface reflection occurs, In the intermediate range, an appropriate binarization threshold value is adopted with a relatively large correction amount. Further, the difference between the average luminance value on the low luminance side and the binarization threshold is not increased, thereby suppressing the generation of noise on the low luminance side.

According to a third aspect of the present invention, in the image binarization apparatus according to the first aspect, the binarization threshold value setting means is calculated by the average luminance value calculation means. The binarization threshold is set by multiplying a local average luminance value by using a correction value of a larger value as the average luminance value becomes larger.

That is, the invention according to claim 3 employs an appropriate binarization threshold value in a luminance range in which surface reflection occurs, with a correction amount that reduces the difference between the average luminance value and the binarization threshold value.

The image binarizing device according to a fourth aspect of the present invention includes a luminance value input means for inputting a luminance value of each pixel constituting a multi-valued image, and a luminance value input means for inputting the luminance value. Average brightness value calculating means for calculating a local average brightness value including a pixel to be binarized based on the first brightness value, and a first correction amount for the local average brightness value calculated by the average brightness value calculating means. Subtraction threshold value calculation means for calculating a first binarization threshold value by subtraction, and multiplying the local average brightness value calculated by the average brightness value calculation means by a second correction amount less than 1 Multiplication threshold value calculation means for calculating a second binarization threshold value, and whether the magnitude of the local average brightness value calculated by the average brightness value calculation means is larger or smaller than a predetermined comparison value Comparing and judging, and comparing the local When it is determined that the average luminance value is larger than the comparison value, it is determined that the first binarization threshold value calculated by the subtraction threshold value calculation means is to be used, and the comparison determination is performed. When the local average luminance value is determined to be smaller than the comparison value by the means, the second binarization threshold calculated by the multiplication threshold calculation means is adopted. A binarization threshold determining unit for determining, and a binarization unit for binarizing the pixel to be binarized by the binarization threshold determined by the binarization threshold determining unit, I do.

That is, the invention according to claim 4 uses a binarization threshold value for an area where surface reflection occurs and a binarization threshold value for an ordinary area to selectively apply the threshold to an area where surface reflection occurs. On the other hand, the binarization threshold value, in which the difference from the average luminance value becomes relatively small as the average luminance value increases, is adopted. For a normal region, the overall tone after the binarization is smooth. Is adopted.

According to a fifth aspect of the present invention, there is provided the image binarization apparatus according to the fourth aspect, wherein the subtraction threshold value calculation means is configured to calculate the local brightness calculated by the average luminance value calculation means. The first binarization threshold is calculated by subtracting a correction amount corresponding to the magnitude of the luminance value from the average luminance value.

That is, the invention according to claim 5 can employ a more appropriate binarization threshold value in a luminance range in which surface reflection occurs.

According to a sixth aspect of the present invention, in the image binarizing apparatus according to the fourth or fifth aspect, the multiplying threshold setting means is calculated by the average luminance value calculating means. The second binarization threshold is calculated by multiplying a local average luminance value by a correction amount corresponding to the magnitude of the luminance value.

That is, the invention according to claim 6 employs a binarization threshold value in which a boundary is not generated and a smooth binary image is obtained in an area where surface reflection does not occur.

According to a seventh aspect of the present invention, there is provided the image binarizing apparatus according to any one of the first to sixth aspects, wherein the block dividing means for dividing the multi-valued image into blocks is provided. The average brightness value calculating means calculates the local average brightness value in units of blocks divided by the block dividing means.

That is, in the invention according to claim 7, an appropriate binarization threshold value is set for each block.

The image binarizing device according to claim 8 is the image binarizing device according to claim 7, wherein the average luminance value of a block adjacent to the block for calculating the local average luminance value is used. And a comparison value determining unit that determines the comparison value based on the comparison value.

That is, in the invention according to claim 8, the binarization threshold is set by more accurately determining the whiteout of the multi-valued image.

According to a ninth aspect of the present invention, there is provided an image binarizing apparatus, comprising: a density value input means for inputting a density value of each pixel constituting a multi-valued image; Average density value calculating means for calculating a local average density value including the pixel to be binarized based on the average density value calculated by the average density value calculating means. A binarization threshold setting unit that sets a binarization threshold higher than the local average density value; and a binarization threshold set by the binarization threshold setting unit, and binarizes the pixel to be binarized. And binarizing means.

That is, according to the ninth aspect of the present invention, a binarization threshold having a different value according to the average density value is used, and an appropriate binarization threshold is used for an area where surface reflection occurs. I do.

According to a tenth aspect of the present invention, in the image binarization apparatus according to the ninth aspect, the binarization threshold value setting means is calculated by the average density value calculation means. A correction amount according to the local average density value,
When the average density value is in a predetermined intermediate range, a binarization threshold is set by adding a correction amount that is a larger value than when the average density value is outside the intermediate range to the local average density value. Features.

That is, according to the tenth aspect of the present invention, in a density range in which surface reflection occurs, an appropriate binarization threshold value is adopted while using a correction amount that reduces the difference between the average density value and the binarization threshold value. In the intermediate range, an appropriate binarization threshold value is adopted with a relatively large correction amount. In addition, the difference between the average density value on the high density side and the binarization threshold is not increased, thereby suppressing the generation of noise on the high density side.

The image binarizing device according to claim 11 is the image binarizing device according to claim 9, wherein the binarization threshold value setting means is calculated by the average density value calculating means. The binarization threshold is set by multiplying the local average density value by using a smaller correction amount as the average density value increases.

That is, in the invention according to the eleventh aspect, in the density range where surface reflection occurs, an appropriate binarization threshold is adopted by a correction amount that reduces the difference between the average density value and the binarization threshold. I do.

According to a twelfth aspect of the present invention, there is provided an image binarization device, comprising: a density value input means for inputting a density value of each pixel constituting a multi-valued image; and a density value input by the density value input means. Average density value calculating means for calculating a local average density value including a pixel to be binarized based on the first density value, and a first correction amount for the local average density value calculated by the average density value calculation means. An addition threshold value calculating means for calculating a first binarization threshold value by adding, and a local average density value calculated by the average density value calculating means multiplied by a second correction amount which is one or more. Multiplication threshold value calculation means for calculating a second binarization threshold value, and whether the magnitude of the local average density value calculated by the average density value calculation means is larger or smaller than a predetermined comparison value Comparing and judging the local area by the comparing and judging means. If it is determined that the magnitude of the average density value is smaller than the comparison value, it is determined that the first binarization threshold calculated by the addition threshold calculation means is to be used, and the comparison is performed. When the determination unit determines that the local average density value is larger than the comparison value, the second binarization threshold calculated by the multiplication threshold calculation unit is employed. And a binarization unit for binarizing the pixel to be binarized by the binarization threshold determined by the binarization threshold determination unit. And

That is, according to the twelfth aspect of the present invention, a binarization threshold value for an area where surface reflection occurs and a binarization threshold value for an ordinary area are selectively used so that an area where surface reflection occurs is used. On the other hand, a binarization threshold value, in which the difference from the average density value becomes relatively smaller as the average density value becomes smaller, is adopted, and for a normal region, the overall tone after the binarization is smooth. Is adopted.

According to a thirteenth aspect of the present invention, in the image binarization device according to the twelfth aspect, the adding threshold value calculating means is configured to calculate the local value calculated by the average density value calculating means. The first binarization threshold is calculated by adding a correction amount corresponding to the magnitude of the density value to the average density value.

That is, the invention according to claim 13 can employ a more appropriate binarization threshold value in the density range in which surface reflection occurs.

According to a fourteenth aspect of the present invention, in the image binarization apparatus according to the twelfth or thirteenth aspect, the multiplication threshold value setting means is calculated by the average density value calculation means. The second binarization threshold is calculated by multiplying a local average density value by a correction amount corresponding to the magnitude of the density value.

That is, the invention according to claim 14 employs a binarization threshold value that can obtain a smooth binary image without generating a boundary in an area where surface reflection does not occur.

According to a fifteenth aspect of the present invention, in the image binarization apparatus according to any one of the ninth to fourteenth aspects, a block dividing means for dividing the multi-valued image into blocks is provided. The average density value calculation means calculates the local average density value based on the blocks divided by the block division means.

That is, in the invention according to claim 15, an appropriate binarization threshold is set for each block.

According to a sixteenth aspect of the present invention, in the image binarization apparatus according to the fifteenth aspect, the average density value of a block adjacent to the block for calculating the local average density value is set. And a comparison value determining unit that determines the comparison value based on the comparison value.

That is, the invention according to claim 16 sets the binarization threshold value by more accurately determining the whiteout of the multi-valued image.

In the image binarizing method according to the seventeenth aspect, a luminance value of a pixel forming a multi-valued image is input, and a luminance value of a local pixel set including a pixel to be binarized is calculated. For a pixel set that is a low-valued binarization threshold and is overexposed, a binarization threshold having a smaller difference from the average luminance value than for a pixel set that is not overexposed is determined. The image of the local pixel set is binarized using the binarization threshold and output.

That is, according to the seventeenth aspect of the present invention, an appropriate binarization threshold is set for an image having surface reflection.

In the image binarizing method according to the eighteenth aspect, a density value of a pixel constituting a multi-valued image is input, and the average density value of a local pixel set including a pixel to be binarized is obtained. For a pixel set that is a high-valued binarization threshold and is overexposed, a binarization threshold having a smaller difference from the average density value than for a pixel set that is not overexposed is determined. The image of the local pixel set is binarized using the binarization threshold and output.

That is, the invention according to claim 18 sets an appropriate binarization threshold even for an image having surface reflection.

An image binarization program according to a nineteenth aspect causes a computer to input a luminance value of a pixel constituting a multi-valued image and calculate an average of a local pixel set including a pixel to be binarized. A binarization threshold value lower than the luminance value, such that the difference from the average luminance value is smaller for a pixel set that is overexposed than for a pixel set that is not overexposed. A threshold is determined, and the image of the local pixel set is binarized using the binarization threshold and output.

That is, the invention according to claim 19 sets an appropriate binarization threshold value even for an image having surface reflection.

According to a twentieth aspect of the present invention, there is provided an image binarization program, wherein a computer inputs a density value of a pixel forming a multi-valued image, and averages a local pixel set including a pixel to be binarized. A binarization threshold value that is higher than the density value and has a smaller difference from the average density value for a pixel set that is overexposed than for a pixel set that is not overexposed. A threshold is determined, and the image of the local pixel set is binarized using the binarization threshold and output.

That is, according to the twentieth aspect of the present invention, an appropriate binarization threshold is set for an image having surface reflection.

[0049]

Embodiments of the present invention will be described below in detail with reference to the drawings. (Embodiment 1) In Embodiment 1, a case will be described in which the image binarizing apparatus of the present invention is applied to a digital camera. FIG. 1 shows an example of an apparatus configuration from input of image data to recording of a binarized image (binary data) when the image binarizing apparatus of the present invention is applied to a digital camera. It is a block diagram. In the following embodiments including this embodiment, an example in which the binarization threshold is set mainly based on the luminance value will be described. However, the binarization threshold may be set based on the density value. The case where the binarization threshold is set based on the density value will be described as appropriate.

The digital camera 100 includes a CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU 108, block buffer 109, average brightness value calculator 120, low brightness threshold value setting device 121, and binarization threshold value setting circuit 122, which will be described later, binarization device 123, compressor 124, image storage memory 125, , Consisting of

The CCD 101 converts the light condensed by the optical system (not shown) of the digital camera 100 into an electric signal, and converts the R of each pixel constituting a multi-valued image as image data.
Outputs a GB analog signal. The output analog signal is converted into a digital signal by the A / D converter 102.
The white balance of the digital signal is adjusted by the white balance adjuster 103. For the image data whose white balance has been adjusted, the pixel interpolator 104 interpolates an R, G or B signal having no information at each pixel. Hereinafter, R, G, and B represent red, green, blue, or red signal values, green signal values, and blue signal values, respectively.

Here, the relationship between the filter of the CCD 101 and the interpolation of the pixel interpolator 104 will be described. FIG.
FIG. 3 is a conceptual diagram illustrating a concept of a light receiving unit of the CCD 101. The R, G, and B filters are applied to the light receiving unit in a fixed pattern, and a difference between colors is identified by the filters. It should be noted that, in general, pixels of green G, which have high sensitivity to human eyes, are arranged more than filters of other colors. The subscript is used as an identifier for identifying the position (filter number). In the drawing, filter numbers are assigned only to the central part.

In FIG. 2, the red R at the position G0
Interpolation signal value R (G0) and blue B interpolation signal value B
(G0) is calculated as in the following equation (1). R (G0) = (R0 + R2) / 2 B (G0) = (B0 + B1) / 2 {(1)

Further, the interpolation signal values G (R0), B (R0), G (B0), and R of the green G at the positions of R0 and B0.
(B0) is calculated as in the following equation (2). G (R0) = (G0 + G1 + G2 + G5) / 4 B (R0) = (B0 + B1 + B4 + B5) / 4 G (B0) = (G0 + G1 + G3 + G6) / 4 R (B0) = (R0 + R2 + R4 + R5) / 4 ‥‥ (2) Pixel interpolator 104 The above interpolation is performed at all pixel positions, and the interpolated RGB signal values are output for each pixel.

The luminance generator 105 generates a luminance signal value Y from each interpolated pixel by the following equation (3). Y = 0.34R + 0.55G + 0.11B (3) A multiplier and an adder are required to calculate a luminance signal value which is a digital value using Expression (3). By approximating by equation (4), the luminance generator 105
Can be configured with only an adder. Y = (2/8) R + (5/8) G + (1/8) B (4) Therefore, when calculating the luminance signal value using the approximate expression (4), the luminance value is calculated using a simple circuit configuration. Y can be calculated, and a digital camera excellent in circuit cost, calculation speed, and power consumption can be provided.

When the binarization threshold is set based on the density value, an inverted value of the luminance signal (that is, the number of gradation ranges-luminance value) can be adopted. For example, A /
When the D converter 102 outputs 256 values (0 to 255) and the luminance signal value is 200, the density value is 55.
(255-200). The density value may be separately calculated from the RGB signals without using such an inverted value.

The luminance signal value Y is calculated by the aperture corrector 106
Thereby, the high-frequency portion of the image data is emphasized. The aperture correction is performed by using a well-known 5 × 5 size high-frequency emphasis filter. The high-frequency emphasized luminance value signal is temporarily stored in the frame memory 107.

The CPU 108 calculates a block size and a sampling interval (sampling cycle), and controls a block buffer 109, an average luminance value calculator 120, and other circuits and parts of the digital camera 100 described below. . Since the image size of the multi-valued image is determined by the size of the CCD (the number of pixels), the CPU 108 determines the block size necessary for binarizing the multi-valued image based on the image size and the sampling interval (sampling period). calculate.

The unevenness of the luminance value due to the optical system such as the lens of the digital camera 100 differs depending on the position and the intensity of the light source, but generally, there is a tendency that the image is bright near the center of the image and darker toward the periphery. Therefore, the CPU 108 calculates the image division pattern in consideration of the vignetting of the optical system. Note that a division pattern can be selected from a certain block division pattern set in a storage unit (not shown) or the like.

FIG. 3 is a diagram showing an example of division of a multivalued image into blocks. In addition to the usual square division,
(A) shows an example in which a multivalued image is divided by a combination of a square, a rectangle, and a triangle with the center of the image (the center of the lens) being point-symmetric. On the other hand, FIG. 3B shows an example in which a multivalued image is divided from the center of the image based on concentric circles. By dividing the block in consideration of the optical system, the brightness in the block becomes more uniform, and the binarization threshold is set for each block as described later. Is possible.

In the following, for the sake of simplicity, a block divided into squares will be used. However, if the average luminance value of a certain block is abnormally high compared to that of an adjacent block, whiteout occurs. Since there is a possibility that the blocks may be divided, the blocks including the adjacent blocks may be subdivided, and the area where the whiteout occurs may be appropriately divided to reconstruct the blocks. By dividing an image into local pixel sets using the above blocks, a high-quality binary image can be obtained.

The block buffer 109 includes a CPU 108
The image is read from the frame memory 107 in block units based on the block division pattern determined by
Remember temporarily. Note that the block stored in the block buffer 109 is referred to as a processing target block.

The average luminance value calculator 120 samples pixels from the image stored in the block buffer 109 at a preset sampling period, and calculates an average luminance value. FIG. 4 is a diagram illustrating an example of a sampling interval for sampling pixels in a block. FIG. 4 (a)
Shows that the CPU 108 sets the block size to 64 × 64 pixels and sets the sampling period to 2 for an image having an image size of 1280 × 960 pixels (in the figure, only 9 × 8 pixels in one block). Is displayed). On the other hand, FIG. 4B shows that the CPU 108
A state in which the block size is set to 128 × 128 pixels and the sampling period is set to 4 for an image of 60 × 1920 pixels is shown (only 9 × 8 pixels in one block are shown in the figure).

The CPU 108 can set the total number of blocks or the sampling interval within a block to a binarized image size in accordance with the processing capacity in consideration of power consumption. Therefore, even if the image size is large (even if the number of all pixels is large), the number of samplings can be kept constant, the processing time until the determination of the binarization threshold is reduced, and the binary Conversion processing becomes possible. Note that the CPU 108 may set a sampling interval for each block. In addition, a block in which whiteout occurs may be finely resampled.

FIG. 5 is a block diagram showing an example of the configuration of the average luminance value calculator 120. The luminance value (v) of the pixel sampled by the CPU 108 is output to a comparator 501 by a low luminance threshold (low luminance threshold th) described later.
1 (i, j) ((i, j) is an index indicating a block number, and this block is B (i, j)). The comparator 501 outputs a signal value 1 when the luminance value v of the sampled pixel is larger than the low luminance threshold thl (i, j), and outputs a signal value 0 when the luminance value v is smaller.

When the signal value 1 is output, the gate 50
7 is opened and the luminance value v is input to the adder 506. Adder 5
In step 06, the value (sumv) of the addition result register 502 is added to the luminance value v of the input pixel, and the addition result register 502 stores a new addition result. On the other hand, the signal 1 of the comparator 501 is also transmitted to the counter 503, and counts the number of luminosity values v (referred to as num) passing through the gate 507.

When the above-described processing is expressed by an equation as an algorithm, the following equation (5) is obtained. if v> thl (i, j) then sumv = sumv + v num = num + 1 else sumv = sumv num = num (5)

Here, when the counter 503 is incremented as a result of the increment (state where the counter indicates a power of 2), the gate 508 is opened and the luminance value v held in the addition result register 502 is read. Sum (su
mv) is transmitted to the shift register 504 and the counter 50
3 shifts to the right by the bit position -1 at which "1" is set. After processing all pixels in the block,
The value stored in the shift register 504 is the average luminance value av
Output as e (i, j). That is, the average luminance value is calculated by the following equation (6). ave (i, j) = sumv '/ num' (6)

Here, num 'is the block B (i, j)
Is less than or equal to the number of samples sampled at
The largest value represented by a power of 2
v 'is the addition result register 5 when num' is counted.
02 represents the value held.

The low-luminance threshold setting unit 121 outputs the average luminance value ave (i−1) of the immediately preceding block (if the current block is B (i, j), for example, B (i−1, j)). ,
j) is multiplied by a predetermined coefficient to calculate a low luminance threshold thl (i, j) used by the average luminance value calculator 120. Assuming that the predetermined coefficient is Ca = 1/4 (one power of 2),
The low-luminance threshold setting unit 121 determines the lower two levels of ave (i, j).
No special circuit is required because the value is the value excluding the bit,
The circuit configuration becomes simple, and processing can be performed at high speed and with low power consumption. The calculation formula described above can be represented by the following formula (7). thl (i, j) = ave (i-1, j) * CaＣａ (7)

Here, when calculating the low luminance threshold thl, the average luminance value ave of only one adjacent block is used. However, depending on how the blocks are divided, the average of all adjacent (for example, up, down, left, and right) blocks is used. Luminance value ave
May be used.

The binarization threshold value setting circuit 122 calculates the average luminance value ave (i, j) calculated by the average luminance value calculator 120.
Is used for binarization of a multi-valued image,
(I, j) is set. FIG. 6 is an explanatory diagram showing one configuration example of the binarization threshold setting circuit. Binary threshold setting circuit 1
22 includes an addition / subtraction unit 601, a multiplication / division unit 602, a comparator 603, and a selector 604.

The addition / subtraction unit 601 subtracts a constant Cm (Cm is a positive value) from ave (i, j). Here, the subtraction is performed because Cm> 0, but the addition and subtraction unit 601 may add the constant Cm as a negative value depending on the use mode.

The multiplication / divider 602 multiplies ave (i, j) by a constant Cb (where Cb is a positive value and less than 1). Here, since 1>Cb> 0, the multiplication is performed. However, depending on the mode of use, the division may be performed by the multiplier / divider 602 with the constant Cb> 1.

The comparator 603 compares ave (i, j) with a predetermined value (comparison value) Ath, and outputs ave (i, j).
If j) is greater than Ath, selector 604 selects the result of addition / subtraction unit 601 and outputs TH (i, j).
Conversely, if ave (i, j) is smaller than Ath, selector 604 selects the result of multiplication / divider 602 and sets TH
(I, j) is output.

The comparison value Ath may be a constant or may be determined based on surrounding blocks. For example, 1.5 times the average of the luminance values of adjacent blocks B may be set as Ath. By determining Ath based on the luminance values of the surrounding blocks in this way, it is determined whether the luminance value of the block B of interest is prominent, that is, whether or not whiteout has occurred in the block B of interest. This makes it easy to determine whether or not the binarization is of high quality.

As described above, in the binarization threshold value setting circuit 122, TH is calculated by subtracting ave (i, j) using the constant Cm for an area that is overexposed. TH is calculated by multiplying Cb with respect to a normal area without overexposure. The reason for switching between these two processes is based on the change in the contrast of overexposed areas. That is, in a normal image, the contrast ratio between the character and the background is substantially constant regardless of the signal level. Therefore,
For a normal image, the binarization threshold value setting circuit 122 can be constituted only by the multiplier / divider 602.

On the other hand, an image in which whiteout occurs is characterized in that the ratio of the contrast between the character and the background is reduced and, at the same time, the average luminance value ave is increased. Therefore, in such a region, it is necessary to adopt or set a binarization threshold TH close to the average luminance value ave in order to appropriately binarize. However, simply multiplying ave by Cb will increase the difference between the binarization threshold TH and the average luminance value ave as the average luminance value ave increases. That is, the multiplier / divider 602
If only the surface reflection occurs, the image will not be correctly binarized.

In order to avoid this, the binarization threshold value setting circuit 122 of the first embodiment sets the binarization threshold value TH so as to cope with the case where there is surface reflection (that is, the case where the average luminance value ave is large). An adder / subtractor 601 is provided so that the difference between the average luminance values ave is relatively small.
The threshold value TH can be selected according to the magnitude of the threshold value.

When the attribute of the image signal is a density signal, if the average density value (denoted as ave for convenience) is larger than a predetermined comparison value (Ath for convenience), the addition / subtraction unit 601 is selected. . Here, ave is a predetermined coefficient C
m is added to determine a binarization threshold (TH for convenience). Conversely, if ave is smaller than Ath, the multiplier / divider 602 is selected, and the predetermined coefficient Cb (1>
The binarization threshold TH multiplied by Cb) is determined.

Note that the predetermined coefficient Cb is represented by Cb = x / 16,
Alternatively, if Cb = x / 8 (x is a default value representing a natural number not exceeding the denominator), the circuit configuration of the multiplier / divider 602 becomes simple, and processing can be performed at high speed and with low power consumption.

The binarizer 123 is provided in the block buffer 10
Each of the nine pixels is compared with a binarization threshold value TH and binarized. The binarized image is subjected to image compression suitable for a binary image such as MH or MR by the compressor 124. The compressed image is stored in the image storage memory 125.

Next, the flow of processing until a multi-valued image is binarized will be described. FIG. 7 is a flowchart showing a processing flow of image data until a multi-value image is binarized. The CPU 108 reads out the size of the multi-valued image stored in the frame memory 107 (step S70).
1). Note that the number of pixels or CCD set in advance in the digital camera 100, not from the frame memory 107,
The information from 101 may be used. Then, CPU1
08 sets the sampling period according to the image size (step S702).

Based on the image size read in step S701 and the sampling period set in step S702, the CPU 108 sets the block size, shape, and division pattern (step S703). Since the number of pixels output from the CCD 101 is usually constant or a rated number of pixels such as 640 × 480 pixels or 800 × 600 pixels specified by mode switching, the CPU 108 determines a sampling period and a block shape in advance. You can also select what you have saved.

Subsequently, one block B (i, j) is set from the plurality of divided blocks, and the image information is transferred to the block buffer 109 (step S704). Pixels (luminance values) are sampled from the block B (i, j) in accordance with the sampling period set in step S702 (step S705). Subsequently, the adjacent block B (i-
Based on the average luminance value ave (i−1, j) of (1, j), the low luminance threshold thl for the block B (i, j)
(I, j) is calculated, and a luminance value less than the low luminance threshold thl (i, j) is excluded from the luminance values of the sampled pixels (step S706). Excluding low luminance values in step S706 contributes to high-quality binarization processing.

The average luminance value ave (i, j) of the block B (i, j) is calculated using the luminance value from which the low luminance value has been excluded (step S707). This average luminance value ave
The process proceeds to a routine for calculating the binarization threshold value TH (i, j) for the block B (i, j) based on (i, j) (step S708). The contents of this routine will be described later. The binarization threshold T calculated in step S708
Using H (i, j), all pixels (g (x, j) of block B (i, j) stored in block buffer 109
y)) is binarized (step S709). Note that x and y are natural numbers indicating the position of each pixel in the block.

Finally, it is determined whether or not the binarization processing has been performed for all the blocks (step S710), and when the binarization of all the blocks has been completed (step S7).
10: YES), the process ends, and when the binarization of all blocks has not ended (step S710: NO)
Sets a block adjacent to B (i, j) (for example, B (i + 1, j)) and repeats steps S704 to S710.

Here, the binarization threshold value calculation routine of step S708 will be described. FIG. 8 is a flowchart illustrating an example of the processing flow of the binarization threshold value calculation routine. First, the comparator 603 outputs the average luminance value ave (i,
j) is compared with the comparison value Ath (step S801).
If the average luminance value ave (i, j) is larger than the comparison value Ath (step S801: YES), the addition / subtraction unit 601 is used.
And the binarization threshold TH (i, j) = ave (i, j) -Cm
Is calculated (step S802).

On the other hand, if the average luminance value ave (i, j) is At
If h is not larger than h (step S801: NO), the multiplication / division unit 602 sets the binarization threshold value TH (i, j) = ave.
(I, j) * Cb is calculated (step S803). Here, since it is assumed that the attribute of the image signal is a luminance signal, −Cm (Cm> 0) and Cb are less than 1. However, when the attribute of the image signal is a density signal, + Cm is set.
(Cm> 0), Cb is a value larger than 1.

In the above example, description has been made mainly using luminance values. However, physical quantities having a strong correlation with luminance values can be handled in the same manner as luminance values. Such physical quantities include, for example, light quantity, photometric value, exposure value, lightness, saturation, and the like.

In the first embodiment, an example in which the image binarizing device of the present invention is applied to a digital camera has been described. This digital camera performs a normal binarization threshold setting process in an area where there is no overexposure, and in a high luminance area where overexposure occurs, the difference from the average luminance value becomes relatively smaller as the luminance increases. Since such a binarization threshold is set, it is possible to binarize an image with high quality. In addition, the processing of the average luminance value calculator used when calculating the binarization threshold value TH can be performed at high speed and with low power consumption.
Digital cameras can perform high-quality binarization processing with high speed and low power consumption.

(Embodiment 2) In Embodiment 2, the CM
A case where an image binarizing device using an OS sensor is applied to a digital camera will be described. FIG. 9 is a block diagram showing an example of a device configuration from the input of an image to the recording of a binarized image in a digital camera using a CMOS sensor for an image input portion. In the present embodiment, the same parts as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted. The different parts from the first embodiment will mainly be described.

The digital camera 900 has a CM
An OS sensor 901 is provided. Therefore, unlike the CCD 101 (see FIG. 1) that can only perform raster scanning,
Since the sensor 901 can perform random access and can perform reading in block units, the frame memory 107 becomes unnecessary, and the circuit configuration is simplified. Furthermore, CCD1
01 is a CMOS which forms another circuit of the digital camera 900.
In addition to requiring a separate power supply from the integrated circuit, CMOS
The sensor 901 can use the same power supply as that of the CMOS integrated circuit, and consumes less power. Therefore, the digital camera 9
Since the circuit scale as 00 is small, the power consumption, processing speed, cost, and the like are more convenient than a system using a CCD. In this embodiment, the CMOS
Although a sensor is used, an image binarizing device having an image input unit capable of accessing other blocks may be used.

(Embodiment 3) In Embodiment 3, a binarization threshold applied to each pixel in a predetermined block is calculated.
A case will be described in which an image binarization device that binarizes image data for each pixel is applied to a digital camera. FIG.
FIG. 2 is a block diagram showing an example of a device configuration of a digital camera that calculates a binarization threshold applied to each pixel and binarizes image data for each pixel. In the present embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and description thereof will be omitted, and portions different from the first embodiment will be mainly described.

The digital camera 1000 receives the average luminance value output from the average luminance value calculator 120 and outputs a binary threshold value (hereinafter, referred to as a block binary threshold value) applied to each block. A threshold value setting circuit 1001 and a memory 1 for storing a block threshold value
002, and a binarization threshold interpolator 1003 for setting a binarization threshold to be applied to each pixel in a predetermined block based on the block binarization threshold. A digital camera 100 according to the first aspect,
It has a similar configuration. Note that a portion for generating a color difference signal is omitted.

FIG. 11 shows a block binarization threshold setting circuit 1.
001 is an explanatory diagram showing one configuration example. FIG. The block binarization threshold setting circuit 1001 includes a multiplication / divider 1011 and a coefficient memory 1012. The block binarization threshold setting circuit 1001 outputs the block binarization threshold of the processing target block to the memory 1002.

The multiplier / divider 1011 calculates the average luminance value ave
(I, j) is multiplied by a coefficient Cb. Coefficient memory 1012
Stores the coefficient used in the multiplication / division unit 1011. The value of the coefficient Cb differs according to the magnitude of the average luminance value ave. FIG. 12A is an explanatory diagram showing an example of coefficients stored in the coefficient memory 1012. As illustrated, Cb increases so as to approach 1 as the average luminance value ave increases.

By using such a coefficient Cb, the block binarization threshold value TH in the high-luminance area is set to the average luminance value av
e can be prevented from greatly deviating from e, and high-quality binarization can be performed even in an area where whiteout occurs. Here, since 1>Cb> 0, the multiplication is performed. However, depending on the use mode, the division may be performed by the multiplication / divider 1011 with the constant Cb being> 1.

When setting the block binarization threshold,
When a density value is used, a coefficient Cb table as shown in FIG. 12B may be used. As shown in the figure, the coefficient Cb has a larger value of 1 or more as the density value (ave for convenience) is smaller. By using this coefficient Cb, the block binarization threshold value TH in the low density region is obtained.
Can be prevented from greatly deviating from the average density value ave, and high-quality binarization can be performed even in an area where whiteout occurs.

The block binarization threshold setting circuit 100
In No. 1, the addition / subtraction unit is not provided, so that the circuit configuration is simple. However, depending on the use mode, the binarization threshold setting circuit 122 (shown in Embodiment 1 as the block binarization threshold setting circuit 1001) 1 (see FIG. 1). Further, as in an embodiment to be described later (for example, Embodiment 7), a configuration may be adopted in which only an addition / subtraction unit is used as the block binarization threshold value setting circuit 1001 and no multiplication / division unit is used.

The memory 1002 sequentially stores all block binarization thresholds output from the block binarization threshold setting circuit 1001.

The binary threshold value interpolator 1003 is
A binarization threshold to be applied to each pixel in a predetermined area is set using the block binarization thresholds of all the blocks stored in 02. In the following description, an interpolation block refers to a predetermined region in which a binarization threshold is set for each of its constituent pixels. Here, the outline of the calculation of the binarization threshold applied to each pixel in the interpolation block will be described. FIG.
It is an explanatory view for explaining an outline of calculating a binarization threshold applied to each pixel in an interpolation block, and FIG. 12A is a diagram showing a relationship between a processing target block and an interpolation block;
FIG. 7B is an explanatory diagram for calculating a binarization threshold value applied to each pixel in the interpolation block.

As is clear from FIG. 13A, the interpolation block BH is composed of four adjacent processing target blocks Ba,
It spans Bb, Bc and Bd. Let the block binarization thresholds of the processing target blocks Ba, Bb, Bc, Bd be a, b, c, d, respectively. Binarization threshold interpolator 100
3 calculates a binarization threshold applied to each pixel in the interpolation block BH using the block binarization thresholds a, b, c, and d.

Referring to FIG. 13B, interpolation block B
A method of calculating a binarization threshold applied to the pixel bp in H will be described. Assuming that the interpolation block BH is a rectangle, its size (the number of pixels) is xbnum in the horizontal direction and y in the vertical direction.
bnum. Further, the position of the pixel bp is (m, l). At this time, the value a at the (0,0) point, the value b at the (xbnum, 0) point, and (0, ybnum) of the interpolation block BH.
It is assumed that there is a value c at the point and a value d at the (xbnum, ybnum) point, and the interval is linearly approximated.

First, assuming that the provisional threshold value on the left boundary line of the interpolation block BH is leftth and the provisional threshold value on the right boundary line is rightth, the following equation (9) leftth = (a (ybnum−1) + cl) ) / Ybnum rightth = (b (ybnum−1) + dl) / ybnum ‥‥ (9) Next, at the (0, l) point, the value leftth,
Considering that the value rightth is at the (xbnum, l) point, the binarization threshold th (m, l) applied to the pixel bp is linearly approximated as in the following equation (10). th (m, l) = (left (xbnum-m) + rightth × m) / xbnum {(10)

When the interpolation block BH is located at the end of the entire image, the number of adjacent blocks to be processed is two or one. In this case, the obtained block binarization threshold value is not sufficient. Equations (9) and (10) can be used by substituting and using the block binarization threshold. For example, at the upper left corner of the image, the obtained block binarization threshold is only d, and the block binarization threshold corresponding to a, b, and c cannot be obtained. Therefore, in this case, a, b,
Substitute the value of c with d.

In the binarizer 123, the frame memory 10
7 or the luminance value of the pixel bp in the interpolation block from the block buffer 109 and the binarization threshold value interpolator 1003
Is compared with the binarization threshold value for the pixel bp calculated in
The luminance value of the interpolation block BH is binarized. The binarized image data is supplied to the compressor 124 as in the first embodiment.
Then, image compression suitable for binarized images such as MH and MMR is performed.

Next, a description will be given of a flow of processing until a multi-value image is binarized in the present embodiment. FIG.
FIG. 4 is a flowchart showing a processing flow of image data until a multi-value image is binarized. The CPU 108
The size of the multi-valued image stored in the frame memory 107 is read (step S1401). Instead of using the frame memory 107, the number of pixels preset in the digital camera 1000 or information from the CCD 101 may be used. Subsequently, the CPU 108 sets a sampling period according to the image size (step S14).
02).

Based on the image size read in step S1401 and the sampling period set in step S1402, the CPU 108 sets the block size, shape, and division pattern (step S1403). CCD101
Since the number of pixels output in is normally constant or the rated number of pixels such as 640 × 480 pixels or 800 × 600 pixels specified by mode switching, the CPU 108 determines the sampling period and block shape in advance. You can also choose what you put.

Subsequently, one block B (i, j) is set from the plurality of divided blocks, and the image information is transferred to the block buffer 109 (step S1404).
Step S1402 from block B (i, j)
Pixel (brightness value) according to the sampling cycle set in
Is sampled (step S1405). continue,
The adjacent block B calculated in the previous routine
Based on the average luminance value ave (i−1, j) of (i−1, j), the low luminance threshold t for the block B (i, j)
hl (i, j) is calculated, and a luminance value less than the low luminance threshold thl (i, j) is excluded from the luminance values of the sampled pixels (step S1406). Step S140
Excluding low luminance values in 6 contributes to high quality binarization processing.

The average luminance value ave (i, j) of the block B (i, j) is calculated using the luminance value from which the low luminance value has been excluded (step S1407). This average luminance value ave
The process proceeds to a routine for calculating a block binarization threshold value TH (i, j) of the block B (i, j) based on (i, j) (step S1408).

FIG. 15 is a flowchart showing an example of the processing flow of the block binarization threshold value calculation routine. The block binarization threshold setting circuit 1001 first determines a coefficient Cb to be used according to the value of the input average luminance value ave (i, j) (step S1501). Subsequently, the multiplication / divider 1011 multiplies the average luminance value ave (i, j) by the coefficient determined in step S1501, and outputs a block binarization threshold TH (i, j) (step S150).
2).

Returning to FIG. 14, the block binarization threshold value TH calculated in step S1502 is stored in the memory 1002 (step S1409). It is determined whether the block binarization threshold has been calculated for all blocks (step S1410). If the calculation has not been completed (step S1410: NO), a block adjacent to B (i, j) is set. (For example, B (i + 1, j)), and repeat steps S1404 to S1410.

When the block binarization threshold has been calculated for all the blocks (step S1410: YE)
S), an interpolation block is set (step S141)
1). This interpolation block is suitable for partially performing high-quality binarization or the like, and may be set in advance by the user, or may be a mode in which an image center portion is set by appropriate mode switching. You may. Binarization threshold interpolator 100
3 is a pixel-based binarization threshold th in the interpolation block using the block binarization threshold TH of the processing target block that spans the interpolation block set in step S1411.
(X, y) is set (step S1412). In addition,
The function f in FIG. 14 conceptually shows Expressions (9) and (10), and x and y are natural numbers representing the positions of the respective pixels of the image.

The pixels read out from the frame memory 107 by the binarizer 123 (g (x, y)) are binarized using the binarization threshold calculated in step S1412 (step S1413). Next, it is determined whether or not the binarization processing has been performed on all the pixels (step S141).
4) If binarization of all pixels is not completed (step S1414: NO), a pixel adjacent to g (x, y) is set (for example, g (x + 1, y)), and step S1412 is performed. To step S1414 are repeated.
If binarization has been completed for all pixels (step S1414: YES), it is determined whether or not binarization has been completed for all interpolation blocks (step S141).
5) If not completed (step S1415: N)
O): Steps S1411 to S1415 are repeated, and if the processing is completed (step S1415: Y
ES), and terminate the process.

The digital camera according to the third embodiment performs block division corresponding to the image size and the optical system.
A luminance value is sampled and extracted from the block, a low luminance threshold is set in consideration of surrounding blocks, a block binarization threshold is performed based on the threshold, and further, interpolation is performed using the block binarization threshold. Since the binarization is set by setting the binarization threshold applied to each pixel in the block,
Higher quality image processing can be performed than the digital camera according to the first embodiment. In particular, when individual characters are small, such as when photographing a bulletin board, or when characters in a part of the region are small, it is possible to partially binarize the portion with high quality.

The digital camera according to the third embodiment is
Since the block binarization threshold value setting circuit is composed of only the multiplier / divider, the circuit configuration is simple and high-speed binarization is possible. In addition, by adjusting the denominator and the numerator with the coefficient Cb using a power of 2, a circuit can be constructed using only the shift register, and a high-speed and power-saving digital camera can be provided.

(Embodiment 4) In Embodiment 4, a binarization threshold value applied to each pixel in a predetermined block is calculated.
A case in which an image binarization device using a CMOS sensor for binarizing image data for each pixel is applied to a digital camera will be described. FIG. 16 shows a CMOS image input portion.
FIG. 2 is a block diagram illustrating an example of a device configuration of a digital camera that uses a sensor to calculate a binarization threshold applied to each pixel and binarizes image data for each pixel. In the present embodiment, the same components as those in the third embodiment will be denoted by the same reference numerals, and detailed description thereof will be omitted. The different portions from the third embodiment will be mainly described.

The digital camera 1600 has a C
It has a MOS sensor 1601. Therefore, unlike the CCD 101 (see FIG. 10) that can only perform raster scanning,
The MOS sensor 1601 is capable of random access, is capable of reading in block units, reads pixel values for each block, calculates a block binarization threshold value,
2 can be stored in the block buffer 10
9 (see FIG. 10) becomes unnecessary, and the circuit configuration is simplified.

At the same time, the pixel values read for each block are converted into luminance signals, subjected to aperture correction, and stored in the frame memory 107. Further, the CCD 101 requires a separate power supply from a CMOS integrated circuit that forms another circuit of the digital camera 1000, while the CMOS sensor 16
01 can use the same power supply as the CMOS integrated circuit and consumes less power. Therefore, the circuit size of the digital camera 1600 is also reduced, so that power consumption, processing speed,
In terms of cost and the like, it is more convenient than a system using a CCD. Although the CMOS sensor is used in the present embodiment, an image binarizing device having an image input unit that can access other blocks may be used.

(Fifth Embodiment) In a fifth embodiment, a case will be described in which an image binarization device that limits the average luminance value to a value within a predetermined range is applied to a digital camera. FIG. 17 is a block diagram showing an example of a device configuration of a digital camera that limits the range of the average luminance value so as to fall within a value within a predetermined range. Note that, in the present embodiment, the same components as those in Embodiment 1 are denoted by the same reference numerals, and description thereof will be omitted.

The digital camera 1700 has a limiter 1701 that limits the average luminance value output from the average luminance value calculator 120 by a preset lower limit value to prevent the average luminance value that is too low from being output. For example, consider a relatively bright character in a dark image, more specifically, an image of black ink drawn on a blackboard. Although the area of the black ink characters is very small compared to the area of the blackboard, the important information in this case is the drawn image of the black ink. Here, if normal binarization is performed, image data of the blackboard other than the black and white drawing image becomes dominant, so that the binarization threshold value is set at the boundary of shading of the black portion of the blackboard. Therefore, the black ink image is naturally determined to be white, but the black ink residue (for example, the locus of blackboard eraser) is also determined to be white.
As a result, noise increases.

The limiter 1701 adjusts the average luminance value so that such noise is not reproduced. Specifically, it is determined whether or not there is an average luminance value within a predetermined range. If not, the average value is replaced with the lower limit of the range width. This limiter 1701 sets the lower limit of the average luminance value, and it is possible to obtain a binarized image with high contrast.

Depending on the mode of use, it is possible to limit the average luminance value to a preset upper limit value so that an excessively high average luminance value is not output. For example, it can be applied to an image marked on a whiteboard. In this case, since the average luminance value becomes high, contrary to the above-described example, the limiter 1701 prevents the excessively high average luminance value from being output.

In this embodiment, the CCD 1
01 was used, but depending on the mode of use, the second embodiment
As shown in, a CMOS sensor 901 (see FIG. 9) may be used.

(Embodiment 6) In Embodiment 6, the binarization threshold value applied to each pixel in a predetermined block is calculated while limiting the average luminance value to a value within a predetermined range. A case will be described in which an image binarization device that binarizes image data for each pixel is applied to a digital camera. FIG. 18 is a block diagram illustrating an example of a configuration of a digital camera according to the present embodiment. In the present embodiment, the same components as those in the fifth embodiment are denoted by the same reference numerals, and description thereof will be omitted.

The digital camera 1800 has a limiter 1801 that limits the average luminance value output from the average luminance value calculator 120 by a preset lower limit value to prevent the output of the average luminance value that is too low. The average luminance value output from limiter 1801 is within a predetermined range. For example, the present invention can be applied to relatively bright characters in a dark image, more specifically, to a black ink image drawn on a blackboard.

In particular, in the present embodiment, the binarization threshold value interpolator 1003 calculates the binarization threshold value applied to each pixel in the interpolation block from the block binarization threshold value of the surrounding processing target block. At this time, the limiter 1801
Since the prominent average luminance value is excluded, individual binarization thresholds can be set more appropriately, thereby enabling higher quality binarization processing. Depending on the mode of use, the average luminance value may be limited to a preset upper limit value so that an excessively high average luminance value is not output even though whiteout does not occur. In addition, the limiter 1801
May set a predetermined range from a block binarization threshold.

In this embodiment, the CCD 1
01 was used, but the fourth embodiment may be used depending on the mode of use.
As shown in, a CMOS sensor 1601 (see FIG. 16) may be used.

(Embodiment 7) In Embodiment 7, an image pickup apparatus having photometric means will be described. FIG.
FIG. 2 is a block diagram showing an example of a device configuration from input of an image to recording of an image subjected to a binarization process in a case where an image pickup device including a photometer is applied to a digital camera. In the present embodiment, since the components of the first embodiment are similar, the same components as those of the first embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted. The different parts will be mainly described.

The digital camera 1900 includes the CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU 108, photometer 1901, smoother 1902
, A memory 1903, and a block read controller 1904
, A binarization threshold setting circuit 1905, and a binarizer 123
, A compressor 124, and an image storage memory 125.

The CCD 101 is a digital camera 1900
Is a part that converts light condensed by an optical system (not shown) into an electric signal.
It is a part that outputs a B analog signal. The output analog signal is converted into a digital signal by the A / D converter 102. The digital signal is sent to the white balance adjuster 10
3. Through the pixel interpolator 104, the luminance generator 105, and the aperture corrector 106, processing such as interpolation and extraction of luminance values is performed, and is temporarily stored in the frame memory 107.

The CPU 108 has a photometer 1901 to be described later.
Based on the luminance information (luminance value) from, a block reading controller 1904 described later is controlled to divide the image stored in the frame memory 107 in the same manner as the screen division used by the photometer 1901 for photometry. The screen division used by the photometer 1901 for photometry may be a fixed one,
8, the division may be performed as shown in FIG. The CPU 108 controls the other digital camera 1
Each circuit 900 is controlled.

The photometer 1901 has an automatic exposure detection mechanism (AE) for performing photometry of a subject before taking an image, and measures the brightness of each screen based on a digital signal output from the A / D converter 102. I do. The photometry is performed, for example, by adding the luminance values of the pixels. At this time, the CPU 108
It is also possible to sample the photometric value used for the addition by the photometer 1901. Note that the amount of light photoelectrically converted by the CCD 101 may be used depending on the mode of use.

The smoother 1902 smoothes the photometric value of each screen obtained by the photometer 1901 and outputs it to the memory 1903 as ave (i, j). As an example of smoothing,
The following processing is included. That is, the average value ave (i, j) of the photometric values of all pixels included in one screen (G (i, j)) in the photometer 1901 (or the screen G
When the average value of the photometric values sampled from (i, j)) is a value protruding from the average value of the photometric values of the surrounding screen, the photometric value of each pixel of the screen G (i, j) is calculated. The correction is performed so that the value does not become prominent compared with the average value of the photometric values of the surrounding screen.

An example of an algorithm for realizing this processing is shown below. The average photometry value of four screens adjacent to the screen G (i, j) is set as ave4 (i, j), and the screen G
The average photometric value ave (i, j) of (i, j) is ave4
If the value is three times or more than (i, j), the photometric value (s (x, y)) of each pixel in the screen G (i, j) is expressed by the following equation (1).
Convert as in 1). if ave (i, j) ≧ 3 * ave4 (i, j) thens (i, j) = ave4 (i, j) + (1/4) * (s (i, j) −ave4 (i, j) )) ‥‥ (11)

The smoothing unit 1902 uses the converted photometric values to calculate the average photometric value ave (i, j) of the screen G (i, j).
j) is calculated again, and the average photometric value is output to the memory 1903. The memory 1903 stores the average photometric value.

The block reading controller 1904 is connected to the photometer 1
An image 901 divides an image stored in the frame memory 107 in the same manner as the screen division used for photometry. This allows
The threshold value by binarization becomes more natural.

The binarization threshold value setting circuit 1905 uses ave
The binarization threshold TH (i, j) is set based on (i, j). FIG. 20 is an explanatory diagram showing an example of the configuration of the binarization threshold setting circuit 1905. Binarization threshold setting circuit 1905
Has an adder / subtracter 2001 and a coefficient memory 2002. The binarization threshold setting circuit 1905 outputs the binarization threshold of the processing target block to the binarizer 123.

The adder / subtracter 2001 calculates the average photometric value ave
The coefficient Cm is subtracted from (i, j). Coefficient memory 200
2 stores a coefficient used in the addition / subtraction unit 2001.
The value of the coefficient Cm varies depending on the magnitude of the average photometric value ave. FIG. 21A shows a coefficient memory 2002.
FIG. 4 is an explanatory diagram showing an example of coefficients stored in the. As illustrated, Cm has a small value when the average photometric value ave is the largest in the middle range and the photometric value is very high or the photometric value is very small.

By using such a coefficient Cm, it is possible to prevent the binarization threshold value TH in a region where the photometric value is large (high-luminance region) from being largely deviated from the average photometric value ave. Even in such an area, high-quality binarization is possible. In the illustrated example, as described above, the coefficient Cm is larger than the average photometric value of the halftone. Most of the photometric values are intermediate values, and are set in consideration of performing binarization with high quality in this region. In addition, although the subtraction is performed with Cm being positive here, addition may be performed by the addition / subtraction unit 2001 with Cm being a negative value depending on the use mode.

In this embodiment, the smoothing unit 190
2, the photometric value in the overexposed area where the photometric value protrudes is smoothed. Therefore, there is a possibility that the difference between the binarization threshold value for the overexposed area and the photometric value before smoothing becomes large. Therefore, Cm when the photometric value is smoothed may be subtracted to 0, and in some cases, may be subtracted as a negative value (see FIG. 21). In such a case where the conversion of the coefficient between positive and negative occurs, the use of a multiplier / divider complicates the processing algorithm. In this regard, in this embodiment, the addition / subtraction unit 2001 having a simple processing algorithm is employed, and it can be said that convenience is high. When Cm is set to 0, the smoothed value can be directly used as the binarization threshold TH, so that the circuit configuration is simplified, and the digital camera 1900 with high speed and low power consumption can be obtained.

When the density value is used for setting the binarization threshold, the coefficient Cm as shown in FIG.
What is necessary is just to add using a table. As shown in the figure, the coefficient Cm table has a smaller coefficient Cm than the coefficient Cm for the intermediate area in an area where the density value (ave for convenience) is small. By using the coefficient Cm, the binarization threshold value TH in the low-density region is set to the average density value ave.
, And high-quality binarization is possible even in an area where whiteout occurs.

In this example, the binarization threshold setting circuit 19
05 has an addition / subtraction unit 2001 and does not include a multiplication / division unit. However, depending on a use mode, the block binarization threshold value setting circuit 100 includes a multiplication / subtraction unit and does not include an addition / subtraction unit.
1 (see FIG. 11), or a binarization threshold setting circuit 122 (see FIG. 6) provided with any of the arithmetic units.

The binarizer 123 outputs a binarization threshold TH (i,
Based on j), each pixel of the block B (i, j) corresponding to the screen G (i, j) is binarized. The smoothed block is binarized by appropriately processing the block using the photometric value after conversion. Compressor 124 is M
Compression suitable for a binary image is performed by H, MR, or the like. The image storage memory 125 stores the compressed image.

Next, the flow of image data until a multi-valued image is binarized will be described. FIG. 22 is a flowchart showing a processing flow of image data until a multi-value image is binarized. First, the CPU reads the image size from the CCD 101 (step S2201). Subsequently, the CPU 108 sets one screen G (i, j) from the photometric values output from the photometer 1901 (step S2202), and calculates an average of the photometric values of the screen (step S2203). When obtaining the average, all photometric values may be used, or sampling may be performed as appropriate.

Next, it is determined whether or not the average value of the photometric values of the screens protrudes from the average value of the photometric values of the adjacent screens (step S2204). If it is protruding (step S2204: YES), the average of the photometric values is smoothed by the smoother 1902 (step S2205). If it has not protruded (step S2204: NO) or if the smoothing process has been performed in step S2205,
The block reading controller 1904 reads the block B (i, j) corresponding to the screen G (i, j) from the frame memory 107 (step S2206), and binarizes based on the photometric value of the screen G (i, j). The process proceeds to a routine for calculating a threshold value (step S2207).

Here, the binarization threshold value calculation routine in step S2207 will be described. FIG. 23 is a flowchart illustrating an example of the processing flow of the binarization threshold value calculation routine. First, the binarization threshold setting circuit 1905 determines a coefficient Cm to be used according to the value of the input average photometric value ave (i, j) (step S2301). continue,
The addition / subtraction unit 2001 subtracts the coefficient determined in step S2301 from the average luminance value ave (i, j) and outputs a binarization threshold TH (i, j) (step S2302).

Returning to FIG. 22, the multi-valued image in the block is binarized using the binarization threshold calculated in step S2207 (step S2208). It is determined whether binarization has been completed for all blocks (step S22).
09), if completed (step S2209: YE)
S), the binarization process ends, and if not (step S2209: NO), the process proceeds to step S2202, and the next screen (for example, G (i + 1, j)) is set.
Thereafter, steps S2202 to S2209 are repeated.

In the seventh embodiment, since the binarization threshold is set using information (photometric value) obtained from the automatic photometer provided in the digital camera, a process for separately setting the binarization threshold is performed. Is unnecessary. Therefore, the circuit configuration is simplified, the cost can be reduced, and the algorithm for calculating the binarization threshold value is simplified, and the power consumption can be reduced. Also, since the binarization threshold setting circuit subtracts a predetermined coefficient according to the average photometric value, it is possible to binarize a multi-valued image with high quality even if the image has a pinpoint reflection by a light source. Becomes possible.

(Eighth Embodiment) In the eighth embodiment, the CM
An image pickup device using an OS sensor will be described.
FIG. 24 is a block diagram showing an example of a device configuration up to binarizing an input image and recording the input image in a digital camera using a CMOS sensor in an image input portion. In the present embodiment, the same components as those in the seventh embodiment are denoted by the same reference numerals, and the description thereof will be omitted. The different components from the seventh embodiment will be described.

The digital camera 2400 has a C
It has a MOS sensor 2401. Therefore, unlike the CCD 101 (see FIG. 19) which can only perform raster scanning,
Since the MOS sensor 2401 can perform random access and can perform reading in units of blocks, the frame memory 107 and the block reading controller 1904 become unnecessary, and the circuit configuration is simplified. In this case, the CPU 108
Have the role of a block read controller.

Further, while the CCD 101 requires a power supply different from that of the CMOS integrated circuit, the CMOS sensor 2
The power supply 401 can use the same power supply as that of the CMOS integrated circuit, and consumes less power. Therefore, the circuit scale of the digital camera 2400 is reduced, and thus the power consumption, processing speed, cost, and the like are more convenient than a system using a CCD. Although the CMOS sensor is used in the present embodiment, an image capturing apparatus having an image input unit capable of accessing other blocks may be used.

(Embodiment 9) Embodiment 9 has photometric means, calculates a binarization threshold applied to each pixel in a predetermined block, and binarizes image data for each pixel. A case where the image pickup apparatus is applied to a digital camera will be described. FIG. 25 is a configuration diagram illustrating an example in which an image capturing apparatus that includes a photometer and sets a binarization threshold value for each pixel is applied to a digital camera. In the present embodiment,
Since the components of the seventh embodiment are similar,
The same components as those described above are denoted by the same reference numerals, and description thereof will be omitted, and portions different from the seventh embodiment will be described.

The digital camera 2500 includes the smoothing unit 19
Screen binarization threshold setting for inputting an average value of photometric values outputted from the memory 02 and stored in the memory 1903 and outputting a value multiplied by a predetermined coefficient for each screen (hereinafter referred to as a screen binarization threshold) A circuit 2501, a memory 2502 for storing a screen binarization threshold, and a binarization threshold interpolator 2503 for setting a binarization threshold to be applied to each pixel in a predetermined block based on the screen binarization threshold And. Note that a portion for generating a color difference signal is omitted.

The screen binarization threshold value setting circuit 2501 multiplies the smoothed photometric value by a coefficient corresponding to the screen. Hereinafter, for convenience of explanation, it is assumed that the screen having the smoothed photometric value a is displayed as Ga, and the predetermined coefficient is displayed as Cb (Ga). Screen binarization threshold setting circuit 2501
Calculates the screen binarization threshold for all screens,
The data is sequentially stored in the memory 2502. The screen binarization threshold setting circuit 2501 has, for example, a circuit configuration similar to that of the block binarization threshold setting circuit 1001 shown in FIG. 11, and stores Cb (Ga) in a coefficient memory.

The binary threshold interpolator 2503 is
02, using the screen binarization thresholds of all screens stored in
A binarization threshold applied to each pixel of the interpolation block BH is set. In the following description, a binarization threshold applied to each pixel is set. The interpolation block BH is set by the block read controller 1904. The setting may be performed by the CPU 108 depending on the mode of use.

Here, the outline of the calculation of the binarization threshold applied to each pixel in the interpolation block will be described. FIG. 26 is a diagram showing the relationship between the interpolation block and the screen. As is clear from the figure, the interpolation block BH extends over four adjacent screens Ga, Gb, Gc and Gd. The average values of the smoothed photometric values of the screens Ga, Gb, Gc, and Gd are a, b, c, and d, respectively. The binarization threshold interpolator 2503 calculates the average value a of the smoothed photometric values,
Using b, c, and d, a binarization threshold to be applied to each pixel in the interpolation block BH is calculated.

A method of calculating a binarization threshold applied to the pixel bp in the interpolation block BH will be described. Assuming that the interpolation block BH is rectangular, the size (the number of pixels) is xbnum in the horizontal direction and ybnum in the vertical direction. Further, the position of the pixel bp is (m, l). At this time, the values a × Cb (Ga), (xb
The value b × Cb (Gb), (0, ybnu) at the point (num, 0)
The value c × Cb (Gc), (xbnum, ybnu)
At the point m), there is a value d × Cb (Gd), which is linearly approximated.

In this embodiment, a different coefficient Cb is used for each screen. This is necessary to enable higher quality binarization using the average of the photometric values of each screen,
In particular, when the block division and the screen division are different, a high-quality screen binarization threshold is set.

The binarization threshold th applied to the pixel bp is calculated by using the above-described equation (10). In addition,
In Equations (9) and (10), a, b, c, and d are a × Cb (Ga), b × Cb (Gb), and c × C
b (Gc), d × Cb (Gd).

In the binarizer 123, the frame memory 10
7 or the block reading controller 1904 outputs the luminance value of the pixel bp in the interpolation block and the binarization threshold value interpolator 25.
The binarization of the luminance value of the interpolation block BH is performed by comparing the binarization threshold value for the pixel bp calculated in 03 with the threshold value. The binarized image data is supplied to the compressor 1 as in the seventh embodiment.
At 24, image compression suitable for binarized images such as MH and MMR is performed.

FIG. 27 is a flowchart showing the processing flow of image data until a multi-valued image is binarized. First, the CPU reads the image size from the CCD 101 (step S2701). Subsequently, the CPU 108
Is one of the photometers output from the photometer 1901.
The screen G (i, j) is set (step S2702), and the average value of the screen is calculated (step S2703). When obtaining the average, all photometric values may be used, or sampling may be performed as appropriate.

Next, it is determined whether or not the average of the photometric values of the screens protrudes from the average of the photometric values of the adjacent screens (step S2704). If it is protruding (step S2704: YES), the average value of the photometric values is smoothed by the smoother 1902 (step S2705). If it has not protruded (step S2704: NO) or if the smoothing process has been performed in step S2705,
A screen binarization threshold is calculated based on the average of the photometric values (step S2706), and the screen binarization threshold is stored in the memory 1702 (step S2707).

Subsequently, it is determined whether the calculation of the screen binarization threshold for all blocks has been completed (step S27).
08), if not completed (step S2708:
NO), proceeding to step S2702, setting the next screen (for example, G (i + 1, j)), and step S2702.
To step S2708 are repeated. If the calculation of the screen binarization threshold is completed for all the screens (step S
2708: YES), an interpolation block is set (step S2709). This interpolation block is suitable for partially performing high-quality binarization or the like, and may be set in advance by the user, or may be a mode in which an image center portion is set by appropriate mode switching. You may.

The binarization threshold value interpolator 2503 performs the processing in step S
Using the screen binarization threshold (for example, a × Cb (Ga)) of the screen over which the interpolation block is set in 2709, the pixel-based binarization threshold th in the interpolation block
(X, y) is set (step S2710).

The pixel (g (x, y)) read from the frame memory 107 by the binarizer 123 is binarized using the binarization threshold calculated in step S2710 (step S2711). Next, it is determined whether or not the binarization processing has been performed for all the pixels (step S271).
2) In a case where binarization of all pixels has not been completed (step S2712: NO), a pixel in an interpolation block adjacent to g (x, y) is set (for example, g (x +
1, y)), steps S2710 to S271
Repeat up to 2. If binarization has been completed for all pixels (step S2712: YES), it is determined whether or not binarization has been completed for all interpolation blocks (step S2713). (Step S2713: NO), Steps S2709 to S2713 are repeated, and if the processing has been completed (Step S2713: YES), the processing is completed.

In the ninth embodiment, since the binarization threshold is set using information (photometric value) obtained from the automatic photometer provided in the digital camera, processing for separately setting the binarization threshold is performed. Is unnecessary. Therefore, the circuit configuration is simplified, the cost can be reduced, and the power consumption can be reduced because a separate binarization threshold calculation process is not required. In addition, it is possible to binarize a multi-valued image with high quality even if the image has a pinpoint reflection by a light source. Further, since the binarization is performed by setting a binarization threshold applied to each pixel in the interpolation block based on the photometric value, higher quality image processing can be performed than the digital camera of the seventh embodiment. Become. In particular, when individual characters are small, such as when photographing a bulletin board, or when characters in a part of the region are small, it is possible to partially binarize the portion with high quality.

(Embodiment 10) In Embodiment 10,
A description will be given of a case where an image capturing apparatus that calculates a binarization threshold value applied to each pixel in a predetermined block using a CMOS sensor and a photometer and binarizes image data for each pixel is applied to a digital camera. I do. FIG. 28 shows an example in which a CMOS sensor and a photometer are used to calculate a binarization threshold applied to each pixel in a predetermined block, and apply an image capturing apparatus that binarizes image data for each pixel to a digital camera. FIG. 2 is a configuration diagram showing an example. In the present embodiment, the same parts as those in the ninth embodiment are denoted by the same reference numerals, and detailed description thereof will be omitted. The different parts from the ninth embodiment will be mainly described.

The digital camera 2800 has a C
It has a MOS sensor 2801. Therefore, unlike the CCD 101 (see FIG. 25) which can only perform raster scanning,
Since the MOS sensor 2801 can perform random access and can read data in units of blocks, the frame memory 107 and the block read controller 1904 become unnecessary, and the circuit configuration is simplified. In this case, the CPU 108
Have the role of a block read controller.

Further, while the CCD 101 requires a power supply different from that of the CMOS integrated circuit, the CMOS sensor 2
The power supply 801 can use the same power supply as that of the CMOS integrated circuit, and consumes less power. Accordingly, the circuit size of the digital camera 2800 is also reduced, so that power consumption, processing speed, cost, and the like are more convenient than a system using a CCD. Although the CMOS sensor is used in the present embodiment, an image capturing apparatus having an image input unit capable of accessing other blocks may be used.

(Eleventh Embodiment) In the eleventh embodiment,
A description will be given of a case where an image capturing apparatus that sets a binarization threshold for each processing target block is applied to a digital camera after adjusting the photometric value within a predetermined range based on the photometric value output from the photometric unit. I do. FIG.
Is based on the photometric value output from the photometric means, after adjusting the photometric value to fall within a predetermined range, and applying an image capturing apparatus that sets a binarization threshold for each processing target block to a digital camera. FIG. 3 is a block diagram showing an example of a device configuration from image input to recording of a binarized image.

The digital camera 2900 includes the CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU 108, photometer 2901, smoother 2902
, A limiter 2905, a memory 2903, a block read controller 2904, a binarization threshold setting circuit 122, a binarizer 123, a compressor 124, and the image storage memory 12
5 A part for generating a color difference signal is omitted.

The photometer 2901 has an automatic exposure detection mechanism (AE) for performing photometry of a subject before photographing an image, and measures the brightness of each screen based on a digital signal output from the A / D converter 102. I do. The photometry is performed by adding the luminance values of the pixels. At this time, the CPU 108 can also sample the photometric value used for the addition by the photometer 2901.

The smoother 2902 smoothes the photometric value of each screen obtained by the photometer 2901 and outputs it to the limiter 2905 as ave (i, j). As an example of smoothing,
The following processing is included. That is, the average value ave (i, j) of the photometric values of all the pixels included in one screen (G (i, j)) in the photometer 2901 (or the screen G
When the average value of the photometric values sampled from (i, j)) is a value protruding from the average value of the photometric values of the surrounding screen, the photometric value of each pixel of the screen G (i, j) is calculated. The correction is performed so that the value does not become prominent compared with the average value of the photometric values of the surrounding screen.

An example of an algorithm for realizing this processing is shown below. The average photometry value of four screens adjacent to the screen G (i, j) is set as ave4 (i, j), and the screen G
The average photometric value ave (i, j) of (i, j) is ave4
When it is three times or more than (i, j), the average photometric value ave (i, j) of the screen G (i, j) is converted as in the following equation (12). if ave (i, j) ≧ 3 * ave4 (i, j) then ave (i, j) = ave4 (i, j) + (1/4) * (ave (i, j) −ave4 (i, j) )) ‥‥ (12)

The smoother 2902 calculates the average photometric value ave (i, j) of the screen G (i, j) as described above. Further, in the limiter 2905, ave (i, j)
Is limited to a preset lower limit to prevent the output of an average photometry value that is too low. After that, the average photometric value is output to the memory 2903. Note that, depending on the mode of use, the photometric value whose range is limited by the limiter 2905 may be output without providing the smoothing unit 2902.

The binarization threshold value setting circuit 122 outputs
The binarization threshold TH (i, j) is set based on (i, j). The setting means sets a predetermined coefficient Cb to ave (i, j)
Is multiplied by Cb = x / 16 or Cb = x / 16
b = x / 8 (x is a default value representing a natural number not exceeding the denominator)
Then, since the binarization threshold value setting circuit 122 can be constituted only by an adder, it is advantageous in terms of cost and speed. Note that the value of Cb is increased so as to approach 1 for a block having a large photometric value, and the difference between the binarization threshold and the average photometric value is reduced for an area where whiteout occurs. In the present embodiment, since there is a smoother 2902, as for the region smoothed by the smoother 2902, Cm = 0 or Cm as described using the seventh embodiment.
The binarization threshold TH may be calculated by performing subtraction as <0.

The binarizer 123 outputs a binarization threshold TH (i,
Based on j), each pixel of the block B (i, j) corresponding to the screen G (i, j) is binarized. Compressor 124 is M
Compression suitable for a binary image is performed by H, MMR, or the like. The image storage memory 125 stores the compressed image.

Next, the flow of processing until a multi-valued image is binarized will be described. FIG. 30 is a flowchart illustrating a processing flow of image data until a multi-valued image is binarized. First, the CPU reads the image size from the CCD 101 (step S3001). Then, CPU
108 sets one screen G (i, j) from the photometric values output from the photometer 2901 (step S300).
2), calculate the average value of the screen (step S300)
3). When obtaining the average, all photometric values may be used, or sampling may be performed as appropriate.

Next, it is determined whether or not the average value of the photometric values of the screens protrudes from the average value of the photometric values of the adjacent screen (step S3004). If it is protruding (step S3004: YES), the average of the photometric values is smoothed by the smoothing unit 2902 (step S3005). If it does not protrude (step S3004: NO) or if the smoothing process is performed in step S3005, it is determined whether the output photometric value is too low (step S3006). If it is too low (step S300
6: YES), the value of the photometric value is replaced with a predetermined value (step S3007). When the value is not too low (step S3006: NO) or when the replacement process is performed in step S3007, a binarization threshold is calculated based on the photometric value of the screen G (i, j) (step S300).
8).

Next, the screen G is read from the frame memory 107.
The block read controller 2904 reads the block B (i, j) corresponding to (i, j) (Step S300)
9), the multi-valued image in the block is binarized using the binarization threshold (step S3010). If binarization has been completed for all blocks (step S301)
1: YES), the binarization process ends, and if not (step S3011: NO), step S300
2 and the next screen (for example, G (i + 1, j)) is set. Thereafter, steps S3002 to S301 are performed.
Repeat 1.

In the eleventh embodiment, since the binarization threshold is set using information (photometric value) obtained from the automatic photometer provided in the digital camera, a process for separately setting the binarization threshold is performed. Is unnecessary. Therefore, the circuit configuration is simplified, the cost can be reduced, and the power consumption can be reduced because a separate binarization threshold calculation process is not required. Further, it is possible to binarize a multi-valued image with high quality even if the image has a pinpoint reflection by a light source. In addition, the binarization threshold is set efficiently by using the smoother and the limiter together.

In the digital camera 2900, CC
Although D101 is used, a configuration of a digital camera 3100 using a CMOS sensor 3101 can be adopted as shown in FIG. With such a configuration, compared to the digital camera 2900,
A frame memory and a block reading controller are not required, and the circuit configuration is simplified. Further, the same power supply as that of the CMOS integrated circuit can be used for the CMOS sensor 3101, and power consumption is reduced. Therefore, the digital camera 3100 can have a smaller circuit scale than the digital camera 2900, and also consumes less power. Therefore, digital camera 2
It is possible to provide a digital camera that is more convenient than 900 in terms of power consumption, processing speed, cost, and the like.

(Twelfth Embodiment) In the twelfth embodiment,
A case will be described in which an image capturing apparatus that calculates a binarization threshold to be applied to individual pixels in a predetermined block based on photometric values limited to a predetermined range is applied to a digital camera. FIG. 32 is a diagram illustrating an example in which an image capturing apparatus that calculates a binarization threshold to be applied to each pixel in a predetermined block based on photometric values limited to a predetermined range is applied to a digital camera. FIG.

The digital camera 3200 includes the CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU 108, photometer 2901, smoother 2902
, Limiter 3201, and block binarization threshold setting circuit 1
001, the memory 1002, and the binarization threshold interpolator 100
3, a binarizer 123, a compressor 124, and an image storage memory 125. A part for generating a color difference signal is omitted.

The CCD 101 is a digital camera 3200
Is a part that converts light condensed by an optical system (not shown) into an electric signal.
It is a part that outputs a B analog signal. The output analog signal is converted into a digital signal by the A / D converter 102. The digital signal is sent to the white balance adjuster 10
3. Through the pixel interpolator 104, the luminance generator 105, and the aperture corrector 106, processing such as interpolation and extraction of luminance values is performed, and is temporarily stored in the frame memory 107.

The CPU 108 is a digital camera 3200
Each circuit and each part are controlled. A photometer 2901 is an automatic exposure detection mechanism (AE) that performs photometry of a subject before capturing an image.
And measures the brightness of each screen based on the digital signal output from the A / D converter 102. The photometry is performed by adding the luminance values of the pixels. At this time CP
U108 can also sample the photometric value used for the addition by the photometer 2901.

The smoothing unit 2902 smoothes the photometric value of each screen obtained by the photometer 2901 and outputs it to the limiter 3201 as ave (i, j). Further, the limiter 320
In step 1, ave (i, j) is limited to a preset lower limit to prevent output of an average photometry value that is too low.

The block binarization threshold value setting circuit 1001
threshold value TH (i, j) based on ave (i, j)
Set. Block binarization thresholds of all blocks are stored in the memory 1002. Next, using the binarization thresholds of all the blocks, the binarization threshold interpolator 1003 calculates a binarization threshold for each pixel. While calculating the binarization threshold value for each pixel, the luminance value of the pixel is stored in the frame memory 10.
7, and the luminance value and the binarization threshold are input to the binarizer 123. The binarizer 123 binarizes the luminance value by comparing the binarization threshold with the luminance value. The binarized image is subjected to image compression suitable for a binary image such as MH or MMR by the compressor 124. The compressed image is stored in the image storage memory 125. Further, a digital camera can be realized with a similar configuration even if a CMOS sensor is used instead of the CCD 101.

Next, the flow of processing until a multi-valued image is binarized will be described. FIG. 33 is a flowchart illustrating a processing flow of image data until a multi-valued image is binarized. First, the CPU reads the image size from the CCD 101 (step S3301). Then, CPU
108 sets one screen G (i, j) from the photometric values output from the photometer 2901 (step S330).
2), calculate the average value of the screen (step S330)
3). When obtaining the average, all photometric values may be used, or sampling may be performed as appropriate.

Next, it is determined whether or not the average value of the photometric values of the screens protrudes from the average value of the photometric values of the adjacent screen (step S3304). If it is protruding (step S3304: YES), the average of the photometric values is smoothed by the smoother 2902 (step S3305). If it does not protrude (step S3304: NO) or if the smoothing process has been performed in step S3305, it is next determined whether the photometric value is too low (step S3).
3306). If it is too low (step S3306: Y
ES), the photometric value is replaced with a predetermined value (step S3307). If the value is not too low (step S33)
06: NO) or when the replacement process is performed in step S3307, based on the photometric value of the screen G (i, j), the block binarization threshold TH of the block corresponding to the screen G (i, j)
(I, j) is calculated (step S3308).

Next, it is determined whether or not the block binarization threshold has been calculated for all the blocks (step S33).
09), if the block binarization thresholds of all blocks have been calculated (step S3309: YES), the process proceeds to the next step, and if not (step S3).
309: NO) sets a block adjacent to G (i, j) (for example, G (i + 1, j)), and proceeds to step S330.
Steps 2 to S3309 are repeated.

When the block binarization threshold has been calculated for all blocks (step S3309: YE
S), and sets an interpolation block (step S331)
0). The binarization threshold interpolator 1003 calculates a binarization threshold th (x, y) for each pixel by interpolation using the binarization threshold (step S3311). Note that x and y are natural numbers indicating the position of each pixel of the image. Further, using the binarization threshold th (x, y), the pixel (g (x, y)) read from the frame memory 107 is binarized (step S3312). It is determined whether or not the binarization processing has been performed for all the pixels (step S3313), and when the binarization of all the pixels has not been completed (step S331).
3: NO) sets a pixel adjacent to g (x, y) (for example, g (x + 1, y)), and repeats steps S3311 to S3313.

If the binarization has been completed for all the pixels (step S3313: YES), it is determined whether the binarization has been completed for all the interpolation blocks (step S33).
14) If not completed (step S3314:
NO), Steps S3310 to S3314 are repeated, and if binarization has been completed for all interpolation blocks (Step S3314: YES), the process ends.

In the twelfth embodiment, since the binarization threshold is set using information (photometric value) obtained from the automatic photometer provided in the digital camera, processing for separately setting the binarization threshold is performed. Is unnecessary. Therefore, the circuit configuration is simplified, the cost can be reduced, and the power consumption can be reduced because a separate binarization threshold calculation process is not required. In addition, it is possible to binarize a multi-valued image with high quality even if the image has a pinpoint reflection by a light source. In addition, the binarization threshold is set efficiently by using the smoother and the limiter together. Furthermore, since binarization is performed by setting a binarization threshold value applied to each pixel in the interpolation block based on the photometric value, high-quality image processing can be performed. In particular, when individual characters are small, such as when photographing a bulletin board, or when characters in a part of the region are small, it is possible to partially binarize the portion with high quality.

(Thirteenth Embodiment) In the thirteenth embodiment,
An image capturing apparatus that performs image binarization by software processing in a CPU will be described. FIG. 34 is a block diagram illustrating an example of a device configuration from image input to recording of a binarized image in a case where an image capturing device that performs image binarization by software processing in a CPU is applied to a digital camera. FIG. In the present embodiment,
The same components as those in the above-described embodiments are denoted by the same reference numerals, and detailed description thereof will be omitted. Different portions will mainly be described.

The digital camera 3400 includes the CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU3401, ROM3402, RAM3403,
An image storage memory 125. A part for generating a color difference signal is omitted.

The CCD 101 is a digital camera 3400
Is a part that converts light condensed by an optical system (not shown) into an electric signal.
It is a part that outputs a B analog signal. The output analog signal is converted into a digital signal by the A / D converter 102. The digital signal is sent to the white balance adjuster 10
3. Through the pixel interpolator 104, the luminance generator 105, and the aperture corrector 106, processing such as interpolation and extraction of luminance values is performed, and is temporarily stored in the frame memory 107.

The CPU 3401 is a digital camera 340
In addition to controlling each circuit and each part of 0, it has a processing function of image binarization. ROM connected to CPU 3401
3402 stores a software program for realizing the binarization function. Further, the coefficient Cb or C
m is also stored. Further, the RAM 3403 stores data such as image data as a work area for performing a binarization process.

[0201] The ROM 3402 contains
The average luminance calculator in the embodiment, a low luminance threshold setting device, a binarization threshold setting circuit, a binarization threshold interpolator, a binarizer, a software program for realizing functions such as a compressor are stored, Average brightness calculation, binarization threshold setting,
Functions such as binarization and compression of a binary image can be realized by executing a program in the CPU 3401.
The image binarized and compressed by the CPU 3401 is stored in the image storage memory 125.

The digital camera 3400 may have a photometer, a smoother, a limiter, a block binarization threshold setting circuit, a binarization threshold interpolator, a binarizer, and a compressor depending on the mode of use. A software program that realizes functions such as binarization threshold setting, binarization, and compression of a binary image is realized by the CPU 3401 executing the program. CP
The image binarized and compressed in U3401 is stored in the image storage memory 125. Also, instead of the CCD 101, C
A MOS sensor may be used.

In the thirteenth embodiment, since each function is realized by the CPU 3401 and the ROM 3402, it is not necessary to separately form each functional unit, and it is possible to reduce the development cost and to provide an inexpensive digital camera. It becomes possible. In addition, by upgrading software, it is possible to always provide the latest algorithm.

(Embodiment 14) In Embodiment 14,
An image pickup apparatus having a photometer that performs image binarization by software processing in a CPU will be described. FIG.
FIG. 3 is a block diagram illustrating a device configuration from image input to recording of a binarized image in a case where an image capturing device that performs image binarization by software processing in a CPU is applied to a digital camera. In the present embodiment, detailed description of the same components as those in the above-described thirteenth embodiment will be omitted, and different portions will be mainly described.

The digital camera 3500 includes the CCD 101
, A / D converter 102 and white balance adjuster 1
03, a pixel interpolator 104, a luminance generator 105, an aperture corrector 106, a frame memory 107,
PU 3501, ROM 3502, RAM 3503, photometer 3504, memory 3505, image storage memory 1
25. A part for generating a color difference signal is omitted.

The CCD 101 is a digital camera 3500
Is a part that converts light condensed by an optical system (not shown) into an electric signal.
This is a part that outputs an analog signal. The output analog signal is converted into a digital signal by the A / D converter 102. The digital signal is sent to the white balance adjuster 10
3. Through the pixel interpolator 104, the luminance generator 105, and the aperture corrector 106, processing such as interpolation and extraction of luminance values is performed, and is temporarily stored in the frame memory 107.

The photometer 3504 has an automatic exposure detection mechanism (AE) for performing photometry of a subject before taking an image, and measures the brightness of each screen based on a digital signal output from the A / D converter 102. I do. The photometry is performed by adding the luminance values of the pixels. At this time, the CPU 3501 can also sample the photometric value used for the addition by the photometer 3504. Since the photometer 3504 can calculate the average luminance simultaneously with the generation of the luminance signal,
The amount of calculation in the PU 3501 is reduced, and the processing from image capturing to storage of a binary image can be performed at a higher speed than in the thirteenth embodiment.

The CPU 3501 is a digital camera 350
In addition to controlling each circuit and each part of 0, it has a function for image binarization. RO connected to CPU 3501
In M3502, a software program for realizing the binarization function is stored. RAM 3503
Stores data such as image data as a work area for performing binarization processing.

The ROM 3502 stores software programs for realizing the functions of the smoother, the limiter, the block binarization threshold setting circuit, the binarization threshold interpolator, the binarizer, and the compressor in the above embodiment. Are stored, and functions such as a binarization threshold setting, binarization, and compression of a binary image are C
This can be realized by executing a program in the PU 3501. Further, the coefficient Cb or Cm is also stored. The image binarized and compressed by the CPU 3501 is stored in the image storage memory 125. Depending on the mode of use, a CMOS sensor may be used instead of the CCD 101.

In the fourteenth embodiment, similarly to the thirteenth embodiment, each function is realized by the CPU 3501 and the ROM 3502, so that it is not necessary to separately form individual functional units, and it is possible to reduce development costs. It is possible to provide an inexpensive digital camera. Also,
Upgrading the software makes it possible to always provide the latest algorithm.

(Embodiment 15) The present invention can also be realized by a software program in addition to the embodiments described above. FIG. 36 is a diagram illustrating a configuration example of a computer system when the present invention is implemented by software.

In the figure, 3601 is a CPU for controlling the entire apparatus based on the control program, 3602 is a ROM storing the control program, and 3603 is an RA
M, 3604 is a display device for displaying the input / output state of the computer, 3605 is a hard disk, 3
Reference numeral 606 denotes a keyboard used for inputting a character string and the like, 3607 denotes a CD-ROM drive, and 3608
Is a CD-ROM as a computer-readable recording medium
It represents a ROM and stores a program for realizing the image binarization method of the present invention.

In the computer system configured as described above, a program for realizing the image binarization method of the present invention is recorded on the CD-ROM 3608. CPU
Under the control and processing of 3601, the above program is read, and the program is started to execute the image binarization processing, and the binarized information is output to the hard disk 3605 or the like.

[0214]

As described above, according to the image binarizing apparatus of the present invention (claim 1), the luminance value input means inputs the luminance value of each pixel constituting the multi-valued image and outputs the average luminance value. Calculating means for calculating a local average luminance value including a pixel to be binarized based on the luminance value input by the luminance value input means; and a binarization threshold setting means for calculating the average luminance value. The binarization threshold is set lower than the local average luminance value according to the magnitude of the local average luminance value calculated by the means, and the binarization means is set by the binarization threshold setting means. Since the pixel to be binarized is binarized by the binarization threshold, a binarization threshold having a different value according to the average luminance value is used, and the binarization threshold is appropriate for an area where surface reflection occurs. A binarization threshold can be adopted, which allows a multi-valued image to be binarized with high quality. It is possible to become.

[0215] Further, according to the image binarizing device of the present invention (claim 2), in the image binarizing device according to claim 1, wherein the binarizing threshold value setting means is calculated by the average luminance value calculating means. The correction amount according to the obtained local average luminance value, and when the average luminance value is in a predetermined intermediate range, the correction amount is larger than that in the case where the average luminance value is outside the intermediate range. Since the binarization threshold is set by subtracting from the average luminance value, in the luminance range where surface reflection occurs, an appropriate binarization threshold is set with a correction amount that reduces the difference between the average luminance value and the binarization threshold. While adopting, an appropriate binarization threshold can be adopted in the intermediate range with a relatively large correction amount, and thereby, a multi-valued image can be binarized with high quality. In addition, by preventing the difference between the average luminance value on the low luminance side and the binarization threshold from being large, it is possible to suppress the occurrence of noise on the low luminance side, whereby the multi-valued image can be made high quality. It is possible to binarize.

[0216] In the image binarizing apparatus according to the present invention (claim 3), in the image binarizing apparatus according to claim 1, the binarization threshold value setting means is calculated by the average luminance value calculating means. Since the binarization threshold is set by multiplying the obtained local average luminance value by using a correction amount of a larger value as the average luminance value becomes a larger value, in a luminance range in which surface reflection occurs, An appropriate binarization threshold value can be adopted by a correction amount that reduces the difference between the average luminance value and the binarization threshold value, which makes it possible to binarize a multi-valued image with high quality. Become.

In the image binarizing device of the present invention (claim 4), the luminance value input means inputs the luminance value of each pixel constituting the multi-valued image, and the average luminance value calculating means outputs the luminance value. Based on the brightness value input by the value input means, a local average brightness value including the pixel to be binarized is calculated, and the subtraction threshold calculation means calculates the local average brightness value calculated by the average brightness value calculation means. A first binarization threshold is calculated by subtracting the first correction amount from the average luminance value, and the multiplication threshold value calculating means calculates the local average luminance value calculated by the average luminance value calculating means by one.
A second binarization threshold is calculated by multiplying by a second correction amount that is a value less than and the comparison determination unit determines that the magnitude of the local average luminance value calculated by the average luminance value calculation unit is a predetermined value. It is determined whether the value is larger or smaller than the comparison value of, and the binarization threshold value determination means determines that the magnitude of the local average luminance value is larger than the comparison value by the comparison determination means. In this case, it is determined that the first binarization threshold value calculated by the subtraction threshold value calculation means is used, and the magnitude of the local average luminance value is larger than the comparison value by the comparison determination means. When it is determined to be small, it is determined to adopt the second binarization threshold calculated by the multiplication threshold calculation means, and the binarization means is determined by the binarization threshold determination means. Binarizing the pixel to be binarized according to the determined binarization threshold Therefore, the binarization threshold for the region where the surface reflection occurs and the binarization threshold for the normal region are selectively used. For the region where the surface reflection occurs, as the average luminance value increases, It is possible to adopt a binarization threshold that makes the difference from the average luminance value relatively small, and to adopt a binarization threshold that makes the overall tone after binarization smooth for a normal region. This makes it possible to binarize a multi-valued image with high quality.

Further, in the image binarizing device according to the present invention (claim 5), in the image binarizing device according to claim 4, the subtraction threshold value calculating means is calculated by the average luminance value calculating means. Since the first binarization threshold is calculated by subtracting a correction amount corresponding to the magnitude of the luminance value from the local average luminance value, a more appropriate threshold value in a luminance range in which surface reflection occurs is obtained. A binarization threshold can be employed, which makes it possible to binarize a multi-valued image with high quality.

[0219] In the image binarizing device according to the present invention (claim 6), in the image binarizing device according to claim 4 or 5, the multiplication threshold value setting means is calculated by the average luminance value calculating means. Since the second binarization threshold is calculated by multiplying the obtained local average luminance value by a correction amount corresponding to the magnitude of the luminance value, in a region where surface reflection does not occur, It is possible to adopt a binarization threshold that allows a smooth binary image to be obtained without generating a boundary, thereby enabling a multivalued image to be binarized with high quality.

The image binarizing device according to the present invention (claim 7) is the image binarizing device according to any one of claims 1 to 6, wherein the block dividing means converts the multi-valued image into blocks. And the average luminance value calculation means calculates the local average luminance value in units of blocks divided by the block division means, so that an appropriate binarization threshold value can be set for each block. This makes it possible to binarize a multi-valued image with high quality.

[0221] Further, in the image binarizing apparatus according to the present invention (claim 8), in the image binarizing apparatus according to claim 7, the comparison value determining means calculates the local average luminance value. Since the comparison value is determined on the basis of the average luminance value of the block adjacent to, the binarization threshold can be set by more accurately determining the overexposure of the multi-valued image. High quality binarization is possible.

In the image binarizing device of the present invention (claim 9), the density value input means inputs the density value of each pixel constituting the multi-valued image, and the average density value calculation means outputs the density value. Based on the density value input by the value input unit, a local average density value including the pixel to be binarized is calculated, and the binarization threshold setting unit calculates the local average density value calculated by the average density value calculation unit. The binarization threshold is set higher than the local average density value in accordance with the magnitude of the average density value, and the binarization unit uses the binarization threshold set by the binarization threshold setting unit. Since the pixels to be binarized are binarized, a binarization threshold having a different value according to the average density value is used, and an appropriate binarization threshold is used for an area where surface reflection occurs. This allows multi-valued images to be binarized with high quality It made.

The image binarizing apparatus according to the present invention (Claim 1)
0) The image binarization device according to claim 9, wherein the binarization threshold value setting unit is a correction amount according to a local average density value calculated by the average density value calculation unit. When the average density value is in a predetermined intermediate range, the binarization threshold is set by adding a correction amount that is a larger value than when the average density value is outside the intermediate range to the local average density value. In the density range where surface reflection occurs, while using an appropriate binarization threshold with a correction amount that reduces the difference between the average density value and the binarization threshold, an appropriate binarization with a relatively large correction amount is used in the intermediate range. A binarization threshold can be adopted, which makes it possible to binarize a multi-valued image with high quality. In addition, by preventing the difference between the average density value on the high density side and the binarization threshold from being large, it is possible to suppress the occurrence of noise on the high density side, thereby improving the quality of the multi-valued image. It is possible to binarize.

The image binarizing apparatus of the present invention (Claim 1)
1) The image binarization device according to claim 9, wherein the binarization threshold value setting unit sets an average density value to a local average density value calculated by the average density value calculation unit. Since the binarization threshold is set by multiplying by using a smaller correction amount as the value becomes larger, the difference between the average density value and the binarization threshold is reduced in the density range where surface reflection occurs. An appropriate binarization threshold value can be adopted by such a correction amount, and thereby, it becomes possible to binarize a multi-valued image with high quality.

The image binarizing apparatus of the present invention (Claim 1)
In 2), the density value input means inputs the density value of each pixel constituting the multi-valued image, and the average density value calculation means performs binarization based on the density value input by the density value input means. The local average density value including the power pixel is calculated, and the addition threshold value calculation means adds a first correction amount to the local average density value calculated by the average density value calculation means to obtain a first average correction value. A binarization threshold value is calculated, and the multiplication threshold value calculation means adds 1 to the local average density value calculated by the average density value calculation means.
A second binarization threshold is calculated by multiplying the second correction amount which is the above value, and the comparison and determination unit determines that the magnitude of the local average density value calculated by the average density value calculation unit is a predetermined value. It is determined whether the value is larger or smaller than the comparison value, and the binarization threshold value determination unit determines that the local average density value is smaller than the comparison value by the comparison determination unit. In this case, it is determined that the first binarization threshold value calculated by the addition threshold value calculation means is to be used, and the magnitude of the local average density value is larger than the comparison value by the comparison determination means. When it is determined to be large, it is determined to adopt the second binarization threshold calculated by the multiplication threshold calculation means, and the binarization means is determined by the binarization threshold determination means. Binarizing the pixel to be binarized according to the determined binarization threshold Therefore, the binarization threshold for the region where the surface reflection occurs and the binarization threshold for the normal region are selectively used. For the region where the surface reflection occurs, as the average density value decreases, It is possible to adopt a binarization threshold that makes the difference from the average density value relatively small, and for a normal region, adopt a binarization threshold that makes the overall tone after binarization smooth. This makes it possible to binarize a multi-valued image with high quality.

Further, the image binarizing apparatus of the present invention (Claim 1)
3) The image binarizing device according to claim 12, wherein
The addition threshold calculating means adds the correction amount according to the magnitude of the density value to the local average density value calculated by the average density value calculating means, thereby obtaining the first
Is calculated, a more appropriate binarization threshold can be adopted in the density range in which surface reflection occurs, whereby a multivalued image can be binarized with high quality.

The image binarizing apparatus of the present invention (Claim 1)
4) The image binarizing device according to claim 12 or 13, wherein the multiplication threshold setting unit sets a local average density value calculated by the average density value calculation unit as:
Since the second binarization threshold is calculated by multiplying by the correction amount corresponding to the magnitude of the density value, a smooth binary image without boundaries is obtained in an area where surface reflection does not occur. It is possible to adopt a binarization threshold value, which enables binarization of a multi-valued image with high quality.

Further, the image binarizing apparatus of the present invention (Claim 1)
5) The image binarizing device according to any one of claims 9 to 14, wherein the block dividing unit divides the multi-valued image into blocks, and wherein the average density value calculating unit includes the block dividing unit. Calculates the local average density value based on the blocks divided by, so that an appropriate binarization threshold value can be set for each block, thereby binarizing a multi-valued image with high quality. It becomes possible.

Further, the image binarizing apparatus of the present invention (Claim 1)
6) The image binarizing device according to claim 15, wherein
The comparison value determination means determines the comparison value based on the average density value of a block adjacent to the block for which the local average density value is calculated. A binarization threshold can be set, which makes it possible to binarize a multi-valued image with high quality.

The image binarization method of the present invention (Claim 1)
7) is a binarization threshold of a value lower than the average luminance value of a local pixel set including a pixel to be binarized, which is input with a luminance value of a pixel forming a multi-valued image. For a set of pixels that are not overexposed, determine a binarization threshold having a smaller difference from the average luminance value than for a pixel set that is not overexposed, and determine the image of the local pixel set by using the binarization threshold. Since binarization is used for output, it is possible to set an appropriate binarization threshold even for an image having surface reflection, thereby binarizing a multi-valued image with high quality. It becomes possible.

The image binarization method of the present invention (Claim 1)
8) is a binarization threshold of a value higher than the average density value of a local pixel set including a pixel to be binarized, which is input with a density value of a pixel forming a multi-valued image. For a set of pixels that are not overexposed, determine a binarization threshold that has a smaller difference from the average density value than for a pixel set that is not overexposed. Since binarization is used for output, it is possible to set an appropriate binarization threshold even for an image having surface reflection, thereby binarizing a multi-valued image with high quality. It becomes possible.

The image binarization program according to the present invention (claim 19) provides a computer which inputs a luminance value of a pixel constituting a multi-valued image, and a local pixel set including a pixel to be binarized. The binarization threshold value is lower than the average luminance value of, and the difference from the average luminance value is smaller for a pixel set that is overexposed than for a pixel set that is not overexposed. Since the binarization threshold is determined and the image of the local pixel set is binarized and output using the binarization threshold, an appropriate binarization threshold is also set for an image having surface reflection. This makes it possible to binarize the multi-valued image with high quality.

The image binarization program according to the present invention (claim 20) provides a computer which inputs a density value of a pixel constituting a multi-valued image, and a local pixel set including a pixel to be binarized. The binarization threshold value is higher than the average density value, and the difference from the average density value is smaller for a pixel set that is overexposed than for a pixel set that is not overexposed. Since the binarization threshold is determined and the image of the local pixel set is binarized and output using the binarization threshold, an appropriate binarization threshold is also set for an image having surface reflection. This makes it possible to binarize the multi-valued image with high quality.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an example of a device configuration from input of image data to recording of a binarized image when the image binarizing device according to the first embodiment is applied to a digital camera. is there.

FIG. 2 is a CCD in the digital camera according to the first embodiment.
FIG. 4 is a conceptual diagram showing the concept of a light receiving unit.

FIG. 3 is a diagram illustrating a division example in which a multivalued image captured by the digital camera according to the first embodiment is divided into blocks.

FIG. 4 is a diagram illustrating an example of a sampling interval for sampling pixels in a block.

FIG. 5 is a block diagram illustrating an example of a configuration of an average luminance value calculator of the digital camera according to the first embodiment.

FIG. 6 is an explanatory diagram illustrating a configuration example of a binarization threshold setting circuit of the digital camera according to the first embodiment;

FIG. 7 is a flowchart showing a processing flow of image data until a multi-value image is binarized in the digital camera according to the first embodiment.

FIG. 8 is a flowchart illustrating an example of a processing flow of a binarization threshold value calculation routine in the flowchart illustrated in FIG. 7;

FIG. 9 is a block diagram showing an example of a device configuration when the image binarization device according to the second embodiment is applied to a digital camera.

FIG. 10 is a block diagram illustrating an example of a device configuration when the image binarizing device according to the third embodiment is applied to a digital camera.

FIG. 11 is an explanatory diagram showing a configuration example of a block binarization threshold value setting circuit of the digital camera according to the third embodiment.

FIG. 12 is an explanatory diagram showing an example of coefficients stored in a coefficient memory.

FIG. 13 is an explanatory diagram illustrating an outline of calculating a binarization threshold value applied to each pixel in an interpolation block of the digital camera according to the third embodiment.

FIG. 14 is a flowchart showing a processing flow of image data until a multivalued image is binarized in the digital camera according to the third embodiment.

FIG. 15 is a flowchart showing an example of a processing flow of a block binarization threshold value calculation routine in the flowchart shown in FIG. 14;

FIG. 16 is a block diagram illustrating an example of a device configuration of a digital camera that uses a CMOS sensor for an image input portion, calculates a binarization threshold applied to each pixel, and binarizes image data for each pixel. FIG.

FIG. 17 is a block diagram illustrating an example of a device configuration when the image binarization device according to the fifth embodiment is applied to a digital camera.

FIG. 18 is a block diagram illustrating an example of a device configuration when the image binarizing device according to the sixth embodiment is applied to a digital camera.

FIG. 19 is a block diagram illustrating an example of a device configuration from image input to recording of a binarized image when the image capturing device according to the seventh embodiment is applied to a digital camera.

FIG. 20 is an explanatory diagram showing a configuration example of a binarization threshold value setting circuit of the digital camera according to the seventh embodiment.

FIG. 21 is an explanatory diagram showing an example of coefficients stored in a coefficient memory.

FIG. 22 is a flowchart showing a processing flow of image data until a multi-valued image is binarized in the digital camera according to the seventh embodiment.

FIG. 23 is a flowchart illustrating an example of a processing flow of a binarization threshold value calculation routine illustrated in FIG. 22;

FIG. 24 is a block diagram showing an example of a device configuration up to a step of binarizing and recording an input image in a digital camera using a CMOS sensor for an image input portion.

FIG. 25 is a block diagram illustrating an example of a device configuration when the image capturing device according to Embodiment 9 is applied to a digital camera.

FIG. 26 is a diagram illustrating a relationship between an interpolation block and a screen in the digital camera according to the ninth embodiment.

FIG. 27 is a flowchart illustrating a processing flow of image data until a multi-valued image is binarized in the digital camera according to the ninth embodiment.

FIG. 28 is a diagram illustrating a digital camera in which a binarization threshold applied to each pixel in a predetermined block is calculated using a CMOS sensor and a photometer and binarization of image data is performed for each pixel. FIG. 2 is a configuration diagram showing an example.

FIG. 29 is a block diagram illustrating an example of a device configuration in a case where the image capturing device according to Embodiment 11 is applied to a digital camera.

FIG. 30 shows a digital camera according to an eleventh embodiment.
9 is a flowchart illustrating a processing flow of image data until a multi-value image is binarized.

FIG. 31 is a block diagram showing an example of another device configuration of the digital camera according to the eleventh embodiment.

FIG. 32 is a block diagram illustrating an example of a device configuration in a case where the image capturing device according to Embodiment 12 is applied to a digital camera.

FIG. 33 shows a digital camera according to a twelfth embodiment.
9 is a flowchart illustrating a processing flow of image data until a multi-value image is binarized.

FIG. 34 is a block diagram illustrating an example of a device configuration from image input to recording of a binarized image in a case where an image capturing device that performs image binarization by software processing in a CPU is applied to a digital camera. FIG.

FIG. 35 is a block diagram illustrating a device configuration from image input to recording of a binarized image in a case where an image capturing device that performs image binarization by software processing in a CPU is applied to a digital camera. .

FIG. 36 is a diagram illustrating a configuration example of a computer system when the present invention is realized by software.

[Explanation of symbols]

100, 900, 1000, 1600, 1700, 18
00, 1900, 2400, 2500, 2800, 29
00, 3100, 3200, 3400, 3500 Digital camera 101 CCD 102 A / D converter 103 White balance adjuster 104 Pixel interpolator 105 Luminance generator 106 Aperture corrector 107 Frame memory 109 Block buffer 120 Average luminance value calculator 121 Low Brightness threshold setting device 122, 1905 Binarization threshold setting circuit 123 Binarization device 124 Compressor 125 Image storage memory 501 Comparator 502 Addition result register 503 Counter 504 Shift register 506 Adder 601, 2001 Addition and subtraction device 602, 1011 Multiplication Divider 603 Comparator 604 Selector 1001 Block binarization threshold setting circuit 1003, 2503 Binarization threshold interpolator 1012 Coefficient memory (Cb) 1901, 2901, 3504 Photometer 190 , 2902 Smoothing unit 1904, 2904 Block readout controller 2002 Coefficient memory (Cm) 2501 Screen binarization threshold setting circuit Cb coefficient, constant Cm coefficient, constant TH Binarization threshold (block binarization threshold) ave Average value ( Average luminance value, average density value) Y luminance signal value (luminance value) th binarization threshold thl low luminance threshold v luminance value

Claims (20)

[Claims] 1. A brightness value input means for inputting a brightness value of each pixel constituting a multi-valued image, and a local value including a pixel to be binarized based on the brightness value input by the brightness value input means. Average brightness value calculating means for calculating an average brightness value, and a binarization threshold lower than the local average brightness value according to the size of the local average brightness value calculated by the average brightness value calculating means. Binarization threshold setting means for setting, and binarization means for binarizing the pixel to be binarized with the binarization threshold set by the binarization threshold setting means, Image binarization device. 2. The method according to claim 1, wherein the binarization threshold value setting unit is a correction amount corresponding to a local average luminance value calculated by the average luminance value calculation unit, and the average luminance value is within a predetermined intermediate range. 2. The image processing apparatus according to claim 1, wherein a threshold value is set by subtracting a correction amount, which is a value larger than a value outside the intermediate range, from the local average luminance value. Pricer. 3. The binarization threshold setting unit uses a correction amount of a larger value with respect to a local average luminance value calculated by the average luminance value calculation unit as the average luminance value becomes larger. 2. The image binarization device according to claim 1, wherein the binarization threshold is set by multiplying the image by a threshold. 4. A luminance value input means for inputting a luminance value of each pixel constituting a multi-valued image, and a local value including a pixel to be binarized based on the luminance value input by the luminance value input means. Average brightness value calculating means for calculating an average brightness value; calculating a first binary threshold by subtracting a first correction amount from the local average brightness value calculated by the average brightness value calculating means; Subtraction threshold value calculation means; multiplication threshold value calculation for calculating a second binarization threshold value by multiplying a local average brightness value calculated by the average brightness value calculation means by a second correction amount less than 1 Means, a comparison determination means for determining whether the local average brightness value calculated by the average brightness value calculation means is larger or smaller than a predetermined comparison value, and The magnitude of the local average luminance value is greater than the comparison value. When it is determined that the first binarization threshold calculated by the subtraction threshold calculation unit is used, the magnitude of the local average luminance value is determined by the comparison determination unit. When it is determined that is smaller than the comparison value, a binarization threshold determining unit that determines to adopt the second binarization threshold calculated by the multiplication threshold calculation unit; An image binarization device, comprising: a binarization unit that binarizes the pixel to be binarized according to the binarization threshold determined by the binarization threshold determination unit. 5. The subtraction threshold value calculation means subtracts a correction amount according to the magnitude of the luminance value from the local average luminance value calculated by the average luminance value calculation means, and 5. The image binarization device according to claim 4, wherein one binarization threshold is calculated. 6. The multiplying threshold setting means multiplies the local average luminance value calculated by the average luminance value calculating means by a correction amount according to the magnitude of the luminance value, and The image binarization device according to claim 4 or 5, wherein a binarization threshold value of 2 is calculated. 7. An image processing apparatus comprising: a block dividing unit that divides a multi-valued image into blocks; wherein the average luminance value calculating unit calculates the local average luminance value based on the blocks divided by the block dividing unit. The image binarization device according to claim 1, wherein: 8. The apparatus according to claim 7, further comprising comparison value determination means for determining the comparison value based on an average luminance value of a block adjacent to a block for calculating the local average luminance value. Image binarization device. 9. A density value input means for inputting a density value of each pixel constituting a multi-valued image, and a local value including a pixel to be binarized based on the density value input by the density value input means. Average density value calculating means for calculating an average density value, and a binarization threshold higher than the local average density value according to the size of the local average density value calculated by the average density value calculating means. Binarization threshold setting means for setting, and binarization means for binarizing the pixel to be binarized with the binarization threshold set by the binarization threshold setting means, Image binarization device. 10. The binarization threshold setting means, wherein the correction value is a correction amount according to the local average density value calculated by the average density value calculation means, and the average density value is in a predetermined intermediate range. 10. The image processing method according to claim 9, wherein a binarization threshold is set by adding a correction amount that is larger than a value outside the intermediate range to the local average density value. Pricer. 11. The binarization threshold value setting means uses a correction amount of a smaller value for the local average density value calculated by the average density value calculation means as the average density value becomes larger. 10. The image binarization device according to claim 9, wherein the binarization threshold is set by multiplying the image. 12. A density value input means for inputting a density value of each pixel constituting a multi-valued image, and a local value including a pixel to be binarized based on the density value input by the density value input means. Average density value calculating means for calculating an average density value, and calculating a first binarization threshold value by adding a first correction amount to the local average density value calculated by the average density value calculating means. Addition threshold value calculation means, and multiplication threshold value calculation for calculating a second binarization threshold value by multiplying the local average density value calculated by the average density value calculation means by a second correction amount that is one or more. Means, a comparison determination means for determining whether the local average density value calculated by the average density value calculation means is larger or smaller than a predetermined comparison value, and the comparison determination means The magnitude of the local average density value is greater than the comparison value. If it is determined to be small, it is determined that the first binarization threshold calculated by the addition threshold calculation means is to be used, and the magnitude of the local average density value is determined by the comparison determination means. When it is determined that is larger than the comparison value, a binarization threshold determination unit that determines to adopt the second binarization threshold calculated by the multiplication threshold calculation unit; An image binarization device, comprising: a binarization unit that binarizes the pixel to be binarized according to the binarization threshold determined by the binarization threshold determination unit. 13. The method according to claim 1, wherein the adding threshold value calculating means adds the correction amount according to the magnitude of the density value to the local average density value calculated by the average density value calculating means. 13. The image binarization device according to claim 12, wherein one binarization threshold is calculated. 14. The multiplication threshold value setting means multiplies the local average density value calculated by the average density value calculation means by a correction amount corresponding to the magnitude of the density value. 14. The image binarization device according to claim 12, wherein a binarization threshold value of 2 is calculated. 15. An image processing apparatus comprising: a block dividing unit that divides a multi-valued image into blocks; wherein the average density value calculating unit calculates the local average density value based on the blocks divided by the block dividing unit. The image binarization device according to any one of claims 9 to 14, wherein: 16. The apparatus according to claim 15, further comprising comparison value determination means for determining said comparison value based on an average density value of a block adjacent to a block for calculating said local average density value. Image binarization device. 17. A binarization threshold value which is lower than an average luminance value of a local pixel set including a pixel to be binarized when a luminance value of a pixel constituting a multi-valued image is input. For the set of pixels, a binarization threshold having a smaller difference from the average luminance value than for the set of pixels not overexposed is determined, and the image of the local pixel set is converted to the binarization threshold. An image binarization method characterized in that the image is binarized and output using. 18. A binarization threshold value which is higher than an average density value of a local pixel set including a pixel to be binarized, which is input as a density value of a pixel constituting a multi-valued image. For the set of pixels, a binarization threshold having a smaller difference from the average density value than for the set of pixels that are not overexposed is determined, and the image of the local pixel set is subjected to the binarization threshold. An image binarization method characterized in that the image is binarized and output using. 19. A binarization threshold having a value lower than an average luminance value of a local pixel set including a pixel to be binarized, the luminance value of a pixel forming a multi-valued image being input to a computer. The binarization threshold is determined such that the difference from the average luminance value is smaller for the pixel set that is overexposed than for the pixel set that is not overexposed, and the image of the local pixel set is determined. Using the binarization threshold to output the image. 20. A binarization threshold having a higher value than an average density value of a local pixel set including a pixel to be binarized, in which a computer inputs a density value of a pixel forming a multi-valued image. The binarization threshold is determined such that the difference from the average density value is smaller for the pixel set that is overexposed than for the pixel set that is not overexposed, and the image of the local pixel set is determined. Binarizing the image using the binarization threshold and outputting the binarized image.