JP2000253255A - Signal processor and signal processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a high speed processing with a small device scale by storing a predetermined reference signal, detecting a difference value between an input signal and the reference signal, signal-processing the difference value, outputting an operation signal and correcting the input signal, based on the operation signal. SOLUTION: A difference detection circuit 101 calculates a difference value Δ(R, G and B) by using a prescribed relation expression based on an input signal Px (R, G and B) and the signal Pj (R, G and B) of an adjacent picture element, which is accumulated in a memory 104. A signal processing circuit 102 executes the operation of edge emphasis, smoothing and enlargement/ reduction by using the difference value and outputs an operation result Δ' (R, G and B). A correction circuit 103 corrects an input signal by using an operation result and outputs a correction result Px' (R, G and B). The difference value detected by the difference detection circuit 101 is used in common for respective signal processings. Thus, a memory access circuit is reduced and speed-up can be realized.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for processing image signals, and more particularly to an image signal processing apparatus for transmitting, storing and outputting a color image signal.

[0002]

2. Description of the Related Art Image data created by inputting using an image input device such as a scanner or a digital camera is
When performing reproduction using an image display device such as a printer or a monitor, many types of signal processing are performed on image data.

[0003] Scaling processing for increasing or decreasing the number of pixels,
Smoothing processing for the purpose of smoothing contours of character figures and the like, edge enhancement processing for the purpose of emphasizing shade boundaries, and the like are widely used. In addition, there is signal processing performed for the purpose of improving image quality, such as switching filter processing based on the magnitude of signal fluctuation of image data.

[0004] These enlargement / reduction processing, smoothing processing, edge enhancement processing, and the like are all image data having a plurality of pixels, and are obtained by measuring signal fluctuations of a plurality of pixels adjacent to a given arbitrary target pixel. Work based on.

That is, in the scaling process, a difference value between the signal level of the target pixel and the adjacent pixel is detected, and the signal level of the pixel to be newly set is set based on the scaling ratio. The smoothing process targets the contour of a character figure having a large difference in shading. In the edge enhancement processing, a difference value between adjacent pixels is replaced with a value amplified by a coefficient value. Further, the process of determining an area based on the type of appearing color may be performed by signal statistical processing.

As described above, many signal processings perform calculations using the difference between signal fluctuations. The same applies to a case where one pixel targets a color image represented by a plurality of types of color signals.

Performing such difference detection of image data independently for signal processing of each color type is widely used for reducing the scale of the apparatus since one signal processing apparatus can be used in common. ing.

[0008]

The conventional signal processing apparatus performs signal processing for a different purpose as described above.
In each case, a difference value between adjacent (reference) pixels is detected and used for calculation. Then, because the purpose of each signal processing is different, each of them is configured using means corresponding to different difference detection.

In the difference detection, it is necessary to read out signals of a target pixel and a plurality of adjacent pixels from a memory and perform an operation. For this reason, providing each of the difference detection means in each of the signal processing has caused problems such as an increase in the device scale and a long processing time. This means that the access to the image data memory connected to the outside of the CPU requires processing time for address calculation and memory access, both in a hardware configuration and in software processing using the CPU. It is an important issue to reduce the processing time in signal processing of a large amount of image data.

Conventionally, the amplitude value of an input signal (for example, 8
The configuration of the signal processing circuit is changed depending on the data resolution (bit, 16 bit, etc.). However, in order to use the circuit configuration or the software module in various application devices, it is important to not depend on the amplitude value of the input signal.

In addition, when a color image is targeted, if the difference detection is executed independently for each color, the image data obtained as a result of the signal processing may deteriorate in image quality. For example, if the detected edge position is independently detected for each color signal, the detected edge position may be shifted for each color signal, and as a result, a color shift occurs.
Further, when the color signal of the color image is sub-sampled, there is a problem that each pixel cannot reproduce an accurate color.

Accordingly, it is an object of the present invention to provide a signal processing device which is small in device size and capable of high-speed processing. Another object of the present invention is to provide a high-definition, high-definition signal processing apparatus without image degradation.

[0013]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a memory for storing a predetermined reference signal, and one difference detection unit for detecting a difference between an input signal and the reference signal. And one or more signal processing units that perform signal processing on the difference value and output a calculation signal, and a correction unit that corrects the input signal based on the calculation signal.

As described above, each signal processing of the signal processing device is performed by 1
By detecting the difference value with the two difference detection units, it is possible to realize a signal processing device having a small circuit size with a simple configuration without unnecessary circuits, and a low-cost, high-speed signal processing device. it can.

Further, by using the difference value, a signal processing device which does not depend on the amplitude value (data resolution) of the input signal can be realized.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Image data is composed of pixels arranged on a two-dimensional plane. The present invention is characterized in that signal processing is performed by mapping a signal of a pixel on the plane to a signal on a color space. In the following description, an embodiment of the present invention will be described for a color image. Although RGB is exemplified as a color signal, the type and number of color signals are not limited, and CMY, LUV, LAB, C
MYK and the like can be used.

In the conventional configuration, even in an apparatus that successively executes a plurality of types of signal processing, some memories and signal processing operate using independent circuit configurations. For example,
For edge enhancement, a reference pixel adjacent to the target pixel is accessed in memory, and an operation based on filter processing is performed.
For the smoothing process, a pattern matching operation is performed by accessing a reference pixel adjacent to the target pixel in memory. As described above, these signal processings are performed on the assumption that they have different functions. In this case, the circuit scale becomes large, and the processing speed also becomes slow. Hereinafter, the present invention will be described with reference to the drawings.

FIG. 1 shows a basic configuration diagram of a signal processing apparatus according to the present invention. Difference detection circuit 101, signal processing circuit 10
2, a correction circuit 103 and a memory 104. Here, the difference detection circuit 101 outputs the input signal Px (R,
G, B) and a signal Pj (R, G, B) of an adjacent (reference) pixel stored in the memory 104 to calculate a difference value Δ (R, G, B) using a predetermined relational expression. I do. The signal processing circuit 102 performs an operation such as edge enhancement, smoothing, and scaling using the difference value, and outputs an operation result Δ '(R, G, B). Correction circuit 103
Performs correction of an input signal using the above calculation result and outputs a correction result Px '(R, G, B).

FIG. 2 is another block diagram of the signal processing device of the present invention. Memory 104 for storing image data of reference pixel Pj (R, G, B), signal processing circuit 102
Is added, but the configuration is the same as the above basic configuration.

Here, the signal processing circuit 102 performs an operation by using a signal of a pixel distributed on a plane such as edge enhancement, smoothing, enlargement and reduction. All of them have been widely used for the purpose of signal processing. To briefly describe each signal processing, the edge enhancement amplifies a signal change between a target pixel and an adjacent pixel to enhance the appearance of an edge. Smoothing smooths the periphery of a pixel such as a character figure (that is, an outline with a large signal change). In the enlargement / reduction, a signal at a new pixel position based on the magnification is calculated from signal values of peripheral pixels. As described above, any signal processing detects and uses a signal change (difference) between the target pixel and the adjacent pixel.

A feature of the present invention is that the difference value detected by the difference detection circuit 101 is commonly used for each signal processing. That is, it is sufficient that only one difference value calculation circuit is provided. As a result, since the number of memory accesses can be reduced, the speed can be increased, and there is an advantage that the circuit scale can be reduced.

1 and 2, a plurality of signal processing circuits 102 for executing different types of signal processing can be easily connected in series. That is, the signal processing circuit 10
2 can be easily modularized (common components).

Hereinafter, the configuration of each circuit will be described.

(1) Difference Detection Circuit 101 First, the configuration and operation of the difference detection circuit 101 will be described. In an input color image, pixel signals are represented by a combination of a plurality of color signals. It is assumed that the signal of each pixel is represented by a coordinate point in a color space having, for example, an RGB signal on a coordinate axis.

FIG. 3A shows a target pixel x on a plane.
And the positional relationship between reference pixels a, b, and c. When a signal is input in a scanning order in line units, a pixel that has been subjected to signal processing can be used as a reference pixel. Alternatively, by using a line memory, it is possible to refer to a pixel (N + 1 line) preceding the pixel of interest. The difference value Δj (R, G, B) detected here is the target pixel x
, And reference pixels j (j = a, b, c
...). This difference value Δj can be represented by a vector in the RGB color space as shown in FIG. The vector Δj (R, G, B) representing the difference value corresponding to the reference pixel j can be calculated from each component of the color signal as in the following equation.

Δj (R, G, B) = ((Rx−Rj),
(Gx-Gj), (Bx-Bj)) The difference Δj between the reference pixels can be calculated in the same manner. Although the above is a vector having a color signal as a component, it can be represented by a length connecting coordinate points in a color space. The numerical value related to the length can be represented by the square of the length of the vector, or the signal having the largest value among the components of the vector can be represented as the difference. If these are represented by the formulas, Δj = (Rx−Rj) ＾ 2 + (Gx−Gj) ＾ 2 + (B
x−Bj) ＾ 2 Δj = SQRT ((Rx−Rj) ＾ 2 + (Gx−Gj) ＾ 2
+ (Bx−Bj) ＾ 2) Δj = MAX ((Rx−Rj), (Gx−Gj), (Bx
-Bj)) Alternatively, a brightness component can be extracted from the color signal and expressed. For example, Δj = (Rx−Rj) + (Gx−Gj) + (Bx−B
j) Δj = signal L ^* (brightness signal of uniform color space defined by the International Institute of Illumination CIE) where the operator “＾” is a square, “SQRT” is a square root,
“MAX” indicates maximum value extraction.

In an apparatus for sequentially scanning and inputting image data, there is a case where it is not necessary to detect a difference between each target pixel and an adjacent reference pixel in all directions. For example, when detecting a difference in the grid direction between the target pixel and four adjacent pixels, a new two-directional difference is detected in the sequentially input target pixel, and the four directions are combined with the already detected difference value. Can be used.

As shown in FIG. 4, the difference Δj is determined by the color signal Px input using the difference detection circuits 101a to 101e.
(R, G, B) and the color signal Pj (R, G, B) of the reference pixel read from the memory. The image data of the target pixel x and the reference pixel j (j = a, b, c...) Are input using the memory access circuit 108. In order to quickly calculate the difference value of the type of the reference pixel using the input signal, a plurality of difference detection circuits 101 are arranged in parallel. When the execution speed may be low, the same result can be obtained by operating the single difference detection circuit 101 a plurality of times. The plurality of types of difference values Δ (R, G, B) calculated in this way are output, and a value to be actually used is selected by an edge determination circuit 134 described later.

The pixel to be calculated by the difference detection circuit 101 is shifted by one pixel based on the scan direction of the target pixel. At this time, the image data of the pixel of interest is input, and the image data of the pixel to be newly referenced is stored in the memory 1.
Read from 04. FIG. 5 shows a configuration example of a memory access circuit 108 provided inside the difference detection circuit 101 for sequentially inputting image data of a target pixel and a reference pixel, and the delay registers 121a, 121b, 1
A signal to be operated is prepared using 21c and 121x. The address generation circuit 120 generates a memory address based on the order of pixel scanning. A new scan input signal is written as input / output data of the memory, and a signal of a new reference pixel is read. When each of the RGB colors is an 8-bit signal, the pixel signal Px (R, G, B) is input to and output from the memory as 24-bit width data. That is, it is only necessary to perform one memory write and one memory read for the signal processing of one target pixel.

The memory 104 stores data of reference pixels. For example, it is sufficient if there is a data capacity of two lines including the target pixel and the Nth line and the (N-1) th line. Furthermore, by accumulating the (N + 1) th line, it is possible to refer to the pixel preceding the target pixel.

Switching the signal processing method depending on the signal characteristics of the image data may be advantageous in terms of improving the image quality. For example, the edge portion of the image where the signal changes rapidly is a place to be visually noticed, so that the selection of the signal processing method is important. For example, as shown in FIG. 6, the smoothing circuit 131 and the edge emphasizing circuit 13
2. The signal processing such as the smoothing circuit 133 is selected by the selection circuit 135.

Here, as shown in FIG. 7, the edge determination circuit 134 receives a plurality of types of difference values Δ (R, G, B) detected by the difference detection circuit 101, and
Is calculated with the threshold value set in. Then, the content of the signal processing is selected based on the magnitude of the difference. An example of the relationship between the difference value and the threshold (when two thresholds are provided) is shown below. Here, the threshold value set in the setting value register 105 is calculated in advance for many sample images based on experiments.

When difference value Δ ≦ threshold value 1: smoothing circuit 1
Select the calculation result of 31. If the difference value Δ> threshold 1: select the calculation result of the edge emphasizing circuit 132. If the difference value Δ> threshold 2: select the calculation result of the smoothing circuit 133. The feature of 134 is that the difference value Δ is determined based on a combination of the RGB color signals of the pixel. If the determination is made separately for the RGB color signals, the determination position may be shifted for each color signal. On the other hand, according to the present invention, it is possible to obtain a determination result of an edge without deviation.

As described above, most of the signal processing that has been conventionally performed separately assuming a different purpose using the detected difference value Δ or the determination result of the edge position, etc., is eventually ended up in the present embodiment. Taking into account the relationship between the two-dimensional plane and the color space, it can be realized by collective signal processing.

(2) Edge Emphasis The edge emphasis process is a widely used technique for enhancing the visual appearance by amplifying a change in shading at an edge detection position. For example, when a printed material is input using a scanner, a signal change at an edge portion of image data may be deteriorated due to factors such as lens characteristics. This is converted into signal characteristics close to those of the original document by edge enhancement processing. However, conventionally,
Edge detection is performed independently for each color signal.

In the present invention, as a combination of color signals,
That is, the edge is detected as a vector in the color space. The edge detection result is a signal indicating a transition portion in a region having a different color. If the color of each region is represented on the RGB color space, the edge portion can be represented by a coordinate point on a vector connecting the regions. The edge detection result can be set by the length of the vector indicating the magnitude of the color change between the regions and the number of coordinate points (the number of pixels) on the vector indicating the gradient of the change. For example,
By adjusting the vector length, a processing result corresponding to a differential filter or an integral filter can be obtained. In addition, even if the vector length is the same, the inclination is large if the number of pixels on the vector is small, and the inclination is small if the number of pixels is large. As described above, the edge enhancement processing is executed by setting the vector length and the number of pixels.
This method has an advantage that color shift between RGB color signals does not occur.

FIG. 8 shows an embodiment of the edge enhancement circuit 132 as a specific example of signal processing for edge enhancement. For simplicity, an example (1) is shown in which a black and white image has a white area and a black area sandwiching an edge portion. FIG. 2 (2) amplifies the signal amplitude by extending the length of the vector, and corresponds to a differentiation process. FIG. 3C shows an example of the result of gradient correction by reducing the number of pixels located on a vector.

The present invention is characterized in that the above-described edge emphasis is collectively processed with other signal processing, thereby eliminating the cause of image quality deterioration and performing high-speed signal processing. Next, the enlargement process will be described as an example.

(3) Enlargement Enlargement / reduction calculates a signal at a pixel position corresponding to the magnification from a pixel signal of an original image.

Here, taking an example of a configuration in which image data captured by a digital camera is printed by a printer, the number of pixels is increased in order to print in accordance with the resolution of the printer.
That is, the enlargement processing will be described.

The principle of the enlargement processing is to calculate the position of a new interpolation pixel Z and the signal level determined by the magnification as shown in FIG. The magnification can be fixedly set in the horizontal and vertical directions, or can be arbitrarily changed based on the position of the pixel. The signal level of the new pixel can be calculated from the signal difference value of the adjacent pixel by interpolation. If this interpolation processing is expressed in the RGB color space, this corresponds to calculating a new coordinate point by dividing a vector connecting coordinate points in the color space based on the magnification. The coordinate points of the new interpolation pixel Z can be obtained by independent proportional distribution calculations for the RGB color signals.

FIG. 10 shows a configuration example of the interpolation coefficient calculation circuit 111 based on the magnification. Although the content of the enlargement processing is not limited, an example will be described below. To simplify the calculation, the magnification value is separated into an integer part and a decimal part. A new pixel based on the integer part of the magnification will be newly located between the input pixels originally located at the lattice points. The decimal part of the magnification is a so-called DDA (Digital Differential Analys
A technique called is) can be used. This is reflected in the magnification when the decimal part of the magnification is added over a plurality of pixels and carried to an integer value. For example, if the decimal part is 0.1, the addition is performed over 10 pixels to carry to an integer 1. Therefore, for the pixel at the time when the carry occurs, the integer part of the magnification is incremented by +1. By setting such a magnification, the magnification can be realized for the entire image.

Here, the image quality is improved by switching the content of the signal processing based on the magnitude of the signal fluctuation of the target image. Image data such as a natural image has relatively little fluctuation. On the other hand, image data made of a character figure such as a document has a large signal fluctuation, and the edge contour shape and the difference value are often significant. Edge correction or smoothing processing can be used to correct such characteristics of the edge portion. Here, the edge enhancement is the signal processing described above.

The smoothing process is a signal process for correcting the contour shape of the edge. As shown in FIG. 11, the signal level of the interpolation pixel Z which is surrounded by the input pixels a, b, c, x as shown in FIG. Decide. Therefore, the difference value Δxa between the pixel x and the pixel a calculated by the difference detection circuit 101
And the like, and using the pattern table 136, the target pixel x
Is calculated. By adding this value to the signal Px of the target pixel, the signal Pz of the new pixel Z is obtained.
= Px + Δzx can be calculated.

In the present invention, when the detected difference Δ is sufficiently large, it is assumed that the target image is a document composed of character figures, and a smoothing process is performed for the purpose of correcting the contour shape of the edge portion. The threshold used for the above determination can be a value close to the signal amplitude range of the image data. This is because, in the data of the generated document image, the character graphic often takes the maximum value of the signal amplitude. The smoothing process can be executed by pattern matching using the difference value detected by the difference detection circuit 101. The smoothing circuit 133 calculates the difference value Δ
Is sufficiently large, the pixel arrangement on the plane is determined by pattern matching, and the signal level of the pixel to be newly set is determined. The resolution of the output pixel can be set based on the set magnification. Specifically, such pattern matching can be configured by the pattern table 136. The pattern table 136 receives the difference value Δj (R, G, B) as an address signal and outputs an interpolation signal as data.
By consulting the table using the difference value Δj (R, G, B), a result independent of the absolute value of the signal level can be obtained. As a result, a smoothing result for the input signal can be obtained. If the difference is small,
Signal processing using the smoothing circuit 131 is performed assuming that the region has a small change in signal level. FIG. 12 shows an example of this circuit configuration. A signal related to the magnification output from the interpolation coefficient calculation circuit 111 and a difference Δ between pixels to be interpolated are input, and an interpolation value is output. Since this is a multiplication, it can be output using a conversion table. The smoothing circuit 131 operates so as to smoothly connect the signal levels of the adjacent pixels at a portion where no edge portion exists. When a new pixel is set based on the magnification, the signal level of the pixel can be calculated based on a conventionally known operation formula called a bilinear method. Specifically, the calculation is performed by distributing the difference value Δ between adjacent pixels based on the magnification without using the signal value itself as a calculation target. Also,
By using the difference Δ between pixels, an interpolation calculation method called a so-called bilinear method can be used.

(4) Smoothing processing + Edge emphasis + Smoothing processing These signal processings have conventionally been performed separately as functionally different contents. However, in the present invention, all of them are adjacent to the target pixel. By focusing on the fact that the processing is performed using pixel signals, batch processing is performed, thereby realizing miniaturization, high-speed operation, and high image quality of the circuit.

Referring to FIG. 13, a description will be given of a circuit configuration for performing signal processing on the smoothing processing, edge enhancement, and smoothing processing in consideration of the magnification at a time. This is because the interpolation coefficient calculation circuit 1 operates based on the magnification in the configuration of FIG.
11 is a device configuration combining the same. The operation of each circuit is as described above. Then, the interpolation coefficient calculation circuit 11
1 sets the contribution ratio between the target pixel and the reference pixel using the newly set position of the interpolation pixel, and transmits the result to the smoothing circuit 131, the edge enhancement circuit 132, and the smoothing circuit 133. The signal processing is performed by inputting the difference value Δ and the interpolation coefficient. The outputs of these arithmetic circuits are output to the selection circuit 13 operating based on the result of the edge determination circuit 134.
5 and output.

An operation example of the circuit configuration of FIG. 13 is shown in FIG.
This is shown in the timing chart of FIG. First, difference detection is performed using the input signal and the memory read signal. The writing of the input signal to the memory is performed at an appropriate timing until the input of the next pixel. A circuit that performs edge enhancement, smoothing, and smoothing simultaneously performs signal processing using the difference value and the set magnification. The signal processing method is selected using the detected magnitude and slope of the difference Δ, and output is performed. As described above, in the signal processing of the target pixel, the number of times of reading and writing of the memory can be reduced to about one each. Further, by operating the processing circuit for a plurality of pixels in a pipeline manner, the processing speed can be increased.

The operation example of the circuit configuration shown in FIG. 13 can be executed by software processing using a CPU as shown in the processing procedure shown in FIG.

As described above, the present invention is characterized in that signal processing is performed using the difference values of the color signals in common. By performing the signal processing using the difference value, the signal processing independent of the amplitude width and the absolute value of the input signal can be performed. For example, if the bit width of the input signal is from 8 bits to 12 bits
The same circuit configuration can be used even for bits. This feature is the same when performing signal processing by software.

(5) Pixel Interpolation Color image input using a single-chip CCD sensor is widely used in digital cameras and the like. In this sensor, color filters are arranged at two-dimensionally arranged pixels. For example, as shown in FIG. 16, the RGB filters are arranged at intervals, and the RGB signals whose positions are shifted in a plane are read. That is, when a single-plate color CCD sensor is used, the output signals of a plurality of
The B signal is imaged. Even with such a sensor,
In the image area where the color signal change is small, the RGB
The color of the area can be determined by a combination of the color signals.
However, when there is a color change in the object to be photographed, the combination of color signals to be read is shifted depending on the positional relationship of the RGB filters.

For example, FIG. 17 shows an example in which a monochrome image is input for simplicity. FIG. 8 shows the correspondence between the sensor position and the edge signal, but this figure shows the relationship between the color filter corresponding to the reading position of the color CCD sensor and the reading signal. A white region and a black region are adjacent to each other, and there is a signal change region (edge portion) between the two. In the uniform white area, the color signals (Rw, Gw, Bw)
In a uniform black area, color signals (Rk, Gk, Bk) are output. However, at the edge where the signal changes from white to black,
A color signal that does not exist is output depending on the positional relationship with the RGB filter. For example, even when a black and white document image or the like is to be input, a color that does not exist in the edge portion is generated. Although such a decrease in the photographing resolution can be avoided by using an independent CCD sensor for each type of color filter, the cost of the apparatus increases. The present invention shows a circuit configuration used to improve the resolution of color image data input by a single-chip CCD sensor using the color signal obtained by the above pixel interpolation using the difference value detected by the difference detection circuit.

In a document or the like, the types of appearing colors can be limited in advance. In this method, the color type to be applied in advance can be limited by a method such as performing statistical processing on the result of pre-scanning an image, or setting the type of a document according to an instruction of an operator. Then, the image can be a combination of regions represented by limited colors, and the edge portion can be a transition portion of the region. If represented in the RGB color space, the limited color of each area is represented by a coordinate point, and the edge between the areas is located on a vector connecting the coordinate points. By interpreting the signal in the color space, the edge portion can be represented by a color signal without color shift. Here, an example of specific signal processing for reading a target image with high definition using an RGB color sensor will be described below. The edge position is determined by converting the RGB color signal into a signal for distinguishing the color region of the target image.

First, it is assumed that the appearance color of the target image is known in advance by statistical processing based on the prescan result of the image or by an instruction from the operator. For example, when a monochrome document image is captured by a single-chip color CCD sensor, the conversion result can be designated in advance as a monochrome signal.

Next, sensitivity correction is performed on the color signal input from the CCD sensor. For example, when the target image is created from white and black areas, the read RGB signal is converted into a signal corresponding to black and white. In addition, each sensor has a variation in sensitivity of a sensor element and a variation in characteristics of an RGB filter. In order to perform such signal conversion and variation correction, for example, as shown in FIG. 18, correction is performed using a conversion table. The principle of signal conversion is to normalize an input signal by the amplitude value of the converted signal.

A case where an input RGB signal is converted into a monochrome signal will be described. The coordinate points of the minimum and maximum values of the black and white signal on the RGB color space are defined as (R0, G0, B0) (R1, G1, B
1), the input signals P (R), P for RGB
For (G) and P (B), the conversion result can be calculated by the following equation.

P (R ') = (P (R) -R0) / (R1
−R0) × 255 P (G ′) = (P (G) −G0) / (G1−G0) × 2
55 P (B ′) = (P (B) −B0) / (B1−B0) × 2
55 The relationship of the above equation can be converted at high speed by putting it into a conversion table. If such sensitivity correction is not performed, sensitivity variation will be included in a signal for each pixel read by a discrete color filter, making it difficult to determine an edge portion. Further, in the above conversion table, correction factors that take into account the sensitivity variation of the sensor element for each pixel or the pseudo characteristics of the optical system are set for each pixel, thereby correcting various factors. It can also be executed collectively. As a result, the read signals from all the pixels can be used as black and white signals. Since the actual RGB filters have overlapping spectral characteristics, even if a signal in the red region of the screen is input, for example, all of the RGB signals change. For this reason,
By performing the sensitivity correction, a combination of multiple colors can be distinguished.

In the above example, the case where the entire screen is a black-and-white image has been described. However, it is also possible to switch the converted signal depending on the position in the screen when the appearance color differs or the screen position. For example, when an image such as black characters on a white background and red characters on a white background is input in another part, signal processing can be performed in the same manner by switching the conversion table for each partial area. As a result, signal reading with a resolution determined by all pixels of the CCD sensor can be realized. As shown in FIG. 18A, this can be realized by inputting coordinates specifying the area of the partial area to the switching device 141 and performing the conversion table 140. The conversion table can be replaced by a circuit that executes an arithmetic expression or a processor that executes software.

By the above conversion, the read signal from the sensor element located at the edge is also converted into a signal that changes between the color regions sandwiching the edge. In terms of the color space, the color signal of the edge portion is set on a vector connecting the coordinate points of the limited color of the area. For example, when converting an input RGB signal to a black and white signal, the conversion result of the pixel at the edge portion is a color intermediate between white and black. Thus, the color signal of the pixel at the edge can be set without color shift.

In the above description, a black-and-white image is read for the purpose of explanation. However, high definition can be realized in reading a color image. For this purpose, it is assumed that the image is decomposed into a plurality of partial areas, and each partial area is composed of an edge portion and two color regions sandwiching the edge portion. The resolution of the input image can be improved by performing the above-described color signal conversion. This can be performed by setting a color area for a partial area around the edge. The specific procedure is as follows: RGB input from CCD sensor
Edge detection is performed by performing difference detection on each of the signals and comparing the magnitude of the detection result with a threshold value.

Combining the detected edge part and its surrounding area to set a partial area as a unit of signal conversion. In this way, the target image is divided into a plurality of partial areas. -For each partial area, a conversion coefficient for sensitivity correction is set using the color signal of the edge area peripheral area. The conversion coefficient is stored in the conversion table 140.

Converting the input RGB signals for each partial area.

Such a processing procedure is repeatedly executed in a partial area around the edge in the screen. In this way, even on a screen where a plurality of types of colors appear, signals can be read at the resolution of the number of CCD pixels.

In the region where no edge portion is detected, it is determined that there is no color change, and pixel signals of color filters arranged at intervals are combined. In this manner, by performing signal processing on a vector connecting coordinate points in a color space, the processing can be performed in combination with processing for increasing the number of pixels by enlargement processing or the like.

In this way, it is possible to generate the same number of RGB read signals as the number of pixels of the sensor from the read signals of the CCD sensor using discrete RGB filters. FIG. 18B shows a CCD sensor 154, an AD converter 155, a screen memory 156, a memory 104, a difference detection circuit 101, a partial area setting circuit 143, and a signal conversion circuit 1.
42 shows an apparatus configuration composed of 42. 1 and 2, the partial area setting circuit 143 is the signal processing circuit 102, and the signal conversion device 142 is
Equivalent to 3. The CCD sensor 154 is a single-plate type color CCD sensor element, and the screen memory 156 stores at least one screen of color image data. Distortion such as shading specific to the CCD sensor can be corrected before being stored in the screen memory 156. To set a partial area of the whole screen,
First, the image data stored in the screen memory 156 is read, edge detection is performed using the detection result of the difference detection circuit 101, and signal processing for setting a partial area is performed by the partial area setting device 143. Using the result,
The image data is read out again from the screen memory 156, and the signal conversion device 142 executes the conversion process and outputs it. By incorporating or combining this circuit in an imaging device such as a digital camera, high-resolution imaging data can be obtained. Further, these signal processings can be created by a program and executed by software processing using a signal processing processor.

Using the same procedure, high-definition image data can be generated by performing signal processing on image data obtained by sub-sampling color signals. This is for example
It can be applied to high definition of (Y, Cr, Cb) signals and the like by the JPEG compression method. Where Y is the brightness signal, Cr and Cd
Represents a color difference signal.

(6) Error Diffusion Since error diffusion is a widely used signal reproduction method, a description of a basic processing procedure is omitted. This is a method suitable for reproducing a uniform level with a limited reproduction signal level, but is not necessarily suitable for an edge portion. In the error diffusion, as shown in FIG. 19A, an error signal generated as a calculation result of the error calculation circuit 106 is converted into an input signal P (R,
G, B) are stored in the memory 10 for correction by the correction circuit 103.
4 is characterized by being returned to. However, as a side effect of the feedback signals E (R), E (G), and E (B), missing dots may occur at an edge portion. For example, when there is a black character on a gray background, an error signal generated in the background portion may be fed back to the black character portion, so that a white dot may enter the edge portion of the black character. The present invention improves the image quality by restricting the feedback of the error signal, particularly for the pixel in which the edge portion is detected by using the difference between the adjacent pixels, particularly for the error diffusion processing of the multi-value output. Can be realized.

As shown in FIG. 19B, the feedback signal E
(R), E (G), and E (B) are temporarily stored using the memory 104 and then referred to for a plurality of pixels. In the present invention, a switch 107 for turning on / off a feedback signal by a control signal based on a difference value is provided. This control signal operates so as to turn off when the difference value is larger than a predetermined threshold value and to turn on when the difference value is smaller than the threshold value to execute a normal error diffusion process. The fact that the difference is large means that the position is an edge portion, and an output signal without fluctuation can be obtained by limiting the feedback signal. The fact that the signal level can be maintained at the edge portion visually has a kind of edge enhancement effect. This method is particularly effective when combined with multi-level error diffusion.

Further, by using the correction processing of the input signal using the feedback signal in the error diffusion processing with the correction circuit 103 having the above-described circuit configuration, the circuit can be downsized.
FIG. 20 shows the common use of the correction circuit 103. This can be shared with signal correction using feedback signals E (R), E (G), and E (B) for error diffusion processing. By adjusting the shift of the processing time using the delay register 121, the correction to the color signal of the same pixel is executed.

In the error diffusion processing, the positional relationship between the target pixel for calculating the error signal and the reference pixel is the same as that of the above-described difference detection circuit. That is, a feedback signal is calculated from a plurality of reference pixels. Therefore, the feedback signal E
A memory for temporarily storing (R), E (G), and E (B) is prepared. This memory may be prepared independently, but since it is accessed sequentially based on the signal processing order of the pixels, the memory 104 used in the difference detection circuit 101 and the memory address generation circuit 120 are used in combination. Can be configured. The memory 104 and the correction circuit 103 described above are shared, and the error calculation circuit 1
FIG. 21 shows an embodiment of a signal processing device in which No. 06 is combined. Then, the speed can be increased by simultaneously or consecutively performing both memory accesses. For this purpose, the data structure used for the memory input / output in the above processing is configured as a plurality of data arrays with consecutive addresses.
Then, by performing burst transfer from the address of the head data, data necessary for signal processing of the pixel can be transferred from the memory. Then, the input data is used in the signal processing unit of each unit.

(7) Digital Camera / Printer Direct Connection Configuration An image input device called a digital camera using a CCD sensor is widely used. In addition, printers that can perform color printing, such as ink jet printers, have become widespread. These are connected, and the photographed image data is printed using a printer. In order to print out with such a circuit configuration, edge enhancement,
Enlargement, smoothing, color conversion, error diffusion processing, etc. are indispensable. According to the present invention, these signal processes can be executed at high speed with a simple circuit configuration. For example, for color conversion, color signal conversion can be realized by mapping a coordinate point in a color space to which a color signal of image data is mapped to a different color space. For example, R
Although there is a method of executing a simple arithmetic expression for converting a GB signal into a CMY signal, mapping in a space represented by a combination of color signals is indispensable in order to reflect non-linear characteristics depending on a device. is there. The data used for this mapping is called, for example, a profile. Many conventional signal processing processes RGB signals in separate procedures, but the present invention, as described above, uses
It operates with a combination of B color signals. Therefore, the color conversion process is also suitable for batch processing using a combination of RGB signals. FIG. 22 shows an embodiment of a signal processing device in which the correction circuit 103 and the color conversion circuit are combined. As shown in FIG. 22, the signal Δ ′ output from the signal processing circuit 102
(R, G, B) is input to the correction circuit 103, and the resulting combination of RGB signals is
09 to output a CMY signal obtained by a mapping process on a color space. The color conversion circuit 109 can be configured by a memory circuit that uses input color signals as address values and uses read data as conversion results. Alternatively, it can be constituted by an arithmetic circuit based on a color conversion equation. Further, as shown in FIG. 23, a color conversion circuit 109 and an error calculation circuit
It is also possible to realize a signal processing device configuration to which the 106 is connected. The operation of each signal processing circuit is as described above.

By the signal processing according to each of the above embodiments, FIG.
As shown in FIG. 4, a digital camera 150, a WWW (World Wide Web) 151, a scanner 152, and the like can be switched and printed on a printer 153 as various types of image sources. In these image sources, since the type of color signal, the number of pixels, the signal level, and the like are different, the operation is performed by rewriting the set value. These setting values can be obtained by communication with a device serving as an image source.

As described above, the circuit configuration of the present invention realizes simplification of the circuit configuration by sharing the difference detection circuit. Here, the signal processing of the present invention can be realized by hardware or software processing of a personal computer. And the same high speed can be realized.

[0074]

According to the above configuration, it is possible to provide a signal processing device having a small device size and capable of high-speed processing. Further, by performing the operation using the difference value, it is possible to realize a modularized circuit configuration or software that does not depend on the amplitude value of the signal.

Further, it is possible to provide a high-quality, high-definition signal processing apparatus without image deterioration.

[Brief description of the drawings]

FIG. 1 is a diagram showing one embodiment of a basic configuration of a signal processing device of the present invention.

FIG. 2 is a diagram showing another embodiment of the basic configuration of the signal processing device of the present invention.

FIG. 3 is a diagram illustrating the operation of the difference detection circuit of the present invention.

FIG. 4 is a diagram showing one embodiment of a configuration of a difference detection circuit of the present invention.

FIG. 5 is a diagram showing one embodiment of a configuration of a memory access circuit of the present invention.

FIG. 6 is a diagram showing one embodiment of a signal processing circuit of the present invention.

FIG. 7 is a diagram showing one embodiment of an edge determination circuit of the signal processing circuit of the present invention.

FIG. 8 is a diagram illustrating the processing of the edge enhancement circuit of the signal processing circuit of the present invention.

FIG. 9 is a diagram showing the principle of the enlargement processing of the signal processing device of the present invention.

FIG. 10 is a diagram showing an embodiment of an interpolation coefficient calculation circuit of the signal processing circuit of the present invention.

FIG. 11 is a diagram showing an embodiment of a smoothing circuit of the signal processing circuit of the present invention.

FIG. 12 is a diagram showing one embodiment of a smoothing circuit of the signal processing circuit of the present invention.

FIG. 13 is a diagram showing another embodiment of the signal processing circuit of the present invention.

14 is a timing chart illustrating the operation of the signal processing circuit of FIG.

FIG. 15 is a flowchart illustrating the operation of the signal processing circuit of FIG. 13;

FIG. 16 is a diagram illustrating pixel interpolation using a CCD sensor in the signal processing device of the present invention.

FIG. 17 is a diagram illustrating signal correction of an edge portion in the signal processing device of the present invention.

FIG. 18 is a diagram illustrating pixel interpolation using a conversion table in the signal processing device of the present invention.

FIG. 19 is a diagram showing an embodiment of a signal processing device provided with a correction circuit and an error calculation circuit according to the present invention.

FIG. 20 is a diagram showing one embodiment of a configuration of a correction circuit of the signal processing device of the present invention.

FIG. 21 is a diagram showing an embodiment of the signal processing device of the present invention.

FIG. 22 is a diagram showing another embodiment of the signal processing device of the present invention.

FIG. 23 is a diagram showing another embodiment of the signal processing device of the present invention.

FIG. 24 is a diagram showing one embodiment of a signal processing system according to the present invention.

[Description of Signs] 101, 101a, 101b, 101c, 101d, 1
01e: difference detection circuit, 102: signal processing circuit, 103
... Correction circuit, 104, memory, 105, set value register, 106, error calculation circuit, 107, switch, 108
... memory access circuit, 109 ... color conversion circuit, 111 ...
Interpolation coefficient calculation circuit, 120... Address generation circuit, 12
1, 121a, 121b, 121c, 121x ... delay register, 131 ... smoothing circuit, 132 ... edge emphasis circuit,
133 ... smoothing circuit, 134 ... edge judgment circuit,
135: selection circuit, 136: pattern table, 140: conversion table, 141: switching device, 142: signal conversion device, 143: partial area setting device, 150: digital camera, 151: WWW, 152: scanner, 153: printer, 154: CCD sensor, 155: A / D converter,
156 Screen memory.

 ──────────────────────────────────────────────────の Continued on the front page F term (reference) 5B057 CA01 CA12 CA16 CB01 CB12 CB16 CD05 CE03 CE05 CE11 CE17 CH11 DB02 DB06 DC16 5C077 LL18 LL19 MP07 NN11 PP02 PP03 PP15 PP20 PP32 PP33 PP37 PP42 PP47 PQ08 PQ22

Claims (8)

[Claims] A memory for storing a predetermined reference signal; one difference detection unit for detecting a difference value between an input signal and the reference signal; a signal processing unit for processing the difference value to output an operation signal A signal processing device comprising: one or more signal processing units; and a correction unit that corrects the input signal based on the operation signal. 2. The signal processing device according to claim 1, wherein the signal processing unit has an edge emphasizing unit that visually clarifies an edge portion of the input image. 3. The signal processing device according to claim 2, wherein the signal processing unit has an enlargement processing unit that enlarges the input image by changing a magnification of the input image. 4. The signal processing device according to claim 1, wherein the signal processing unit including a plurality of signal processing circuits has a selection unit that selects an arbitrary signal processing circuit from the plurality of signal processing circuits. 5. The signal processing device according to claim 1, further comprising: an error diffusion unit that calculates an error of the correction signal corrected by the correction unit, outputs an error signal, and feeds back the error signal to the correction circuit. Signal processing device. 6. The signal processing device according to claim 1, further comprising a color conversion unit that converts the correction signal corrected by the correction unit into a different color signal. 7. The signal processing device according to claim 6, wherein the color conversion unit converts an RGB color signal into a YMC color signal. 8. A memory for storing a predetermined reference signal, one difference detector for detecting a difference value between an input signal and the reference signal, signal processing of the difference value, and outputting an operation signal. A signal processing system having one or more signal processing units, a correction unit that corrects the input signal based on the operation signal, an input unit that reads an input image, and an image recording unit that records the input image. Signal processing system.