JP5025392B2 - Image processing apparatus and image processing method - Google Patents

Description

  The present invention relates to an image input device such as a digital camera or an image scanner, an image output device such as a display or color printer for displaying or printing an input image, and the gradation of the image according to the characteristics of the image input device or the image output device. The present invention relates to an image processing technique to be corrected.

  In an image input device such as a digital camera or image scanner equipped with an image sensor such as a CCD (Charge Coupled Devices), an image output device such as a display or a color printer that displays or prints an image input by the image input device, In general, image processing for correcting the gradation of an image in accordance with the characteristics of a printing device is performed in the apparatus. Among such image input / output devices, a conventional method will be described with respect to a procedure for correcting gradation in a digital camera.

  As the simplest and general configuration for correcting gradation, a gradation conversion characteristic as shown in FIG. 13 is defined in advance in a lookup table or the like, and a pixel to be image-processed (hereinafter referred to as a target pixel). ) Can be realized by referring to and outputting the converted value for the input value. However, when this method is used, in the luminance region where the input signal is less than Ix in FIG. 13, signal amplification exceeding 1 time is performed by gradation conversion, so that noise included in the image signal is also amplified simultaneously. was there.

  In addition, the target pixel value is separated by a threshold value for each signal level, and a low luminance low pass filter is applied to the low luminance portion and a high luminance low pass filter is applied to the high luminance portion while suppressing noise. A method for gradation conversion is proposed in Japanese Patent Laid-Open No. 06-113172.

  However, the conventional signal processing that performs tone conversion by the above method in the image input / output device applies a strong low-pass filter to suppress noise in the low luminance region where the signal amplification factor is large, and the resolution is lowered in the dark subject. However, there was a problem that the image was blurred.

Japanese Patent Laid-Open No. 06-113172

  The present invention has been made to solve such a problem, and an object of the present invention is to realize image processing that performs gradation conversion while suppressing noise amplification without reducing the resolution of a dark subject.

An image processing apparatus according to the present invention includes a line buffer unit that holds an input image signal for a plurality of lines, and a low-pass filter that extracts a plurality of image signals including a processing target pixel and its peripheral pixels from the line buffer unit and performs a filtering process. A first adding unit that subtracts the output value of the low-pass filter unit from the image signal value of the processing target pixel from the line buffer unit; and a first unit that performs gradation conversion on the output value of the low-pass filter unit . A gradation conversion means; a second addition means for adding the output from the first addition means to the signal value after gradation conversion; and a second floor for performing gradation conversion on the pixel to be processed from the line buffer means. A selector means capable of switching and outputting the output signal value from the tone converting means, the second adding means and the signal from the second gradation converting means in accordance with the input signal value;
The selector means refers to the gradation conversion characteristics defined by the first gradation conversion means and the second gradation conversion means, and the gain of the output value corresponding to the input signal value is less than 1.0. In this case, the signal from the second gradation converting means is selected, and when it is 1.0 or more, the output signal value from the second adding means is selected.

According to the image processing apparatus of the present invention, the low-pass filter unit performs noise removal processing on the target pixel to be processed and a plurality of pixels in the horizontal and vertical directions around the target pixel, and the low-pass filter processing data is processed. The tone conversion is performed by the first tone conversion means, and the difference value between the target pixel value and the noise-removed value after the low-pass filter processing is added to the tone conversion output value from the first tone conversion means. Output
On the other hand, the target pixel to be processed is subjected to gradation conversion by the second gradation conversion means and output, and is defined by the first gradation conversion means and the second gradation conversion means by the selector means. Referring to the gradation conversion characteristics, when the gain of the output value corresponding to the input signal value is less than 1.0, the signal from the second gradation conversion means is selected, and when the gain is 1.0 or more, the first value is selected. Since the output signal value from the second adding means is selected, in the luminance region where noise is amplified by non-linear gradation conversion, noise amplification in the conventional gradation conversion processing, or in the low luminance region to prevent a decrease in resolution, the effect noise can be carried out tone conversion processing for high resolution without being amplified there is, in the luminance region where noise by the tone conversion is suppressed by the general tone conversion Output with suppressed noise Effect there Ru that selectively obtain a.

Embodiment 1 FIG.
Embodiment 1 of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an image processing apparatus according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 1 denotes a line buffer means for holding data for a plurality of lines of an image, and 2 denotes noise removal using a plurality of horizontal pixels and a plurality of vertical pixels supplied from the line buffer means 1. Low-pass filter means 3 is a look-up table as gradation conversion means for performing gradation conversion on data after low-pass filter processing, and 4 is a target pixel to be processed in the low-pass filter means 2 supplied from the line buffer means 1 The first adding means for calculating the difference between the value and the value from which noise has been removed by the low-pass filter means 2, and 5 is the difference value between the target pixel calculated by the first adding means 4 and the value after the low-pass filter processing. Is added to the gradation conversion output value from the look-up table 3.

The operation of the image processing apparatus will be described below based on the block diagram of FIG.
For example, a case where a 3-tap filter of 3 horizontal pixels and 3 vertical pixels is used as a noise removal filter in the low-pass filter means 2 will be described. The line buffer means 1 holds image data for three lines, and pixel data for three taps centered on the pixel to be processed in the center line is transferred to the low-pass filter means 2. The contents of taps in the low-pass filter means 2 are shown in FIG. In FIG. 2, the center pixel Y11 corresponds to the target pixel to be processed for gradation conversion.

  In the low-pass filter means 2, for example, low-pass filter processing is performed according to equation (1) to calculate the output value Ylpf.

  The output value Ylpf of the low-pass filter means 2 is branched and supplied to the next stage processing. That is, one is supplied to the first addition means 4 for calculating the difference value between the target pixel value Y11 supplied from the line buffer means 1 and the output value Ylpf of the low-pass filter means 2, and the other is supplied to the lookup table 3. The

  From the look-up table 3, an output value LUT (Ylpf) obtained by performing gradation conversion processing on the output value Ylpf of the low-pass filter means 2 is output. The output value LUT (Ylpf) from the look-up table 3 is input to the second adding means 5 and added with the output value Y22-Ylpf from the first adding means 4 to obtain the final output Yout.

  The above processing is summarized as shown in Equation (2).

  3 to 5 show changes in signal characteristics when the gradation conversion process described in this embodiment is performed. In FIG. 3, (a) shows a state in which random noise is mixed into a pixel signal that is originally flat, and (b) shows a conventional simple floor to the random noise mixed pixel signal shown in (a). This is a result of applying the tone correction. FIG. 3B shows that noise is amplified. 4A shows the output signal Ylpf of the low-pass filter means 2, and FIG. 4B shows the LUT (Ylpf) signal obtained by performing tone correction on the output signal Ylpf of the low-pass filter means 2 by the look-up table 3. FIG. 5 shows the result of adding the output value Y22-Ylpf from the first addition means 4 to the signal LUT (Ylpf) after gradation correction, that is, the output signal of the second addition means 5. From FIG. 3B and FIG. 5, when tone correction is performed by the image processing apparatus according to the present embodiment, noise is suppressed from being amplified due to the influence of nonlinear tone correction, and after tone conversion. It can be seen that the S / N similar to that before conversion is obtained in the signal.

  As described above, according to the image processing apparatus of the present embodiment, conventionally, the tone conversion processing in which noise is amplified or the resolution of the low luminance region is reduced can be increased without increasing noise. There is an effect that can be implemented.

Embodiment 2. FIG.
The second embodiment will be described below with reference to the drawings. FIG. 6 is a block diagram showing an image processing apparatus according to Embodiment 2 of the present invention. In FIG. 6, reference numerals 1 to 5 have the same configurations as those in FIG. 6 is different from the first embodiment shown in FIG. 1 in that the multiplication means 6 is provided.

Next, the operation of the image processing apparatus in the second embodiment will be described.
In the second embodiment, the difference between the target pixel signal Y11 to be processed and the signal Ylpf from which noise has been removed by the low-pass filter unit 2 around the target pixel signal Y11 is calculated by the first addition unit 4 and then externally output by a register or the like. The multiplication means 6 multiplies a predetermined coefficient α given by An arithmetic expression of the gradation correction process in the second embodiment is shown in Expression (3).

  By configuring the image processing device as described above, for example, when performing gradation correction with a digital camera, if a low-sensitivity image captured with low sensitivity is input, a large value is given to the coefficient α to sharpen it. When a noisy image captured with high sensitivity is reproduced by reproducing an edge, it is possible to give a small value to the coefficient α and perform gradation correction with an emphasis on noise removal.

Embodiment 3 FIG.
The third embodiment will be described below with reference to the drawings. FIG. 7 is a block diagram showing an image processing apparatus according to Embodiment 3 of the present invention. In FIG. 7, 1 and 3 to 6 have the same configuration as that of FIG. 6 in the second embodiment. In FIG. 7, the configuration of the low-pass filter means 2 is different from that of the second embodiment shown in FIG. FIG. 8 is a diagram showing the internal configuration of the low-pass filter means 2 in FIG. In FIG. 8, 7 is an edge detection filter for detecting the presence or absence of an edge passing through the target pixel of processing target using each pixel of three taps centered on the target pixel of interest Y11 of FIG. 2, and 8 is each of three taps. This is a low-pass filter using pixels. 7 is different from the second embodiment in that the edge detection result of the edge detection filter 7 is input to the multiplication means 6 in FIG.

  Next, the operation in this embodiment will be described. Hereinafter, the operation of the low-pass filter means 2 which is a characteristic operation in the present embodiment will be described first. For example, a Sobel filter is applied to each 3-tap pixel input to the edge detection filter 7 to determine whether the target pixel to be processed is on the edge. An example of the coefficients of the Sobel filter is shown in FIG. In the Sobel filter, when the pixel value of FIG. 2 is multiplied by the coefficient of the same coordinate in FIG. 9 and added, the target pixel to be processed is on the edge, and when the addition result is small, the target of attention is Since it can be determined that the pixel is not on the edge, edge detection is possible. For example, when the filter coefficient in FIG. 9A is multiplied, it can be determined whether or not the target pixel to be processed is on the horizontal edge, and when the filter coefficient in FIG. 9B is multiplied, the target pixel to be processed is vertical. It can be determined whether or not it is on the edge. An equation for calculating the edge detection value Edge in the edge detection filter 7 is shown in Expression (4). Here, Kxy is a filter coefficient at coordinates (x, y).

  For example, the maximum edge strength Edge in each direction detected by the edge detection unit 7 is compared with a predetermined threshold THedge, and if it is larger than this, it is determined that the target pixel to be processed is on the edge. A signal (for example, “1”) instructing execution of multiplication using normal α is sent from the low-pass filter unit 2 to the multiplication unit 6, and if the pixel is smaller than a predetermined threshold THedge, the target pixel to be processed is on the edge The low-pass filter means 2 sends a signal (for example, “0”) instructing the multiplication execution with α = 0 to the multiplication means 6. As a result, the expression (3) is executed as it is in the edge region, and in the non-edge region, the first term for adding the noise level offset in the equation (3) is canceled, so the tone correction is performed on the low-pass filter value. LUT (Ylpf) is finally output.

  In the present embodiment, there is an effect that an edge region and a non-edge region are detected, and a gradation correction result in which only the edge is emphasized and noise is further suppressed according to the region can be obtained.

  In the third embodiment, the case where a Sobel filter is used as the edge detection filter 7 has been described. However, the present invention is not limited to this, and the same effect can be obtained as long as the processing can extract edge information in an image such as a Laplacian filter.

Embodiment 4 FIG.
Hereinafter, the fourth embodiment will be described with reference to the drawings. The block configuration of the fourth embodiment is FIG. 7 as in the third embodiment. However, the configuration of the low-pass filter means 2 is different from that of the third embodiment shown in FIG. FIG. 10 shows the configuration of the low-pass filter means 2 according to the fourth embodiment. The low pass filter means 2 of FIG. 10 is different from FIG. 8 in that the edge detection result in the edge detection filter 7 is input to the low pass filter 8.

  Next, the operation in this embodiment will be described. Hereinafter, the operation of the low-pass filter means 2 which is a characteristic operation in the present embodiment will be described first. The edge detection filter 7 detects, for example, whether the target pixel Y11 to be processed is on the edge and the edge direction when it is on the edge using the Sobel filter of FIG. In the low-pass filter 8, based on the detection result of the edge detection filter 7, for example, an optimum filter coefficient is selected from a set of low-pass filter coefficients shown in FIG. For example, when the detection value in the edge detection filter 7 exceeds the maximum value in FIG. 9A and the edge threshold THedge, it is determined that the target pixel to be processed exists on the horizontal edge, and the low-pass filter 8 in FIG. ) To select a filter coefficient that emphasizes the horizontal resolution and perform low-pass filter processing. If the detection value in the edge detection filter 7 exceeds the maximum value in FIG. 9B and the edge threshold value THedge, it is determined that the target pixel to be processed exists on the vertical edge, and the low-pass filter 8 in FIG. ) Is selected and the low-pass filter processing is performed. If the maximum value of the detection result of the edge detection filter 7 does not exceed the threshold value THedge, it is determined that the target pixel to be processed is in the non-edge region, and the noise removal-oriented filter coefficient in FIG. Use low pass filter.

  By configuring the image processing apparatus as described above, there is an effect that a noise removal effect can be sufficiently exerted in a non-edge region while noise near the edge is favorably removed while placing importance on the resolution.

  In the fourth embodiment, the case where the Sobel filter is used as the edge detection filter 7 has been described. However, the present invention is not limited to this, and the difference between the process that can extract edge information in the image, such as a Laplacian filter, and the upper and lower pixels of the target pixel to be processed The same effect can be obtained by combining processing that can detect the correlation direction by comparing the value and the difference value between the left and right pixels.

Embodiment 5 FIG.
The fifth embodiment will be described below with reference to the drawings. FIG. 12 is a block diagram showing an image processing apparatus according to Embodiment 5 of the present invention. In FIG. 12, 1 to 5 have the same configurations as those in FIG. 12 differs from FIG. 1 in that two look-up tables 3a and 3b and a selector 9 are provided. The lookup table 3 b performs general gradation conversion on the target pixel to be processed output from the line buffer unit 1, and the selector 9 outputs the output from the lookup table 3 b and the second addition unit 5. Select the output result of and output.
Note that the look-up table 3b is used as the second gradation converting means, and the gradation conversion characteristics thereof are the same as those of the lookup table 3a.

  Next, the operation in this embodiment will be described. Hereinafter, the operations of the two look-up tables 3a and 3b and the selector 9 which are characteristic operations in the present embodiment will be described. The gradation conversion method using 1, 2, 3a, 4 and 5 in FIG. 12 is the same as the operation described in the first embodiment. As an operation characteristic of the present embodiment, the look-up table 3b is the same as the gradation conversion characteristic defined in the look-up table 3a for the target pixel Y11 to be processed output from the line buffer means 1. Performs general gradation conversion by characteristics and outputs LUT (Y11). Further, the output value LUT (Ylpf) is output from the lookup table 3a as in the first embodiment, and Yout of the above equation (2) is output from the second adding means 5. Here, two values of the gradation conversion, Yout and LUT (Y11), are obtained, and one of them is selected using the selector 9 at the point Ix where the gain value in FIG. 13 becomes 1.0. Select the output value of and output. A specific operation in the selector 9 will be described below. Note that lx is given in advance from the gradation conversion characteristics.

(i) When Y11> Ix Output value: LUT (Y11)
(ii) When Y11 ≦ lx Output value: Yout

  By configuring the image processing apparatus as described above, in a luminance region where noise is amplified by nonlinear gradation conversion, an output with good noise suppression can be obtained by using the gradation conversion apparatus of the present invention. In a luminance region where noise is suppressed by gradation conversion, an output in which noise is suppressed by general gradation conversion can be selectively obtained.

  In the above-described embodiment, the configuration in which lx is given in advance has been described. However, the present invention is not limited to this, and a configuration in which the value is automatically calculated from the set values of the lookup tables 3a and 3b in which the gradation conversion characteristics are set may be employed.

  Moreover, although the said embodiment demonstrated the structure corresponding to Embodiment 1, it can be set as the structure corresponding to any form of Embodiment 2-4. In any form, a look-up table 3b for performing general gradation conversion is provided for the target pixel Y11 to be processed from the line buffer means 1, and the output of the look-up table 3b and the second addition means 5 are provided. It is sufficient to have a configuration including a selector 9 that selectively switches and outputs the output from the input value of the target pixel to be processed.

  Further, in all the above embodiments, the example of the 3-tap is shown as the low-pass filter means 2, but this is not limited to this, and an arbitrary size is set according to the image size and noise characteristics to be processed, A low-pass filter may be provided so as to be switched according to the detected frequency band of the edge.

  In the fourth embodiment and the fifth embodiment, an example in which a 3-tap Sobel filter is used as the edge detection filter 7 has been described. However, the present invention is not limited to this. An effect can be obtained.

  In all the embodiments described above, the configuration using the look-up table 3 in which the gradation conversion characteristics are described in advance as the gradation conversion means has been shown. However, this is not restrictive, and only representative points of the conversion curve are used. A method of registering and interpolating between points by linear interpolation or the like, or defining a conversion curve as a multi-order curve and calculating and outputting an output value during conversion processing may be adopted.

  The image processing technique of the present invention can be applied to an image input device such as a digital camera or an image scanner, an image output device such as a display or a color printer for displaying or printing an input image.

1 is a block diagram illustrating an image processing apparatus according to Embodiment 1 of the present invention. It is content explanatory drawing of the tap in a low-pass filter means. (a) is an original pixel signal characteristic diagram mixed with random noise, and (b) is a characteristic diagram obtained by gradation correction of the original pixel signal of (a) by a conventional method. (a) is an output signal characteristic diagram of the low-pass filter, and (b) is a characteristic diagram after gradation correction of the low-pass filter output signal. It is an output signal characteristic view of the 2nd adder. It is a block diagram which shows the image processing apparatus by Embodiment 2 of this invention. It is a block diagram which shows the image processing apparatus by Embodiment 3 of this invention. It is an internal block diagram of the low-pass filter means by Embodiment 3. In the explanatory diagram of the Sobel filter coefficient example, (a) is a coefficient for determining whether or not the target pixel is on the horizontal edge, and (b) is a coefficient for determining whether or not the target pixel is on the vertical edge. It is an example. It is an internal block diagram of the low-pass filter means by Embodiment 4. FIG. 6 is an explanatory diagram of an example of a filter coefficient at the time of low-pass filter processing. (A) is an example of a filter coefficient when the horizontal resolution is emphasized, (b) is an importance of the vertical resolution, and (c) is an example of a filter coefficient when the noise removal is emphasized. . It is a block diagram which shows the image processing apparatus by Embodiment 5 of this invention. It is a gradation conversion characteristic figure preserve | saved at the lookup table of a conventional apparatus.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1; Line buffer means, 2; Low-pass filter means, 3, 3a, 3b; Look-up table, 4; First addition means, 5; Second addition means, 6; Multiplication means, 7: Edge detection filter, 8 Low-pass filter, 9; selector.

Claims (5)

Line buffer means for holding an input image signal for a plurality of lines, low-pass filter means for extracting a plurality of image signals including a pixel to be processed and its peripheral pixels from the line buffer means, and filtering processing; and line buffer means First addition means for subtracting the output value of the low-pass filter means from the image signal value of the pixel to be processed; first gradation conversion means for performing gradation conversion on the output value of the low-pass filter means; and gradation conversion Second addition means for adding the output from the first addition means to the subsequent signal value, second gradation conversion means for performing gradation conversion on the pixel to be processed from the line buffer means, and second addition Selector means capable of switching and outputting the output signal value from the means and the signal from the second gradation conversion means according to the input signal value;
The selector means refers to the gradation conversion characteristics defined by the first gradation conversion means and the second gradation conversion means, and the gain of the output value corresponding to the input signal value is less than 1.0. In this case, an image processing apparatus is configured to select a signal from the second gradation converting means, and to select an output signal value from the second adding means when 1.0 or more. .   The second adding means comprises multiplying means for multiplying the output signal value of the first adding means added to the signal value after gradation conversion by the gradation converting means by a predetermined coefficient. Item 6. The image processing apparatus according to Item 1.   The low-pass filter unit is configured to extract a plurality of image signals including the processing target pixel and its peripheral pixels from the line buffer unit, perform edge detection and filter processing, and the multiplying unit receives the first addition unit. 3. An image processing apparatus according to claim 2, wherein the output signal value is multiplied by a coefficient based on the edge detection value detected by the low-pass filter means. The low-pass filter unit is configured to extract a plurality of image signals including a processing target pixel and its peripheral pixels from the line buffer unit, perform edge detection, select a filter coefficient based on the detected edge detection value, and perform filter processing The multiplier means is configured to multiply the output signal value from the first adder means by a coefficient based on the edge detection value detected by the low-pass filter means. Image processing apparatus. A holding step for holding the input image signal in a line buffer means for holding a plurality of lines, and a plurality of image signals including a pixel to be processed and its peripheral pixels are extracted from the line buffer means, and low- pass filter processing is performed by the low- pass filter means. A low-pass filter processing step to be performed; a first addition step of subtracting the output value after the low-pass filter processing from the image signal value to be processed extracted from the line buffer means by the first addition means; and an output value of the low-pass filter processing A first gradation conversion step for performing gradation conversion using the first gradation conversion means, and a first addition step for the signal value after the first gradation conversion processing by the second addition means . a second adding step for adding outputs from, floors to the processing target pixel from the line buffer unit by the second gradation conversion means A second gradation conversion step for converting comprises a second switching output possible selectors steps in accordance with the output signal value and the signal of the second gradation conversion step to an input signal value of the addition step by the selector means,
The selector step refers to the gradation conversion characteristics defined in the first gradation conversion step and the second gradation conversion step, and the gain of the output value corresponding to the input signal value is less than 1.0. In the image processing method , the signal from the second gradation conversion step is selected if the output signal value is 1.0, and the output signal value from the second addition step is selected if 1.0 or more .

Images