JP2010050607A - Image processor, and noise reducing method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor capable of improving an S/N by more reducing noise of an image signal, and to provide a noise reducing method and a program. <P>SOLUTION: When m is ≥0. 5 and <1, brightness values of every pixel included in the image signal are integrated over a plurality of frames, and the integration result is divided by the m-th power of the number of frame integration. Otherwise, a brightness value of each pixel included in the image signal is integrated by a brightness value of each pixel in a prescribed range from the position of the pixel, and the integration result is divided by the m-th power of the number of integration pixels. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, a noise reduction method, and a program for reducing noise in an image signal.

  In recent years, imaging devices such as digital cameras, mobile phones, and video cameras, or image display devices such as displays and projectors have been reduced in size, increased in pixel count, and increased in functionality, and dedicated to high-speed CPUs and required processing. Various circuits such as LSI to be executed are mounted. For this reason, in an imaging device, an image display device, and the like, there are problems that noise generation sources increase, and the quality of a reproduced image is deteriorated by superimposing the generated noise on the captured image signal and the image signal to be displayed.

  Various methods for reducing noise superimposed on an image signal have been conventionally studied. For example, a frame in which a luminance value for each pixel is integrated over a plurality of frames and the integration result (frame integration value) is divided by the number of frame integrations. Integration, or pixel integration that integrates the luminance value of each pixel with the luminance value of each pixel in a predetermined range (average value filter) from the pixel position, and divides the integration result (pixel integration value) by the pixel integration number is known. It has been.

  A technique for improving the S / N of an image signal using frame integration is described in Patent Document 1, for example. A technique for improving the S / N of an image signal using pixel integration is described in Patent Document 2, for example.

  The following formula (1) is used by the background art image processing apparatus for frame integration.

Here, k is the horizontal pixel number (1, 2, 3,..., Kmax), l is the vertical pixel number (1, 2, 3,..., Lmax), and i is the frame number (i = 1, 1). 2, 3,..., N), n represents the number of frame integrals. Y (k, l) represents the luminance value after frame integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, and x _i (k, l) represents the i-th pixel. The luminance values before frame integration of the kth pixel in the horizontal direction and the lth pixel in the vertical direction are shown. Here, kmax is the number of pixels in the horizontal direction, and lmax is the number of pixels in the vertical direction.

  Also, the equation (2) below is used by the background art image processing apparatus for pixel integration.

Here, k is the horizontal pixel number (1, 2, 3,..., Kmax), l is the vertical pixel number (1, 2, 3,..., Lmax), and h is the horizontal direction in the average value filter. I is the pixel number in the vertical direction in the average value filter, j is the number of pixels in the vertical and horizontal directions of the average value filter (j × j pixels), and n is the number of pixel integrations (j ² ) Yes. Y (k, l) represents the luminance value after pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, and x (k, l) represents the k-th pixel in the horizontal direction. The luminance value before pixel integration of the l-th pixel and the l-th pixel in the vertical direction is shown. Here, kmax is the number of pixels in the horizontal direction, and lmax is the number of pixels in the vertical direction.
JP 2008-214662 A JP 2006-074190 A

  FIG. 6 shows the relationship of the S / N ratio of the image signal after frame integration with respect to the number of frame integrations obtained by the applicant as an experimental result for the frame integration of the background art described above.

  FIG. 6 shows image signals of the image sensors (CMOS image sensors) a to e after frame integration with respect to the frame integration number when the S / N value when the frame integration number n is 1 is normalized to 1. The measured values of S / N are respectively shown.

  When the noise superimposed on the image signal is assumed to be white noise, the S / N of the image signal after frame integration is theoretically improved in proportion to √n (n is the number of frame integration). “WN” shown in FIG. 6 indicates this theoretical characteristic.

  As shown in the experimental results of FIG. 6, the S / N of the image signals of the image sensors a to e after the frame integration is almost improved even if the number of frame integrations is increased to 16 or more regardless of the characteristic variation due to individual differences. Not. This is considered to be one factor that noise due to quantization error caused by image signal processing cannot improve S / N. Further, it is considered that semi-fixed pattern noise such as 1 / f noise generated due to the imaging element being a semiconductor element is included in the image signal. Such a phenomenon in which the S / N is not improved even if the number of integrations is increased also occurs in the pixel integration described above.

  The present invention has been made to solve the above-described problems of the prior art, and provides an image processing apparatus, a noise reduction method, and a program that can further reduce the noise of an image signal to improve the S / N. The purpose is to provide.

In order to achieve the above object, an image processing apparatus of the present invention is an image processing apparatus for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
A processing device that integrates the luminance value for each pixel included in the image signal over a plurality of frames, divides the integration result by the mth power of the frame integration number, and outputs the luminance value for each pixel after the division;
A memory for storing the image signal;
Have

Alternatively, an image processing apparatus for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
The luminance value for each pixel included in the image signal is integrated with the luminance value of each pixel within a predetermined range from the pixel position, the integration result is divided by the mth power of the pixel integration number, and the pixel value for each pixel after the division is A processing device for outputting a luminance value;
A memory for storing the image signal;
Have

The noise reduction method of the present invention is a noise reduction method for reducing noise of an image signal,
When m is 0.5 or more and less than 1,
Integrating a luminance value for each pixel included in the image signal over a plurality of frames;
Dividing the integration result by the mth power of the frame integration number;
This is a method of outputting the luminance value for each pixel after the division.

Or a noise reduction method for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating the luminance value of each pixel included in the image signal with the luminance value of each pixel within a predetermined range from the pixel position;
Dividing the integration result by the mth power of the pixel integration number;
This is a method of outputting the luminance value for each pixel after the division.

The program of the present invention is a program for causing a computer to execute processing for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating a luminance value for each pixel included in the image signal over a plurality of frames;
Dividing the integration result by the mth power of the frame integration number;
This is for outputting the luminance value for each pixel after the division.

Or a program for causing a computer to execute processing for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating the luminance value of each pixel included in the image signal with the luminance value of each pixel within a predetermined range from the pixel position;
Dividing the integration result by the mth power of the pixel integration number;
This is for outputting the luminance value for each pixel after the division.

  According to the present invention, it is possible to reduce the noise and improve the S / N of the image signal as compared with the image processing apparatus of the background art.

Next, the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing the configuration of the image processing apparatus according to the first embodiment.

  As shown in FIG. 1, the image processing apparatus according to the first embodiment can be realized by a processing device (computer) 10 including a CPU 11 that executes processing according to a program. The CPU 11 and the processing of the CPU 11 are necessary. A main storage device 12 that temporarily stores information; a recording medium 13 that stores a program for causing the CPU 11 to execute a required process; a data storage device (memory) 14 that stores image signals in units of frames; To transmit / receive data to / from various devices included in the memory control interface unit 15 that controls data transfer with the main storage device 12, the recording medium 13, and the data storage device 14, and the device in which the image processing device shown in FIG. 1 is incorporated. The plurality of interface units 16 (two in FIG. 1). The CPU 11 is connected to the memory control interface unit 15 and the interface unit 16 via a bus 18. Although FIG. 1 shows a configuration example in which the processing apparatus 10 includes a data storage apparatus 14 that stores image signals in units of frames, the data storage apparatus 14 is provided outside the processing apparatus 10 independently. Also good.

  The image processing apparatus executes processing for reducing noise of an image signal described below by executing processing by the CPU 11 in accordance with a program stored in the recording medium 13. The recording medium 13 may be a magnetic disk, a semiconductor memory, an optical disk, or other recording medium.

When the image signal is an analog signal, the image processing apparatus shown in FIG. 1 includes an A / D (Analog to Digital) converter for converting the image signal into a digital signal and analog data after processing. D / A (digital to
(Analog) converter may be provided.

  Further, the image processing apparatus can be realized by a logic circuit including an LSI or the like that performs arithmetic processing as long as the noise reduction processing described below can be executed.

  Next, the operation of the image processing apparatus according to the first embodiment will be described with reference to the drawings.

  The image processing apparatus according to the first embodiment is an example in which the above-described frame integration is performed as a process for reducing noise.

However, in this embodiment, the frame integration value shown in the above equation (1) is divided by √n instead of dividing by the frame integration number n. When the frame integration value is divided by √n in this way, the signal component S is multiplied by √n, so that the S / N of the image signal after the frame integration becomes closer to the theoretical characteristic WN shown in FIG. Dividing the frame integral value by √n is desirable because noise is most reduced. However, the frame integral value is not limited to √n, and may be divided by ^nm (0.5 ≦ m <1). That is, in this embodiment, frame integration is executed using the following equation (3).

Here, k is the horizontal pixel number (1, 2, 3,..., Kmax), l is the vertical pixel number (1, 2, 3,..., Lmax), and i is the frame number (i = 1, 1). 2, 3,..., N), n represents the number of frame integrals. Y (k, l) represents the luminance value after frame integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, and x _i (k, l) represents the i-th pixel. The luminance values before frame integration of the kth pixel in the horizontal direction and the lth pixel in the vertical direction are shown. Here, kmax is the number of pixels in the horizontal direction, and lmax is the number of pixels in the vertical direction.

  FIG. 2 shows S / N measurement values of image signals of image sensors (CMOS image sensors) a to e after frame integration with respect to the number of frame integrals when the frame integral value is divided by √n. In FIG. 2, as in FIG. 6, the S / N measurement of the image sensors a to e with respect to the frame integral number when the S / N value when the frame integral number n is 1 is normalized to 1. Each value is shown. Further, “WN” shown in FIG. 2 indicates the S / N theoretical characteristic with respect to the number of frame integrals when the noise is white noise.

  As shown in FIG. 2, by dividing the frame integration value by √n, the S / N of the image signals of the image sensors a to e after the frame integration does not reach the theoretical characteristic WN, but is shown in FIG. It can be seen that this is an improvement over the background art. Further, the S / N of the image signal of each of the image sensors a to e after the frame integration is gradually improved as the number of frame integration increases.

  For example, when the number of frame integration is 256, the S / N of the image signal is improved by about 1.7 times in the image processing apparatus of the present embodiment compared to the background technology in which the frame integration value is divided by n.

  FIG. 3 shows noise data output from a background image processing apparatus that divides the frame integral value by n, and FIG. 4 shows an output from the image processing apparatus according to the first embodiment that divides the frame integral value by √n. An example of noise data to be processed is shown. Note that the noise data shown in FIGS. 3 and 4 is obtained by measuring the same image sensor at the same temperature. The number of frame integrations is 256 for both.

  As can be seen by comparing the graphs of FIGS. 3 and 4, the noise data output from the image processing apparatus of the present embodiment has a smaller amplitude value than the noise data output from the image processing apparatus of the background art, and RMS. (Root Mean Square) It turns out that noise becomes small.

According to the image processing apparatus of this embodiment, by dividing the frame integral value of the luminance value for each pixel by the mth power of the frame integration number n (0.5 ≦ m <1), the image processing apparatus of the background art. In addition, the S / N of the image signal can be improved by reducing noise.
(Second Embodiment)
The image processing apparatus according to the second embodiment is an example in which the above-described pixel integration is performed as a process for reducing noise. Since the configuration of the image processing apparatus is the same as that of the first embodiment, description thereof is omitted.

In this embodiment, the pixel integration value expressed by the above equation (2) is divided by √n instead of being divided by the pixel integration number n. When the pixel integration value is divided by √n in this way, the signal component S is multiplied by √n as in the first embodiment, so that the S / N of the image signal after pixel integration approaches the theoretical characteristics. Note that dividing the pixel integral value by √n is desirable because noise is reduced most. However, the pixel integral value is not limited to √n, and may be divided by ^nm (0.5 ≦ m <1). That is, in this embodiment, pixel integration is executed using the following equation (4).

Here, k is the horizontal pixel number (1, 2, 3,..., Kmax), l is the vertical pixel number (1, 2, 3,..., Lmax), and h is the horizontal direction in the average value filter. I is the pixel number in the vertical direction in the average value filter, j is the number of pixels in the vertical and horizontal directions of the average value filter (j × j pixels), and n is the number of pixel integrations (j ² ) Yes. Y (k, l) represents the luminance value after pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, and x (k, l) represents the k-th pixel in the horizontal direction. The luminance value before pixel integration of the l-th pixel and the l-th pixel in the vertical direction is shown. Here, kmax is the number of pixels in the horizontal direction, and lmax is the number of pixels in the vertical direction.

  FIG. 5 shows measured values of S / N of image signals of the image sensors (CMOS image sensors) A to E after pixel integration with respect to the number of pixel integrals when the pixel integral value is divided by √n. In FIG. 5, as in FIG. 6, when the S / N value when the pixel integration number n is 1, the S of the image sensors A to E after pixel integration with respect to the pixel integration number is normalized. The measured values of / N are shown respectively.

  As shown in FIG. 5, in the image processing apparatus of this embodiment, it can be seen that the S / N of the image signals of the image sensors a to e after pixel integration is improved as the number of pixel integrations increases.

  According to the image processing apparatus of the present embodiment, by dividing the pixel integral value by the mth power of the pixel integral number n (0.5 ≦ m <1), the background art is the same as in the first embodiment. It is possible to improve the S / N of the image signal by reducing noise as compared with the image processing apparatus.

1 is a block diagram illustrating a configuration of an image processing apparatus according to a first embodiment. It is a graph which shows the measured value of S / N of the image signal with respect to the frame integral number of the image processing apparatus of 1st Embodiment. It is a graph which shows the noise data output from the image processing apparatus of background art. It is a graph which shows an example of the noise data output from the image processing apparatus of 1st Embodiment. It is a graph which shows the measured value of S / N of the image signal with respect to the pixel integration number of the image processing apparatus of 2nd Embodiment. It is a graph which shows the measured value of S / N of the image signal with respect to the frame integration number of the image processing apparatus of background art.

Explanation of symbols

10 processor 11 CPU
12 Main Storage Device 13 Recording Medium 14 Data Storage Device 15 Memory Control Interface Unit 16 Interface Unit 18 Bus

Claims (15)

An image processing apparatus for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
A processing device that integrates the luminance value for each pixel included in the image signal over a plurality of frames, divides the integration result by the mth power of the frame integration number, and outputs the luminance value for each pixel after the division;
A memory for storing the image signal;
An image processing apparatus. The processor is
k is the pixel number in the horizontal direction, l is the pixel number in the vertical direction, i is the frame number, n is the frame integration number, y (k, l) is the kth pixel in the horizontal direction and the lth pixel in the vertical direction. When the luminance value after frame integration of pixels, x _i (k, l), is the luminance value before frame integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction of the i-th frame. ,

The image processing apparatus according to claim 1, wherein y (k, l) calculated in step 1 is output. An image processing apparatus for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
The luminance value for each pixel included in the image signal is integrated with the luminance value of each pixel within a predetermined range from the pixel position, the integration result is divided by the mth power of the pixel integration number, and the pixel value for each pixel after the division is A processing device for outputting a luminance value;
A memory for storing the image signal;
An image processing apparatus. The processor is
k is a horizontal pixel number, l is a vertical pixel number, h is a horizontal pixel number in the predetermined range, i is a vertical pixel number in the predetermined range, and j is a vertical direction in the predetermined range. And the number of pixels in the horizontal direction, n is the number of integrated pixels, y (k, l) is the luminance value after pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, x (k, l ) As the luminance value before pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction,

The image processing apparatus according to claim 3, wherein y (k, l) calculated in step (1) is output.   The image processing apparatus according to claim 1, wherein m is 0.5. A noise reduction method for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating a luminance value for each pixel included in the image signal over a plurality of frames;
Dividing the integration result by the mth power of the frame integration number;
A noise reduction method for outputting a luminance value for each pixel after the division. k is the pixel number in the horizontal direction, l is the pixel number in the vertical direction, i is the frame number, n is the frame integration number, y (k, l) is the kth pixel in the horizontal direction and the lth pixel in the vertical direction. When the luminance value after frame integration of pixels, x _i (k, l), is the luminance value before frame integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction of the i-th frame. ,

The noise reduction method according to claim 6, wherein y (k, l) calculated by the step (a) is output. A noise reduction method for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating the luminance value of each pixel included in the image signal with the luminance value of each pixel within a predetermined range from the pixel position;
Dividing the integration result by the mth power of the pixel integration number;
A noise reduction method for outputting a luminance value for each pixel after the division. k is a horizontal pixel number, l is a vertical pixel number, h is a horizontal pixel number in the predetermined range, i is a vertical pixel number in the predetermined range, and j is a vertical direction in the predetermined range. And the number of pixels in the horizontal direction, n is the number of integrated pixels, y (k, l) is the luminance value after pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, x (k, l ) As the luminance value before pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction,

9. The noise reduction method according to claim 8, wherein y (k, l) calculated by the step is output.   The noise reduction method according to claim 6, wherein m is 0.5. A program for causing a computer to execute processing for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating a luminance value for each pixel included in the image signal over a plurality of frames;
Dividing the integration result by the mth power of the frame integration number;
A program for outputting a luminance value for each pixel after the division. k is the pixel number in the horizontal direction, l is the pixel number in the vertical direction, i is the frame number, n is the frame integration number, y (k, l) is the kth pixel in the horizontal direction and the lth pixel in the vertical direction. When the luminance value after frame integration of pixels, x _i (k, l), is the luminance value before frame integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction of the i-th frame. ,

12. The program according to claim 11, for outputting y (k, l) calculated by the following. A program for causing a computer to execute processing for reducing noise in an image signal,
When m is 0.5 or more and less than 1,
Integrating the luminance value of each pixel included in the image signal with the luminance value of each pixel within a predetermined range from the pixel position;
Dividing the integration result by the mth power of the pixel integration number;
A program for outputting a luminance value for each pixel after the division. k is a horizontal pixel number, l is a vertical pixel number, h is a horizontal pixel number in the predetermined range, i is a vertical pixel number in the predetermined range, and j is a vertical direction in the predetermined range. And the number of pixels in the horizontal direction, n is the number of integrated pixels, y (k, l) is the luminance value after pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction, x (k, l ) As the luminance value before pixel integration of the k-th pixel in the horizontal direction and the l-th pixel in the vertical direction,

14. The program according to claim 13, for outputting y (k, l) calculated by the following.   The program according to claim 11, wherein m is 0.5.

Images