JP2009302850A - Noise removal device of solid-state image sensor, image capturing apparatus, noise removal method of solid-state image sensor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a noise removal device of a solid-state image sensor, etc. capable of removing fixed pattern noise with comparatively high precision without causing size increase of an image sensor and great increase in photography time. <P>SOLUTION: The noise removal device of the solid-state image sensor includes: a pixel part 12 having an OB pixel region and an effective pixel region; a first correction data operation part and first correction processing part 14 which calculates first correction data by every line by performing round averaging of pixel signals in the column direction, corrects the pixel signals by every line using the first calculated correction data regarding the OB pixel region, and corrects pixel signals in each line using the first correction data finally obtained after processing the OB pixel region regarding the effective pixel region; and a second correction data operation part and second correction processing part 15 which calculates second correction data by performing averaging of the pixel signals in the column direction regarding the OB pixel region after first correction processing, and corrects the effective pixel region after the first correction processing based on the second correction data regarding the effective pixel region. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a noise removing device for a solid-state imaging device in which a plurality of pixels are arranged in a row direction and a column direction, an imaging device, and a noise removing method for the solid-state imaging device.

  In a solid-state image pickup device used for a digital camera, a digital video camera, or the like, in particular, a MOS type solid-state image pickup device, fixed pattern noise having a vertical stripe pattern may occur. This fixed pattern noise is generated due to variations in characteristics of transistors included in a pixel or variations in characteristics of readout circuits in each column.

  More specifically, the fixed pattern noise appears as noise of the same level at the same position in the horizontal direction in each row of the image sensor, and becomes a vertical stripe pattern when the entire image is viewed.

  Various techniques for correcting such fixed pattern noise have been proposed in the past, but the process is mainly divided into two stages. That is, a detection process of fixed pattern noise and a removal process of detected fixed pattern noise.

  First, with respect to detection of fixed pattern noise, which is the first stage, a read signal from an OB (optical black) region that is a light-shielding portion of the image sensor is used, or the image sensor is temporarily used by using a mechanical shutter or the like. The readout signal from the effective pixel region in this light shielding state is used. However, since the readout signal from the image sensor generally contains random noise, an undesired pattern may occur due to random noise unless fixed pattern noise from which this random noise is removed is detected. There is. Therefore, by using the fact that the random noise amount is 1 / √N when the number of samples is N, the read signal for a plurality of rows is used to reduce line dam noise. That is, fixed pattern noise in which random noise components other than fixed pattern noise are reduced is extracted by calculating an average of a plurality of pixels belonging to the same column for a predetermined column of the image sensor.

  Next, regarding the removal of the fixed pattern noise, which is the second stage, processing such as recording the fixed pattern noise detected as described above as correction data and subtracting the correction data from the photographed image data is performed. Is called.

  As such a fixed pattern noise correction, for example, the technique proposed in Japanese Patent Laid-Open No. 10-313428 can be cited as an example. When generating fixed pattern noise correction data, the technique described in this publication requires a large number of rows of read data in an OB pixel area or a light-shielded effective pixel area in order to remove random noise other than fixed pattern noise. It has become a thing. If this technique is to be realized based only on read data from the OB pixel area, the number of lines used as the OB pixel area must be increased. However, increasing the OB pixel area increases the readout area other than the effective pixel area, which increases the chip size (and consequently increases the cost of the image sensor) and decreases the imaging frame rate. Will end up.

Japanese Patent Application Laid-Open No. 2004-208240 proposes to solve the above-described problem of Japanese Patent Application Laid-Open No. 10-313428. The technique described in the publication is not an increase in the number of rows in the OB pixel region (that is, read data from the OB pixel region obtained in units of one frame remains the same as the number of rows in the conventional solid-state imaging device) The readout data of the OB pixel area corresponding to many rows is obtained by accumulating the pixels of the OB pixel area for a plurality of frames, and the fixed pattern noise correction data is generated based on the readout data. Yes.
JP 10-313428 A JP 2004-208240 A

  However, in the technique described in Japanese Patent Application Laid-Open No. 2004-208240, since it is necessary to shoot a plurality of frames when generating the fixed pattern noise correction data, in the case of still image shooting where only one frame is shot However, it is impossible to remove fixed pattern noise with high accuracy.

  As described above, in the conventional technique, if the fixed pattern noise is to be removed with high accuracy, the size of the image sensor is increased, or the time required for photographing is significantly increased.

  The present invention has been made in view of the above circumstances, and noise of a solid-state imaging device capable of removing fixed pattern noise with relatively high accuracy without causing an increase in the size of the imaging device and a significant increase in shooting time. An object of the present invention is to provide a removal device, an imaging device, and a noise removal method for a solid-state imaging device.

  In order to achieve the above object, a solid-state image sensor noise removal device according to a first aspect of the present invention includes a solid-state image sensor having a pixel portion in which a plurality of pixels are arranged in a row direction and a column direction, and the pixel. 1st correction data which calculates the 1st correction data which is the addition average value of the pixel signal of a column direction about each column of the pixel part about the pixel of a plurality of rows contained in the 1st predetermined field where the light in the part does not enter An arithmetic unit, a first correction processing unit that corrects a pixel signal of an arbitrary pixel based on the first correction data corresponding to a column to which the arbitrary pixel belongs, and the pixel corrected by the first correction processing unit Second correction data for calculating, for each column of the pixel unit, second correction data that is an average value of pixel signals in the column direction for a plurality of rows of pixels included in a second predetermined region where no light is incident on the unit An arithmetic unit; A second correction processing unit that corrects a pixel signal of an arbitrary pixel corrected by the first correction processing unit based on the second correction data corresponding to the column to which the arbitrary pixel belongs. .

  An image pickup apparatus according to a second aspect of the present invention includes the noise removing device for a solid-state image pickup element according to the first aspect.

  Furthermore, in the noise removal method for a solid-state imaging device according to the third aspect of the present invention, light in the pixel portion of the solid-state imaging device having a pixel portion in which a plurality of pixels are arranged in the row direction and the column direction is not incident. A first correction data calculating step for calculating first correction data for each column of the pixel unit, which is an average value of pixel signals in the column direction for a plurality of rows of pixels included in the first predetermined region; A first correction processing step for correcting the pixel signal based on the first correction data corresponding to the column to which the arbitrary pixel belongs, and light in the pixel portion corrected by the first correction processing step is not incident A second correction data calculation step of calculating, for each column of the pixel unit, second correction data that is an addition average value of pixel signals in the column direction for a plurality of rows of pixels included in the second predetermined region; A second correction process step of correcting the basis of serial arbitrary pixel signals of the pixels corrected by the first correction processing step to said second correction data corresponding to the column in which the arbitrary pixel belongs, is a method with.

  According to the solid-state image pickup device noise removing device, the image pickup apparatus, and the solid-state image pickup device noise removing method of the present invention, fixed pattern noise is relatively highly accurate without causing an increase in the size of the image pickup device and a significant increase in shooting time. Can be removed.

  Embodiments of the present invention will be described below with reference to the drawings.

[Embodiment 1]
1 to 19 show the first embodiment of the present invention, and FIG. 1 is a block diagram showing the configuration of the imaging apparatus.

  The present embodiment is an embodiment in which a noise removing device for a solid-state imaging device is applied to an imaging device.

  The imaging apparatus includes a lens 1, an imaging unit 2, an image processing unit 3, a display unit 4, a camera operation unit 5, and a camera control unit 7. Since the memory card 9 is configured to be detachable from the imaging apparatus, the memory card 9 may not have a configuration unique to the imaging apparatus.

  The lens 1 is used to form an optical image of a subject on the pixel unit 12 (see FIGS. 2 to 4 and the like) of the imaging unit 2.

  The imaging unit 2 includes a solid-state imaging device including the pixel unit 12 described above, and photoelectrically converts an optical image of a subject formed by the lens 1 and outputs the optical image.

  The image processing unit 3 converts an image signal (see FIGS. 2 and 4) that is digitized and output from the imaging unit 2, or an image signal (see FIG. 3) that is digitized after being output from the imaging unit 2, Various digital image processing (this image processing includes, for example, compression / decompression processing for recording / reproduction) is performed.

  The display unit 4 displays an image based on a signal image-processed for display by the image processing unit 3.

  The memory card 9 is a recording medium for storing a signal image-processed for recording by the image processing unit 3.

  The camera operation unit 5 is for performing various operation inputs to the imaging apparatus. The camera operation unit 5 includes a sensitivity input unit 6 for inputting the sensitivity of the imaging apparatus.

  The camera control unit 7 is a control unit that controls the entire imaging apparatus including the imaging unit 2, the image processing unit 3, the memory card 9, and the like based on an operation input from the camera operation unit 5. The camera control unit 7 is, for example, automatically controlled based on the luminance value of the subject calculated based on the image signal obtained from the imaging unit 2 (or the luminance value of the subject measured by a photometric device not shown), or as described above. A sensitivity setting unit 8 is provided for setting the sensitivity of the imaging device by manual setting input from the sensitivity input unit 6. Here, the sensitivity of the imaging device is exemplified by an amplification (gain) of an analog signal in the imaging unit 2 or a digital signal amplification in the image processing unit 3.

  Next, with reference to FIGS. 2 to 4, some examples of function distribution in the imaging unit 2 and the image processing unit 3 will be described.

  First, FIG. 2 is a block diagram illustrating a configuration example in which a part of correction data calculation and correction processing is performed by the imaging unit 2 and the other part is performed by the image processing unit 3, respectively.

  The imaging unit 2 includes a TG unit 11, a pixel unit 12, an amplification unit, a CDS unit, an A / D conversion unit 13, a first correction data calculation unit, and a first correction processing unit 14 (in the following, a simple Therefore, the first correction unit 14 and the like are appropriately described).

  Here, the TG unit 11 is a timing generator that generates a timing signal for synchronizing the operation of each unit in the imaging unit 2.

  The pixel unit 12 is configured by two-dimensionally arranging a plurality of pixels in the row direction and the column direction, and photoelectrically converts the subject optical image formed by the lens 1 described above to obtain an electrical imaging signal. Is generated. In the case where the imaging apparatus is for obtaining a color image and has a so-called single plate configuration, a Bayer array color filter, for example, is disposed on the lens 1 side of the pixel unit 12.

  The amplification unit / CDS unit / A / D conversion unit 13 is an amplification unit that amplifies the imaging signal obtained from the pixel unit 12 in an analog manner, a CDS unit that performs correlated double sampling on the analog signal, and a digital imaging signal. An A / D conversion unit for converting the image signal into the image signal.

  The first correction unit 14 performs a first-stage fixed pattern noise removal process, which will be described later, on the digital image signal output from the amplification unit / CDS unit / A / D conversion unit 13.

  Next, the image processing unit 3 includes a second correction data calculation unit / second correction processing unit 15 (hereinafter, appropriately described as the second correction unit 15 or the like for the sake of simplicity), a demosaicing unit 16, A white balance / γ correction unit 17 and a compression / expansion unit 18.

  The second correction unit 15 performs a second-stage fixed pattern noise removal process, which will be described later, on the digital image signal from which the fixed pattern noise output from the first correction unit 14 has been removed to some extent. Thus, a digital image signal from which fixed pattern noise has been removed is generated with higher accuracy.

  The demosaicing unit 16 generates a so-called three-plate image signal from the digital image signal (single-plate image signal) output from the second correction unit 15.

  The white balance / γ correction unit 17 performs white balance processing and γ correction processing on the image signal output from the demosaicing unit 16.

  The compression / decompression unit 18 performs compression processing on the image signal output from the white balance / γ correction unit 17. The compression / decompression unit 18 also performs a process of decompressing the compressed image signal read from the memory card 9.

  Next, FIG. 3 is a block diagram illustrating a configuration example in which the correction processing and correction processing are performed by the image processing unit 3.

  In the configuration shown in FIG. 3, the imaging unit 2 includes a TG unit 11, a pixel unit 12, and an amplification unit / CDS unit 13a.

  The analog imaging signal output from the amplification unit / CDS unit 13a of the imaging unit 2 is converted into a digital image signal by the A / D conversion unit 13b provided outside the imaging unit 2. ing.

  Further, the image processing unit 3 is further provided with a first correction unit 14 in addition to the configuration shown in FIG. 2, and the digital image signal from the A / D conversion unit 13 b is the first correction unit. 14 is input.

  The image signal processed by the first correction unit 14 is then processed by the second correction unit 15 in the same manner as described above.

  Next, FIG. 4 is a block diagram showing a configuration example in which correction data calculation and correction processing are performed by the imaging unit 2.

  In the configuration shown in FIG. 4, the imaging unit 2 includes a TG unit 11, a pixel unit 12, an amplification unit / CDS unit / A / D conversion unit 13, a first correction unit 14, and a second correction unit 15. And. On the other hand, the image processing unit 3 has a configuration in which the second correction unit 15 is deleted from the configuration shown in FIG.

  Any of the configurations shown in FIGS. 2 to 4 described above may be adopted. However, the configuration for removing fixed pattern noise is conventionally performed only on the image processing unit side and only on the imaging unit side. Since it is known what to do, the configuration as shown in FIG. 2 has an advantage in that both can be achieved without the need for a special new design or the like. Yes. Further, the configuration shown in FIG. 3 has an advantage in that the imaging unit 2 can be reduced in size, and further, it is not necessary to incorporate a digital processing circuit in the imaging unit 2. The configuration shown in FIG. 4 has an advantage in that it conforms to a technical trend that is becoming possible to incorporate an electronic circuit into a solid-state imaging device.

  Next, FIG. 5 is a diagram illustrating a configuration of the pixel unit 12 and a reading order.

  The pixel unit 12 includes an OB (optical black) pixel region 12ob and an effective pixel region 12ef. Reading of the pixel data in the pixel unit 12 is performed for each horizontal line data 12LD as shown in the hatched portion, and is read in the order shown by the arrow AR1. That is, pixel data is read out for the OB pixel region 12ob first, and then for the effective pixel region 12ef in units of one horizontal row.

  Next, FIG. 6 is a diagram illustrating a pixel configuration of the OB pixel region 12ob.

  Here, it is assumed that the number of pixels in the horizontal direction is m, and the OB pixel region 12ob is equivalent to k lines (refer to the above-mentioned JP-A-10-31428 for the number of lines in the OB pixel region 12ob). Unlike the example of multiple rows described above, it is assumed that the number of lines is as high as that provided in a normal solid-state image sensor. At this time, the pixels arranged in the OB pixel area 12ob are P (1, 1) to P (m, k) as shown in the figure. Note that pixel data of a pixel P (i, j) (where i and j are integers) is represented by P (i, j) using the same symbols (this notation is an effective pixel). The pixel in the region 12ef is also used).

  In the present embodiment, correction data used for removing fixed pattern noise is calculated using only the read data of the OB pixel area 12ob as shown in FIG.

  Next, FIG. 7 is a diagram illustrating a configuration of the first correction data calculation unit / first correction processing unit 14 and the second correction data calculation unit / second correction processing unit 15.

  In the present embodiment, since the configuration of the first correction unit 14 and the configuration of the second correction unit 15 are basically the same, these configurations will be described together with reference to FIG.

  The first and second correction units 14 and 15 include a switch 21, an addition unit 22, a calculation unit 23, a delay unit 24, an initial value giving unit 25, a fixed pattern noise correction data recording unit 26, and a subtraction unit. 27.

  The input data is connected to the switch 21 and the subtracting unit 27. The switch 21 is connected to the adding unit 22. The addition unit 22 is connected to the calculation unit 23. The calculation unit 23 is connected to the delay unit 24 and the fixed pattern noise correction data recording unit 26. The initial value giving unit 25 is also connected to the delay unit 24. The delay unit 24 is connected to the addition unit 22. The fixed pattern noise correction data recording unit 26 is connected to the subtracting unit 27. Then, output data is output from the subtracting unit 27.

  The switch 21 is a switch that is turned on when the input data is read data of the OB pixel region 12ob, and is turned off when the input data is read data of the effective pixel region 12ef.

  The adder 22 adds the input data and the data delayed by one line by the delay unit 24, for example.

  The calculation unit 23 divides the data output from the addition unit 22 by, for example, 2 in this embodiment.

  The delay unit 24 delays the data output from the calculation unit 23 by, for example, a reading time for one line in the present embodiment, and includes a line buffer, for example.

  The initial value giving unit 25 gives in advance data (initial value) output from the delay unit 24 when the input data is read data of the first line of the OB pixel region 12ob.

  The fixed pattern noise correction data recording unit 26 is configured as a line buffer for one line, and when the next line data is output from the calculation unit 23, it overwrites and stores the already stored line data. It has become a thing. Then, when the readout data of the OB pixel area 12ob is completed and the data of the next line is no longer output from the calculation unit 23 when the switch 21 is turned off, the result of processing the last line in the OB pixel area 12ob is stored. It keeps memorizing the line data.

  The subtracting unit 27 uses the fixed pattern noise correction data stored in the fixed pattern noise correction data recording unit 26 (hereinafter, the fixed pattern noise correction data related to the first correction unit 14 as the first correction data and the first correction data from the input data. 2) subtracting the fixed pattern noise correction data related to the correction unit 15 (referred to as second correction data, etc.) to output output data in which the fixed pattern noise is reduced.

  FIG. 8 is a diagram for explaining a state of processing in a main part of the first correction data calculation unit / first correction processing unit 14. The processing in the first correction unit 14 will be described with reference to FIG. 7 and FIG. 8 described above.

  The first correction data applied in the first correction unit 14 for the pixel P (i, j) is C1 (i, j). Then, the first correction data calculated in the first correction unit 14 corresponds to the pixel array P (1, 1) to P (m, k) of the OB pixel region 12ob as shown in FIG. (1, 1) to C1 (m, k). Then, the first correction data C1 (1, k) to C1 (m, k) finally obtained after the processing related to the OB pixel region 12ob is performed on the readout pixels of each line from the effective pixel region 12ef. Will be applied.

  At this time, the calculation of the first correction data C1 (i, j) is performed by the following cyclic addition average.

  First, before starting the processing, the initial value assigning unit 25 sets the initial value C1 (i, 0) of the first correction data for one line (where i = 1,... m).

When the pixel data P (i, 1) of the first line is input, the adding unit 22 adds the initial value C1 (i, 0) from the delay unit 24, and the addition result is calculated by the calculating unit 23. Is divided by 2, the first correction data C1 (i, 1) for the first line is calculated as shown below.
C1 (i, 1) = {C1 (i, 0) + P (i, 1)} / 2
The first correction data C1 (i, 1) calculated in this way is overwritten and stored in the fixed pattern noise correction data recording unit 26, and is output to the delay unit 24 to delay the readout time for one line.

Similarly, the processing for the second line and thereafter is performed as follows.
C1 (i, 2) = {C1 (i, 1) + P (i, 2)} / 2
C1 (i, 3) = {C1 (i, 2) + P (i, 3)} / 2
...
C1 (i, k) = {C1 (i, k-1) + P (i, k)} / 2

  In this way, when the processing of all the lines in the OB pixel area 12ob is completed, the switch 21 is turned off, and the first correction data C1 (i, k) is stored in the fixed pattern noise correction data recording unit 26 as it is. Continue to be.

  On the other hand, the subtraction unit 27 subtracts the first correction data stored in the fixed pattern noise correction data recording unit 26 from the input data.

First, pixel data after correction that is output data from the first correction unit 14 is set as P1 (i, j). Then, when the input data is the pixel data of the OB pixel region 12ob, the output data P1 (i, j) is calculated by the subtractor 27 as follows.
P1 (i, j) = P (i, j) -C1 (i, j)

Specifically, when the input data is the first line of the OB pixel area 12ob, the output data P1 (i, 1) is as follows.
P1 (i, 1) = P (i, 1) -C1 (i, 1)

When the input data is pixel data of the effective pixel region 12ef, the output data P1 (i, j) from the first correction unit 14 is calculated by the subtraction unit 27 as follows.
P1 (i, j) = P (i, j) -C1 (i, k)

  Such corrected pixel data P <b> 1 (i, j) from the first correction unit 14 is input to the second correction unit 15.

  Next, FIG. 9 is a diagram for explaining the processing in the main part of the second correction data calculation unit / second correction processing unit 15. The processing in the second correction unit 15 will be described with reference to FIG. 7 and FIG. 9 described above.

  As described with reference to FIG. 7, the configuration of the second correction unit 15 is basically the same as the configuration of the first correction unit 14. Therefore, the processing performed in the second correction unit 15 is the first This is the same as the processing described above in the correction unit 14. However, the difference is that the input data in the second correction unit 15 is data obtained by performing the first-stage fixed pattern noise removal processing by the first correction unit 14.

That is, when the input data to the second correction unit 15 is data corresponding to the pixel data of the OB pixel region 12ob, the second correction data C2 (i, j) is calculated as follows (note that In the second correction unit 15, the initial value giving unit 25 gives the initial value C2 (i, 0)).
C2 (i, j) = {C2 (i, j-1) + P1 (i, j)} / 2

Then, assuming that the corrected pixel data to be output data from the second correction unit 15 is P2 (i, j), if the input data is data corresponding to the pixel data in the OB pixel area 12ob, The output data P2 (i, j) is calculated by the subtractor 27 as follows.
P2 (i, j) = P1 (i, j) -C2 (i, j)

When the input data is data corresponding to the pixel data of the effective pixel region 12ef, the output data P2 (i, j) from the second correction unit 15 is calculated by the subtraction unit 27 as follows. .
P2 (i, j) = P1 (i, j) -C2 (i, k)

  In the case where the processing as described above is performed, random noise is better removed in the last row of the OB pixel region 12ob than in the first row of the OB pixel region 12ob.

  In the above description, the first and second correction units 14 and 15 perform cyclic addition averaging for each line while reading the pixel data of the OB pixel area 12ob in units of lines to remove random noise. . In this case, at the stage where the processing up to the k-th row which is the last row of the OB pixel region 12ob is completed, the contribution ratio of the random noise to the fixed pattern noise correction data is 1/2 in the k-th row of the OB pixel region 12ob. However, the first row of the OB pixel region 12ob is 1/2 ^ k (here, the symbol "^ k" means k-th power), and the maximum reduction of random noise when the number of samples k is used. The effect 1 / √k cannot be obtained. Therefore, when it is desired to further increase the effect of reducing random noise, circuit design may be performed so that the contribution of random noise to the fixed pattern noise correction data is equally divided for all lines.

Specifically, this can be realized by providing a second calculation unit between the delay unit 24 and the addition unit 22 and making the function of the calculation unit 23 different (hereinafter referred to as calculation unit 23 ′). That is, when the pixel data P (i, j) on the j-th line in the OB pixel area 12ob is input, the second calculation unit multiplies the value from the delay unit 24 by j, and then the addition unit 22 The data from the second calculation unit and the input pixel data P (i, j) are added, and then the calculation unit 23 ′ may divide the output of the addition unit 22 by (j + 1). By such processing, for example, the first correction data C1 obtained by the first correction unit 14 is sequentially as follows (the same applies to the second correction data C2).
C1 (i, 1) = {C1 (i, 0) + P (i, 1)} / 2
C1 (i, 2) = {C1 (i, 1) × 2 + P (i, 2)} / 3
C1 (i, 3) = {C1 (i, 2) × 3 + P (i, 3)} / 4
...
C1 (i, k) = {C1 (i, k-1) * k + P (i, k)} / (k + 1)

  Alternatively, if a buffer or a delay circuit corresponding to the number of lines in the OB pixel region 12ob is provided instead of the above configuration, a higher random noise reduction effect can be obtained.

  However, the configuration as shown in FIG. 7 is a simple one in which the processing by the calculation unit 23 is simply divided by 2, and the buffer is only about a line buffer, so that the circuit configuration is simpler. A certain high effect can be obtained. For this reason, the configuration of FIG. 7 is adopted in the present embodiment in consideration of application to an actual device.

  Subsequently, an experimental example of fixed pattern noise removal by the first correction data calculation unit / first correction processing unit 14 and the second correction data calculation unit / second correction processing unit 15 will be described with reference to FIGS. To do. FIG. 10 relates to fixed pattern noise, fixed pattern noise on which random noise is superimposed, first correction data, pixel data corrected by the first correction data, second correction data, and pixel data corrected by the second correction data. Chart showing several numerical examples, FIG. 11 is a diagram showing the state of fixed pattern noise, FIG. 12 is a diagram showing the state of fixed pattern noise superimposed with random noise, and FIG. 13 shows the state of the first correction data. FIG. 14 is a diagram showing the state of pixel data after correction by the first correction data, FIG. 15 is a diagram showing the state of second correction data, and FIG. 16 is a pixel after correction by the second correction data. It is a diagram which shows the mode of data.

  In this experiment, the number k of lines in the OB pixel region 12ob is set to 20.

  When performing an experiment for removing fixed pattern noise, first, fixed pattern noise as shown in FIG. 11 was prepared as a sample. This fixed pattern noise is a noise of a pattern in which an offset from the black level is generated and a gentle curve is drawn such that the levels at the left and right ends of one line are low and the center level is high. The specific minimum value, maximum value, average value, median, standard deviation, and value width (maximum value-minimum value) of the fixed pattern noise are as described in the FPN data column of FIG. When this fixed pattern noise is visualized, the center is observed as a vertical stripe pattern brighter than the left and right.

  Next, FIG. 12 shows a state when random noise is superimposed on the fixed pattern noise shown in FIG. In the diagram of FIG. 12A, the central curve indicates fixed pattern noise as shown in FIG. 11, and the upper and lower curves indicate the degree of random noise amplitude. That is, the superimposed random noise has an average amplitude smaller than that of the fixed pattern noise and a change cycle shorter than the change cycle of the fixed pattern noise (for example, changes for each pixel). Yes. The state of random noise with respect to such fixed pattern noise (FPN) is shown as a partially enlarged view in FIG. Specific numerical examples of the fixed pattern noise on which the random noise is superimposed are as described in the FPN + RN column of FIG.

  Subsequently, in FIG. 13, the fixed pattern noise data on which the random noise as illustrated in FIG. 12 is superimposed is processed by the first correction unit 14 by the number of lines of the OB pixel region 12ob (the above-described k lines). The state of the first correction data obtained later is shown. In FIGS. 13 to 16, only the magnitude of the amplitude including random noise is shown by two curves. As shown in the figure, it can be seen that the average amplitude of random noise is reduced due to the statistical effect of using data for k lines. Specific numerical examples of the first correction data are as described in the column of the first correction data in FIG.

  Further, FIG. 14 shows a state of data after subtracting the first correction data as shown in FIG. 13 from the fixed pattern noise data on which the random noise as shown in FIG. 12 is superimposed. As shown in the figure, it can be seen that the offset from the black level is almost eliminated and the fixed pattern noise having a pattern with a higher central level compared to the left and right is considerably reduced. However, it can also be seen that some level of fixed pattern noise still remains. Specific numerical examples of the data after performing the first fixed pattern noise correction are as described in the column of the first corrected data in FIG.

  FIG. 15 shows the second correction data obtained after the first corrected data as shown in FIG. 14 is processed by the second correction unit 15 by the number of lines in the OB pixel region 12ob (for the above-mentioned k lines). The state of is shown. As shown in the figure, the average amplitude of random noise is further reduced due to the statistical effect of using k-line data that has undergone the first-stage fixed pattern noise removal processing. I understand. Specific numerical examples of the second correction data are as described in the column of the second correction data in FIG.

  In addition, FIG. 16 shows the state of the data after subtracting the second correction data as shown in FIG. 15 from the first correction data as shown in FIG. As shown in the figure, each pixel value on the line almost coincides with the black level, and it can be seen that the offset from the black level and the fixed pattern noise are reduced with high accuracy. Specific numerical examples of the data after performing the second fixed pattern noise correction are as described in the column of second corrected data in FIG.

  Based on the specific numerical example shown in FIG. 10, when the effect of the processing by the first correction unit 14 and the effect of the processing by the second correction unit 15 are compared, for example, the following It becomes like this.

  In FIG. 10, focusing on the average value, the average value in the FPN + RN column is 1.636, whereas the average value in the first corrected data column is −0.001734, the average value in the second corrected data column. Is -0.001302. This is because the fixed pattern noise reduction effect due to the processing of the first correction unit 14 is 59.5 dB, and the fixed pattern noise reduction effect due to the processing of the second correction unit 15 is 62.0 dB. It is shown that. Therefore, the improvement effect which can be understood by visually comparing FIG. 14 and FIG. 16 also appears on the numerical value.

  Next, FIG. 17 is a flowchart showing a flow of processing for reducing fixed pattern noise.

  When this process is started, first, data is read out from the pixel unit 12 of the imaging unit 2 for each line in the order shown by the arrow AR1 in FIG. 5 (step S1).

  Then, the first correction unit 14 sequentially generates first correction data for each line (step S2).

  Subsequently, a process of subtracting the first correction data generated in step S2 from the line data read in step S1 by the first correction unit 14 is sequentially performed for each line (step S3).

  Furthermore, the second correction data is sequentially generated for each line by the second correction unit 15 based on the data that has undergone the first-stage fixed pattern noise removal processing in step S3 (step S4).

  Thereafter, the second correction unit 15 sequentially subtracts the second correction data generated in step S4 for each line from the line data subjected to the first-stage fixed pattern noise removal processing in step S3. This is performed (step S5).

  Then, when the process up to step S5 is completed for all lines, this process is terminated.

  FIG. 18 is a timing chart illustrating an example of an imaging sequence of the imaging apparatus corresponding to the fixed pattern noise reduction process.

  In the description of FIG. 18, the imaging apparatus controls an optical shutter that controls the time for the subject light from the lens 1 to reach the imaging unit 2 and an image (moving image) captured by the imaging unit 2 in almost real time. It is assumed that a live view function displayed on the display unit 4 is provided. FIG. 18 shows a state in which a still image is taken when a release button or the like of the photographing apparatus is operated during live view display.

  When performing live view display (the corresponding shooting operation is live view shooting), the optical shutter remains open, and frame images are continuously captured. At this time, the output of the image from the imaging unit 2 is performed for each frame in the order indicated by the arrow AR1 in FIG. 5, that is, in the order of the OB pixel area 12ob and then the effective pixel area 12ef.

  Next, when a release button or the like included in the camera operation unit 5 is operated, the optical shutter is closed, and then the curtain travels (exposure to the pixel unit 12) so that the optical shutter is opened for the set shutter time. Is done. Note that the image signal related to the subject image is not output from the imaging unit 2 when the shutter is closed before exposure and when exposure is performed (however, the readout signal for reset noise removal is excluded) ).

  When the exposure is completed, the optical shutter is closed, and in the order indicated by the arrow AR1 in FIG. 5, first, the image signal of the OB pixel area 12ob is read, and then the image signal of the effective pixel area 12ef is read.

  Then, after a period of no output continues for a time period such as resetting the photodiode of each pixel, the optical shutter is opened again, and live view display is started.

  Next, FIG. 19 is a flowchart showing a processing example in which whether or not the processing by the second correction data calculation unit / second correction processing unit 15 is performed according to the photographing sensitivity.

  When this process is started, first, the imaging apparatus is set to a shooting mode (step S10). Note that this imaging apparatus can be set to a shooting mode and a playback mode.

  Next, it is determined whether the sensitivity setting is automatic or manual (step S11). Here, in the case of being automatic, for example, the camera control unit 7 according to a photometric result performed based on an image photographed by the imaging unit 2 or a photometric result by a photometric device provided separately. The sensitivity setting unit 8 automatically sets the photographing sensitivity (step S12).

  In step S11, when the sensitivity setting is manual, the sensitivity setting unit 8 of the camera control unit 7 performs shooting based on the sensitivity information manually input from the sensitivity input unit 6 of the camera operation unit 5. Sensitivity is set (step S13).

  When the sensitivity setting is thus completed, the operation waits for the release button of the camera operation unit 5 to be operated (step S14).

  Then, when the release button is operated, the imaging unit 2 captures an image and acquires an image (step S15).

  Next, the acquired images are read out in the order as indicated by the arrow AR1 in FIG. 5, the first correction data is calculated by the first correction unit 14, and the first correction process is performed (step S16).

  Subsequently, the camera control unit 7 determines whether or not the sensitivity set as described above by the sensitivity setting unit 8 is equal to or higher than a predetermined sensitivity (step S17). Here, the predetermined sensitivity is a sensitivity for determining whether only the first correction process is sufficient or the second correction process is necessary, and is determined in advance through experiments or the like according to the type or individual of the solid-state imaging device. It has been made. That is, according to the set sensitivity, the amplification factor when analogly amplifying the imaging signal is set as described above. However, if the setting sensitivity is high, the amplification factor also increases. Therefore, the random noise included in the imaging signal is increased. Will be amplified in the same way. Therefore, whether or not there is a lot of random noise in the image signal is determined based on the set photographing sensitivity. Since the criterion for determining whether or not there is a lot of random noise does not depend only on the set sensitivity, it is of course possible to use another criterion.

  If it is determined in step S17 that the set sensitivity is equal to or higher than the predetermined sensitivity, the second correction unit 15 calculates the second correction data and performs the second correction process (step S18).

  On the other hand, if it is determined in step S17 that the set sensitivity is less than the predetermined sensitivity, the process by the second correction unit 15 is not performed.

  Thus, when the processing in step S18 is completed or it is determined in step S17 that the sensitivity is less than the predetermined sensitivity, the processed image is subjected to various image processing, and then recorded in the memory card 9 or the like. Save (step S19), the process is terminated.

  In the above description, the reading of the OB pixel area 12ob is performed in the same manner as the normal pixel reading. However, the present invention is not limited to this, and dummy reading, that is, charge reading from the pixel itself is not performed (or each pixel is read). In this case, only the transfer operation of the horizontal transfer line provided under the pixel region may be performed without reading out the charge from the photodiode included in the pixel area. Even when the pixel signal is such dummy data, the fixed pattern noise correction process can be performed in the same manner as described above.

  According to the first embodiment, the fixed pattern noise correction data is calculated based on only the pixel values of the OB pixel area having the normal number of lines in one frame image, and the fixed pattern noise is removed. The time required for reading out the OB pixel region is the same as that of a normal solid-state image sensor and does not become long. In addition, since it is not necessary to read out images for a plurality of frames, the reading time does not increase.

  In addition to performing the correction process by the first correction unit 14, the correction process is performed by the second correction unit 15, so that the fixed pattern noise can be corrected with high accuracy.

  In this way, it is possible to remove the fixed pattern noise with higher accuracy without increasing the image readout time or without providing a large number of OB pixel regions in the solid-state imaging device.

  Furthermore, since fixed pattern noise can be removed based on image data for one frame, it can be applied not only to still images but also to moving images.

[Embodiment 2]
20 to 23 show the second embodiment of the present invention, and FIG. 20 is a diagram showing the configuration of the pixel unit 12 and the reading order. In the second embodiment, the same parts as those in the first embodiment are denoted by the same reference numerals, description thereof is omitted, and only different points will be mainly described.

  In the first embodiment described above, the fixed pattern noise correction data is calculated based only on the read data from the OB pixel region 12ob. However, in the present embodiment, the light shielding is performed in addition to the read data from the OB pixel region 12ob. Fixed pattern noise correction data is calculated based on read data from a part of the effective pixel region 12ef in the state.

  Therefore, in this embodiment, it is not possible to acquire an image with fixed pattern noise corrected by reading only one frame of image data, and it is necessary to read image data of one frame + α. . Here, the amount of plus alpha (+ α) corresponds to a part of the above-described OB pixel region 12ob and the light-shielded effective pixel region 12ef. However, in practice, the number of lines in the effective pixel area 12ef is significantly larger than the number of lines in the OB pixel area 12ob, and only a part of the effective pixel area 12ef needs to be read. The reading time can be greatly shortened compared with reading all of the image data, and the reading time of image data for one frame can be approached.

  That is, as shown in FIG. 20, in the pixel unit 12 having the same configuration as that of the first embodiment described above, image data relating to the subject is read as indicated by an arrow AR1 (this reading is hereinafter referred to as main reading). Prior to (), after the pixel portion 12 is in a light-shielded state, as indicated by an arrow AR2, the pixel data of the OB pixel region 12ob and several lines (FIG. 2) adjacent to the OB pixel region 12ob in the effective pixel region 12ef ( The pixel data of the number of lines q is read out (this reading is hereinafter referred to as pre-reading). Here, the data read out by the pre-reading is used for calculating the first correction data by the first correction unit 14 as described later.

  Thereafter, in order to read out image data relating to the subject, the pixel data of the OB pixel area 12ob and the pixel data of all lines of the effective pixel area 12ef are read out as indicated by an arrow AR1. By subtracting the first correction data calculated by the first correction unit 14 from the pixel data of the OB pixel region 12ob, the line data subjected to the first-stage fixed pattern noise removal processing is calculated and calculated. Based on the line data, the second correction unit 15 calculates second correction data. Then, the first correction data described above is subtracted from the line data of the effective pixel area 12ef read out thereafter, and the second correction data is further subtracted, thereby removing the fixed pattern noise in the second stage. The processed line data can be obtained.

  Of the first correction unit 14 and the second correction unit 15 that perform such a processing procedure, the configuration of the second correction unit 15 is the same as that shown in FIG. Since the configuration of the correction unit 14 is slightly different, it will be described with reference to FIG.

  FIG. 21 is a diagram illustrating the configuration of the first correction data calculation unit / first correction processing unit 14.

  The first correction unit 14 includes a switch 21, an addition unit 22, a calculation unit 23A, a delay unit 24, a fixed pattern noise correction data recording unit 26A, and a subtraction unit 27.

  The input data is connected to the switch 21 and the subtracting unit 27. The switch 21 is connected to the adding unit 22. The addition unit 22 is connected to the calculation unit 23 </ b> A and the delay unit 24. The delay unit 24 is connected to the addition unit 22. The calculation unit 23A is connected to the fixed pattern noise correction data recording unit 26A. The fixed pattern noise correction data recording unit 26A is connected to the subtracting unit 27. Then, output data is output from the subtracting unit 27.

  The switch 21 is turned on in the case of pre-reading, that is, when the input data is read data of the OB pixel area 12ob and read data of the effective pixel area 12ef in a light-shielded state with a predetermined number of lines. This is a switch that is turned off when the read data is read data.

  The adder 22 adds the input data and the data delayed by one line by the delay unit 24, for example.

  In this embodiment, the arithmetic unit 23A converts the data output from the adder unit 22 from, for example, the total number of pre-read lines, that is, the number k of lines in the OB pixel region 12ob and the effective pixel region 12ef in the light-shielded state in pre-read. The number of lines to be read q is divided by the number of lines added (k + q).

  The delay unit 24 delays the data output from the adder 22 by the read time for one line, and includes a line buffer, for example.

  The fixed pattern noise correction data recording unit 26A is configured as a line buffer for one line, and when the next line data is output from the calculation unit 23A, it overwrites and stores the already stored line data. It has become a thing. When the pre-reading is completed and the switch 21 is turned off so that the next line data is not output from the arithmetic unit 23A, the line data stored as a result of the processing of the last line in the pre-reading is used as it is. Continue to remember.

  The subtraction unit 27 subtracts the fixed pattern noise correction data stored in the fixed pattern noise correction data recording unit 26A from the input data, and outputs output data in which the fixed pattern noise is reduced. However, in this embodiment, the data at the time of pre-reading need not be output.

  Next, the first correction data calculated by the first correction unit 14 configured as shown in FIG. 21 will be described.

  First, before starting the processing, the line buffer in the delay unit 24 is initialized to zero.

  When the pixel data P (i, 1) on the first line of the OB pixel area 12ob is input, the adder 22 adds the value from the delay unit 24. Here, the adder 22 is initialized to 0. Therefore, the pixel data P (i, 1) of the first line is output as it is to the delay unit 24 and the calculation unit 23A.

  The data output to the arithmetic unit 23A is divided by (k + q) and recorded in the fixed pattern noise correction data recording unit 26A, but this data is overwritten every time the next line data is input. In the following, description will be omitted until final useful data is obtained.

  Next, when the pixel data P (i, 2) for the second line is input, the addition unit 22 adds the pixel data P (i, 1) for the first line stored in the delay unit 24. Done.

  The addition by the addition unit 22 is performed in the same manner for the third and subsequent lines, and this addition is performed up to the q-th line which is the final line in the pre-reading of the shielded effective pixel region 12ef. As a result of performing the processing up to the q-th line of the effective pixel region 12ef, the data output from the adding unit 22 is ΣP (i, j). Here, the addition represented by the symbol “Σ” is performed for j = 1 to (k + q).

  When this addition result ΣP (i, j) is output from the adder 22, it is divided by (k + q) by the calculator 23A and is overwritten and recorded in the fixed pattern noise correction data recorder 26A.

Thus, the final first correction data C1 (i, k + q) obtained by the first correction unit 14 by performing pre-reading is
C1 (i, k + q) = ΣP (i, j) / (k + q)
It becomes. That is, the first correction unit 14 calculates a simple addition average of all the lines in the OB pixel area 12ob and some lines used for pre-reading the light-shielded effective pixel area 12ef.

  When the pre-reading is completed, the switch 21 is turned off, and then the main reading is started.

  Then, the subtraction unit 27 performs the above-described subtraction processing of the first correction data C1 (i, k + q) described above from the main read data input in units of lines. As a result, the corrected signal of the OB pixel region 12ob and the corrected signal of the effective pixel region 12ef are output to the second correction unit.

  The subsequent second correction unit 15 calculates the second correction data by taking a cyclic addition average based on the read data from the OB pixel area 12ob, as in the first embodiment. Thereby, the fixed pattern noise remaining after correction by the first correction unit 14 can be removed with higher accuracy.

  Next, FIG. 22 is a flowchart showing a flow of processing for reducing fixed pattern noise, and FIG. 23 is a timing chart showing an example of a shooting sequence of the imaging apparatus corresponding to the fixed pattern noise reduction processing.

  In the present embodiment, since it is necessary to perform pre-reading before the main reading, the flow of noise reduction processing and the shooting sequence are slightly different from those in the first embodiment.

  That is, when the process shown in FIG. 22 is started by operating the release button or the like of the camera operation unit 5, the optical shutter is first closed (step S21).

  Next, pre-reading is performed, and pre-read data (that is, read data from a part of the OB pixel area 12ob and the effective pixel area 12ef) is sequentially output (see also FIG. 23) (step S22).

  Subsequently, based on the pre-read data, the first correction unit 14 calculates the first correction data as described above (step S23).

  Thereafter, curtain travel (exposure to the pixel unit 12) is performed so that the optical shutter is opened for the set shutter time (step S24). Note that after pre-reading, the image signal related to the subject image is not output from the imaging unit 2 until the exposure is completed.

  When the exposure is completed and the shutter is closed again, the image data is sequentially read out line by line in the order indicated by the arrow AR1 in FIG. 5, that is, in the order of the OB pixel area 12ob and the effective pixel area 12ef (step). S25).

  Then, first, the first correction data generated in step S23 is sequentially subtracted from the data output from the imaging unit 2 by the first correction unit 14 (step S26).

  Next, the second correction unit 15 sequentially processes the data related to the OB pixel region 12ob in units of lines in the data subjected to the first-stage fixed pattern noise removal processing from the first correction unit 14. By going, final second correction data is calculated when the processing up to the last line in the OB pixel region 12ob is completed (step S27).

  Subsequently, when the effective pixel region 12ef subjected to the first-stage fixed pattern noise removal process is output from the first correction unit 14, the second correction unit 15 sequentially subtracts the second correction data for each line. Thus, the fixed pattern noise removal process in the second stage is performed (step S28).

  In this way, when the second stage fixed pattern noise removal processing is performed for all the lines of the effective pixel region 12ef, this processing is terminated.

In the present embodiment, the first correction unit 14 calculates the first correction data by the simple addition average, and the second correction unit 15 calculates the second correction data by the cyclic addition average. The correction unit 15 may be configured to perform simple addition averaging (in the first embodiment described above, the second correction unit 15 may be configured to perform simple addition averaging). In this case, the second correction unit 15 also has a configuration as shown in FIG. 21, but the read data from the imaging unit 2 used to calculate the second correction data is the same as that at the time of the main reading. Only read data from the OB pixel area 12ob is obtained. At this time, the second correction data C2 (i, k) applied to the effective pixel area 12ef for the main reading is as follows.
C2 (i, k) = Σ {P (i, j) −C1 (i, k + q)} / k
Here, the addition represented by the symbol “Σ” is performed for j = 1 to k.

  Also in the present embodiment, as in the first embodiment described above, a part of the OB pixel area 12ob and the light-shielded effective pixel area 12ef at the time of pre-reading and a reading of the OB pixel area 12ob at the time of main reading are performed. May be used for dummy reading.

  According to the second embodiment, since the correction process is performed by the second correction unit 15 in addition to the correction process by the first correction unit 14, all the lines in the OB pixel region are shielded from light. It is possible to correct the fixed pattern noise with relatively high accuracy by only reading out some lines in the effective pixel region in the state of being set. Even in this case, it is possible to shorten the reading time and the manufacturing cost as compared with the conventional example in which image data of a plurality of frames is read out and the conventional example in which a large number of OB pixel regions are provided. It becomes possible.

  Furthermore, since the first correction unit 14 (or the second correction unit 15 in addition thereto) performs simple addition averaging, it can be expected that the effect of reducing random noise is higher than the cyclic addition average. And there is an advantage that the circuit configuration becomes simpler.

  In the above description, the noise removing device for the solid-state imaging device and the imaging device including the noise removing device for the solid-state imaging device have been mainly described. However, the noise removing method for the solid-state imaging device for performing the same processing may be used. I do not care.

  The present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, you may delete a some component from all the components shown by embodiment. Furthermore, the constituent elements over different embodiments may be appropriately combined. Thus, it goes without saying that various modifications and applications are possible without departing from the spirit of the invention.

1 is a block diagram illustrating a configuration of an imaging apparatus according to Embodiment 1 of the present invention. FIG. 3 is a block diagram illustrating a configuration example in which a part of correction data calculation and correction processing is assigned to an imaging unit and the other part is assigned to an image processing unit in the first embodiment. FIG. 3 is a block diagram illustrating a configuration example in which correction data calculation and correction processing are performed by an image processing unit in the first embodiment. FIG. 3 is a block diagram illustrating a configuration example in which correction data calculation and correction processing are performed by an imaging unit in the first embodiment. FIG. 3 is a diagram illustrating a configuration of a pixel portion and a reading order in the first embodiment. FIG. 3 is a diagram illustrating a pixel configuration of an OB pixel area in the first embodiment. The figure which shows the structure of the 1st correction data calculating part and 1st correction process part in the said Embodiment 1, and a 2nd correction data calculating part and a 2nd correction process part. In the said Embodiment 1, the figure for demonstrating the mode of a process in the principal part of a 1st correction data calculating part and a 1st correction process part. In the said Embodiment 1, the figure for demonstrating the mode of the process in the principal part of the 2nd correction data calculating part and a 2nd correction process part. In the first embodiment, fixed pattern noise, fixed pattern noise superimposed with random noise, first correction data, pixel data corrected by the first correction data, second correction data, pixel data corrected by the second correction data The figure which shows some numerical examples regarding each. In the said Embodiment 1, the diagram which shows the mode of fixed pattern noise. In the said Embodiment 1, the diagram which shows the mode of the fixed pattern noise which superimposed the random noise. In the said Embodiment 1, the diagram which shows the mode of 1st correction data. In the said Embodiment 1, the diagram which shows the mode of the pixel data after correction | amendment by 1st correction data. In the said Embodiment 1, the diagram which shows the mode of 2nd correction data. In the said Embodiment 1, the diagram which shows the mode of the pixel data after correction | amendment by 2nd correction data. 5 is a flowchart showing a flow of processing for reducing fixed pattern noise in the first embodiment. 4 is a timing chart showing an example of a shooting sequence of the imaging apparatus corresponding to fixed pattern noise reduction processing in the first embodiment. 9 is a flowchart illustrating a processing example in which whether or not processing by the second correction data calculation unit and the second correction processing unit is performed in accordance with the shooting sensitivity in the first embodiment. The figure which shows the structure of the pixel part and reading order in Embodiment 2 of this invention. The figure which shows the structure of the 1st correction data calculating part and 1st correction process part in the said Embodiment 1. FIG. 5 is a flowchart showing a flow of processing for reducing fixed pattern noise in the first embodiment. 4 is a timing chart showing an example of a shooting sequence of the imaging apparatus corresponding to fixed pattern noise reduction processing in the first embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Lens 2 ... Imaging part 3 ... Image processing part 4 ... Display part 5 ... Camera operation part 6 ... Sensitivity input part 7 ... Camera control part (control part)
DESCRIPTION OF SYMBOLS 8 ... Sensitivity setting part 9 ... Memory card 11 ... TG part 12 ... Pixel part 12ef ... Effective pixel area 12ob ... OB pixel area 13 ... Amplifying part / CDS part / A / D conversion part 13a ... Amplifying part / CDS part 13b ... A / D conversion unit 14... First correction data calculation unit / first correction processing unit (first correction unit)
15 ... 2nd correction data calculating part and 2nd correction process part (2nd correction part)
16 ... Demosaicing unit 17 ... White balance / γ correction unit 18 ... Compression / decompression unit 21 ... Switch 22 ... Addition unit 23, 23A ... Calculation unit 24 ... Delay unit 25 ... Initial value giving unit 26, 26A ... Fixed pattern noise correction Data recording unit 27 ... subtraction unit

Claims (9)

A solid-state imaging device having a pixel portion in which a plurality of pixels are arranged in a row direction and a column direction;
The first correction data, which is the addition average value of the pixel signals in the column direction, is calculated for each column of the pixel unit for a plurality of rows of pixels included in the first predetermined region where no light is incident on the pixel unit. A first correction data calculation unit;
A first correction processing unit that corrects a pixel signal of an arbitrary pixel based on the first correction data corresponding to a column to which the arbitrary pixel belongs;
Second correction data, which is an addition average value of pixel signals in the column direction, is obtained for a plurality of rows of pixels included in the second predetermined region where light is not incident on the pixel unit corrected by the first correction processing unit. A second correction data calculation unit for calculating each column of the pixel unit;
A second correction processing unit that corrects a pixel signal of an arbitrary pixel corrected by the first correction processing unit based on the second correction data corresponding to a column to which the arbitrary pixel belongs;
A noise removing device for a solid-state imaging device, comprising: The pixel unit includes an optical black pixel region into which subject light is not incident and an effective pixel region into which subject light can be incident, and the solid-state imaging device receives a pixel signal of the optical black pixel region as a pixel in the effective pixel region. It is driven to read before the signal,
2. The solid-state image pickup device noise removal device according to claim 1, wherein both the first predetermined region and the second predetermined region are the optical black pixel region.   The pixel signals processed by the first correction data calculation unit, the first correction processing unit, the second correction data calculation unit, and the second correction processing unit are processed after the exposure by the pixel unit. The noise removal apparatus for a solid-state image pickup device according to claim 2, wherein the pixel signal is obtained by one reading from the image pickup device. The first correction data calculation unit calculates the first correction data for each row by cyclic addition with respect to the optical black pixel region,
For the optical black pixel region, the first correction processing unit cyclically outputs the pixel signal of the arbitrary pixel for each row by the first correction data calculation unit corresponding to the column to which the arbitrary pixel belongs. Correction is performed based on the calculated first correction data, and for the effective pixel region, the pixel signal of the arbitrary pixel is processed on the optical black pixel region by the first correction data calculation unit. 4. The solid-state image sensor noise removal device according to claim 3, wherein the correction is performed based on the first correction data finally obtained as a result. The pixel unit includes an optical black pixel region into which subject light is not incident and an effective pixel region into which subject light can be incident, and the solid-state imaging device receives a pixel signal of the optical black pixel region as a pixel in the effective pixel region. The first readout in the light-shielded state before the exposure by the pixel unit and the second readout after the exposure by the pixel unit are driven so as to be read out before the signal And is driven to do
The first predetermined area is an area of the optical black pixel area and a part of the effective pixel area in a light-shielded state, and the second predetermined area is the optical black pixel area,
The first correction data calculation unit calculates the first correction data based on the pixel signal obtained by the first reading,
The first correction processing unit corrects the pixel signal of an arbitrary pixel obtained by the second reading based on the first correction data corresponding to the column to which the arbitrary pixel belongs. The noise removing device for a solid-state imaging device according to claim 1, wherein The first correction data calculation unit and the first correction processing unit are provided inside the solid-state imaging device,
The noise removal apparatus for a solid-state image sensor according to claim 1, wherein the second correction data calculation unit and the second correction processing unit are provided outside the solid-state image sensor. A sensitivity setting unit for setting the sensitivity related to the amplification degree of the pixel signal obtained from the solid-state imaging device;
A control unit for controlling the second correction data calculation unit and the second correction processing unit to perform processing only when the sensitivity set by the sensitivity setting unit is equal to or higher than a predetermined sensitivity;
The noise removal device for a solid-state imaging device according to claim 1, further comprising:   An imaging apparatus comprising the solid-state imaging device noise removing apparatus according to claim 1. A pixel signal in a column direction for a plurality of rows of pixels included in a first predetermined area where light is not incident on the pixel unit of a solid-state imaging device having a pixel unit in which a plurality of pixels are arranged in a row direction and a column direction. A first correction data calculation step for calculating the first correction data, which is an addition average value, for each column of the pixel unit;
A first correction processing step of correcting a pixel signal of an arbitrary pixel based on the first correction data corresponding to the column to which the arbitrary pixel belongs;
Second correction data, which is an addition average value of pixel signals in the column direction, is obtained for a plurality of rows of pixels included in the second predetermined region in which light is not incident in the pixel portion corrected in the first correction processing step. A second correction data calculation step for calculating each column of the pixel portion;
A second correction processing step of correcting a pixel signal of an arbitrary pixel corrected by the first correction processing step based on the second correction data corresponding to the column to which the arbitrary pixel belongs;
A method for removing noise from a solid-state imaging device.

Images