JP4952548B2 - Noise detection device, imaging device, and noise detection method - Google Patents

Description

  The present invention relates to a noise detection device that detects noise from an output signal from a solid-state imaging device, an imaging device having such a function, and a noise detection method, and particularly, when an XY address scanning type solid-state imaging device is used. The present invention relates to a noise detection device, an imaging device, and a noise detection method suitable for the above.

  In a digital imaging apparatus, a light-shielded area called an optical black (OPB) area is provided on the imaging apparatus, and an output signal of an effective image area is corrected based on an output signal from the OPB area. It has been broken. As a typical example, there is known a method of detecting an added value or an average value of the output level from the OPB area at the time of imaging and matching the black level of the data in the effective image area using the detection result.

In particular, when an XY address scanning type imaging device such as a CMOS (Complementary Metal-Oxide Semiconductor) image sensor is used, the detection result in the OPB region (VOPB region) adjacent to the effective image region in the vertical direction. A method has been proposed for removing vertical streak noise caused by variation in the characteristics of each column. As an example of this, there has been one in which defective pixels are detected from the output signal of the VOPB region, and the vertical streak noise is accurately removed by using the output signals from the pixels excluding those defective pixels. (For example, refer to Patent Document 1).
JP 2006-25148 A

  By the way, some image sensors on the XY address operation side are provided with an A / D converter and a CDS (Correlated Double Sampling) circuit for each column. In the imaging device having such a configuration, a phenomenon has been found in which pixel data output from an output stage of a specific column has a random variation for each pixel. This phenomenon is that the column where the variation appears is fixed, but it is not known where and when the data variation occurs in the column. The cause of the occurrence is provided at the output stage of the column. It is considered to depend on the operation of a transistor such as an A / D converter.

  In this phenomenon, the level variation becomes conspicuous particularly in an image at a high gain with a small dark current, and therefore it is desired to appropriately correct the variation. However, in the column where the variation occurs, almost the same number of pixels that are higher and lower than the peak of the appearance frequency of the pixel level in the column are generated, so that the pixel data in the VOPB area of the column is conventionally generated. When trying to detect based on the added value or the average value, there is a problem that the variation is buried and it cannot be detected accurately.

  The present invention has been made in view of such points, and a noise detection apparatus capable of accurately detecting random noise generated in an output signal from a specific column of an XY address scanning type solid-state imaging device, An object is to provide an imaging device and a noise detection method.

  In the present invention, in order to solve the above-described problem, in a noise detection apparatus that detects noise from an output signal from an XY address scanning type solid-state imaging device, an output from an effective image region that receives incident light on the solid-state imaging device. Before capturing the signal, a black dummy signal is output on the solid-state image sensor so that the output signal system for each column of the solid-state image sensor does not include a signal component corresponding to light incident on the solid-state image sensor. A signal capturing unit that captures in line units, a reference value acquiring unit that acquires a reference value corresponding to an average value of the signal level of the black dummy signal, and inputs of the black dummy signal for a predetermined number of lines from the signal capturing unit Accept and determine the column in which random noise is generated based on the signal level for each column of the black dummy signal and the reference value from the reference value acquisition unit That the noise determination section, the noise detection apparatus characterized by having provided.

  In such a noise detection apparatus, before capturing an output signal from an effective image area that receives incident light on the solid-state image sensor, the output signal system for each column of the solid-state image sensor responds to the incident light to the solid-state image sensor. The black dummy signal output so as not to include the signal component is captured in units of lines on the solid-state imaging device by the signal capturing unit. The reference value acquisition unit acquires a reference value corresponding to the average value of the signal levels of the black dummy signal. The noise determination unit receives input of a black dummy signal for a predetermined number of lines from the signal capturing unit, and random noise is generated based on the signal level for each column of the black dummy signal and the reference value from the reference value acquisition unit. Determine which column is occurring.

  According to the noise detection device of the present invention, by using the reference value corresponding to the average value of the signal level of the black dummy signal and the signal level for each column of the input black dummy signal, the incident on the solid-state imaging device. It becomes possible to analyze the degree of variation in the signal level for each column in the output signal so as not to include a signal component corresponding to light, and thereby it is possible to accurately detect a column in which random noise is generated.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing a main configuration of an imaging apparatus according to the embodiment.
The imaging apparatus shown in FIG. 1 is realized as a digital still camera or a digital video camera. The imaging apparatus includes a camera block 11, an image processing unit 12, an SDRAM (Synchronous Dynamic Random Access Memory) 13, a display unit 14, a control unit 15, a flash memory 16, an external memory controller 17, and an operation unit 18.

  The camera block 11 is provided with an image sensor 11a for converting incident light from a subject into an electrical signal. In the present embodiment, an XY address scanning solid-state image sensor such as a CMOS image sensor is used as the image sensor 11a. The camera block 11 further drives a lens for condensing light from the subject onto the image sensor 11a, a drive mechanism for moving the lens to perform focusing and zooming, a shutter mechanism, an iris mechanism, and the image sensor 11a. For example, a TG (Timing Generator), an A / D converter circuit for converting an output signal from each pixel of the image sensor into a digital signal, and the like are provided. These functions operate under the control of the control unit 15.

  The image processing unit 12 performs detection processing for AF (Auto Focus), AE (Auto Exposure), various image quality correction processing on the image signal output from the camera block 11, and the control unit 15 based on detection information. Signal processing such as pixel interpolation, mixing, and image quality correction according to the control signal is performed. Further, the image signal after signal processing is converted into a signal such as Y / Cb / Cr and subjected to compression encoding processing to create an image file for recording. Further, an image signal for display is generated and output to the display unit 14. In the present embodiment, a noise detection / correction function, which will be described later, is provided in the image processing unit 12.

  The operation of the image processing unit 12 is controlled by the control unit 15. In addition, data such as parameters necessary for the processing of the image processing unit 12 is basically stored in the flash memory 16, and these data are read into the SDRAM 13 as necessary. The image processing unit 12 operates while using the SDRAM 13 as a buffer.

The display unit 14 includes, for example, an LCD (Liquid Crystal Display), and displays an image based on an image signal output from the image processing unit 12.
The control unit 15 is configured as a microcomputer including a memory such as a CPU (Central Processing Unit) and a RAM (Random Access Memory). The control unit 15 performs overall control of the imaging apparatus by executing a program stored in the flash memory 16 or the like.

  The external memory controller 17 controls reading and writing of data with respect to the portable external memory 17a attached to the imaging apparatus. As the external memory 17a, for example, a memory card composed of a flash memory is applied. In the external memory 17a, an image file generated by the image processing unit 12 is written. In addition, as a portable recording medium for recording an image file, a magnetic tape, an optical disk, or the like may be used. Alternatively, an HDD (Hard Disk Drive) may be used.

The operation unit 18 includes various input switches, and outputs a control signal corresponding to an operation input by the user to these input switches to the control unit 15.
In the imaging apparatus having the above configuration, a signal received and photoelectrically converted by the imaging element 11a is converted into a digital image signal and sequentially supplied to the image processing unit 12. The image processing unit 12 performs image quality correction processing on the image signal supplied from the camera block 11, and the processed image signal is converted into a display image signal and supplied to the display unit 14. As a result, the currently captured image (monitoring image) is displayed on the display of the display unit 14, and the photographer can view the image and check the angle of view.

  In this state, when the control unit 15 is instructed to record an image by pressing the shutter release button of the operation unit 18, the image processing unit 12 compresses the image signal after the image quality correction. An image file is generated by encoding processing. The generated image file is stored in the external memory 17a via the control unit 15 and the external memory controller 17. The image processing unit 12 generates an image file based on the input image signal for one frame when recording a still image. When recording a moving image, the image processing unit 12 The compression encoding process is continuously performed.

FIG. 2 is a diagram illustrating a configuration example of the image sensor and its peripheral circuits in the camera block.
FIG. 2 shows a case where a CMOS image sensor is used as the image sensor 11a. In this example, color filters are arranged in a so-called Bayer arrangement, and the letters “R”, “Gr”, “Gb”, and “B” indicating the components of the color filters are described on the photodiode 101 of each pixel. Yes.

  In the imaging device 11a, the pixel circuit of each pixel arranged in a matrix mainly includes a photodiode 101 that accumulates charges generated by photoelectric conversion, and an amplification amplifier 102 that amplifies an output signal from the photodiode 101. The transfer circuit 103 controls the transfer of the amplified signal. The output of the transfer circuit 103 is connected to a column transfer line 104. The transfer circuit 103 transfers signals in accordance with the control voltage supplied from the vertical scanning circuit 105 through the line (row) selection line 106. That is, when a control voltage is applied to the line selection line 106, the transfer circuit 103 of each pixel circuit on the same line is turned on, and an output signal from the photodiode 101 is output to the column transfer line 104. By applying a control voltage to the line selection line 106 in order from the upper side of the light receiving surface by the vertical scanning circuit 105, an output signal from the pixel circuit of each line is sequentially output to the column transfer line 104.

  In the example of FIG. 2, a so-called column A / D circuit is provided in which an A / D conversion circuit 110 is individually provided at the output stage of each column transfer line 104. The A / D conversion circuit 110 includes a comparator 111 that compares the voltage of the column transfer line 104 with the reference voltage Vref, a counter 112 that counts the output of the comparator 111, a latch circuit 113 that latches the output of the counter 112, and the like. The output signal from each pixel circuit transferred to the column transfer line 104 is converted into a digital signal. Although an example in which an A / D conversion circuit is provided for each column is shown here, a CDS (Correlated Double Sampling) circuit is provided for each column, or both the A / D conversion circuit and the CDS circuit are provided. It may be provided for each column.

  An output signal from the A / D conversion circuit 110 is output to the output signal line 122 via the column selection circuit 121, further amplified by the output amplifier 123, and then output to the image processing unit 12. The column selection circuit 121 for each column controls signal output in accordance with a control voltage supplied through the column selection line 125 from the horizontal scanning circuit 124. That is, by applying a control voltage to the column selection line 125 in order from the left side of the light receiving surface by the horizontal scanning circuit 124, output signals from pixel circuits on the same line in each column are sequentially output to the output signal line 122. .

FIG. 3 is a diagram schematically showing the configuration of the light receiving element region of the imaging element.
As shown in FIG. 3, the light receiving element region 150 of the image sensor 11a includes an effective image region 151 that receives incident light, and an OPB region that is adjacent to the effective image region 151 on the vertical and horizontal sides. A certain VOPB region 152 and a HOPB region 153 are formed. Among these, the VOPB region 152 and the HOPB region 153 are both shielded from light, and the pixel signals that are output from these regions and do not contain a signal component corresponding to incident light are mainly black in the image processing unit 12. Used for level correction. As will be described later, in this embodiment, the pixel signal output from the VOPB region 152 is also used for noise detection for each column. The image file and display image signal generated by the image processing unit 12 are generated based on the pixel signal from the effective image area 151 in the drawing.

[First Embodiment]
Next, a processing function for removing random noise generated in the image sensor 11a and its peripheral circuits will be described. In the present embodiment, this processing function is provided in the image processing unit 12. In addition, the random noise to be removed here is obtained when an XY address scanning image sensor is used as described above, and an A / D conversion circuit or CDS circuit is provided for each column. , Which occurs randomly in a particular column. Hereinafter, such noise is referred to as “column random noise”.

  This column random noise is considered to occur mainly in a transistor provided in the output stage of the image sensor. Specifically, it is considered that this occurs in an A / D conversion circuit provided for each column at the output stage of the column transfer line or a transistor inside the CDS circuit, like the A / D conversion circuit 110 shown in FIG. ing.

  In the present embodiment, basically, the presence / absence and generation level of column random noise are detected based on a signal that is not affected by incident light, such as an output signal from the VOPB region 152, and the detection result. Based on the above, the output signal from the column in which noise is generated is corrected among the output signals from the effective image area 151.

  However, it is impossible to predict whether such column random noise will appear in each column, and even if it appears, at what level in which pixel in that column. Therefore, in order to detect the occurrence and level of column random noise, it is not only necessary to detect the output signals from all columns, but also as many pixels as possible in the vertical direction in each column. It is desirable to detect this signal. For example, even in a column in which column random noise has occurred, there is a possibility that noise-generating pixels are not included only by detecting pixel signals of about several lines.

  Therefore, in the present embodiment, detection for noise detection uses not only the output signal from the VOPB region 152 but also dummy data obtained by color-reading the output of the image sensor 11a, thereby increasing the number of detections. Random column noise is accurately detected based on the signal obtained from the line.

FIG. 4 is a diagram schematically illustrating a detection region in the image sensor.
As described with reference to FIG. 3, the effective image area 151, the VOPB area 152, and the HOPB area 153 are formed in the light receiving element area 150 of the imaging element 11 a. The light receiving element region 150 is scanned in order from the upper line of the light receiving element region 150 in accordance with the timing signal generated from the TG in the camera block 11 under the control of the control unit 15. A signal is output.

  Furthermore, in this embodiment, immediately before scanning the light receiving element area 150, a dummy area 160, which is an area for color-reading the output of the image sensor 11a, is virtually formed. In the dummy area 160, an output signal that does not include a signal component corresponding to the incident light on the image sensor 11 a is output from the camera block 11 to the image processing unit 12.

  Specifically, when an image signal for one frame (or field) from the effective image area 151 in FIG. 4 is taken into the image processing unit 12, scanning is sequentially performed from the leading line of the dummy area 160 according to the timing signal. The signal of each line in the dummy area 160 is sequentially output from the camera block 11. In this period, for example, only the A / D conversion circuit 110 of each column operates in accordance with the timing signal, and the A / D conversion circuit 110 of each column outputs a signal with no input. When scanning of the dummy region 160 is completed in this way, scanning of the light receiving element region 150 is started, and a signal corresponding to the accumulated charge of the photodiode 101 of each pixel on the light receiving element region 150 is output.

  As a result, like the output signal from the VOPB region 152, the output signal from the dummy region 160 does not include a signal component corresponding to the incident light on the image sensor 11a, and only the output characteristics after the A / D conversion circuit 110 are included. Only the components that depend on are included. Accordingly, it can be considered that column random noise appears in the output signal from the dummy region 160 in exactly the same manner as the output signal from the VOPB region 152.

  As described above, it is desirable to use as many pixel signals as possible in order to detect column random noise. Here, the VOPB area 152 provided in advance in the image sensor 11a is usually an area of, for example, about several tens of lines, and this area is of course fixed and the number of lines cannot be changed. . On the other hand, the number of lines in the dummy area 160 can be arbitrarily set under the control of the control unit 15. That is, the dummy region 160 is set so that the VOPB region 152 and the dummy region 160 are combined to secure a sufficient number of lines for detecting column random noise.

  In this embodiment, the VOPB region 152 and the dummy region 160 are used for detecting column random noise. For example, the number of lines in the VOPB region 152 is sufficient to detect column random noise. In this case, it is not necessary to use the output signal of the dummy area 160. Further, the output signal of the VOPB area 152 may be used only for detecting the black level, for example, and only the output signal of the dummy area 160 may be used for detecting the column random noise. That is, as in the VOPB region 152 and the dummy region 160, output signals that do not include a signal component corresponding to incident light are arranged at the output stage of the image sensor 11a by the number of lines necessary for detecting column random noise. What is necessary is just to make it output with respect to the image process part 12 from a digital signal output circuit.

  Next, a method for detecting column random noise will be specifically described. In the following detection method examples, basically, as a reference value for noise detection, an average value is first calculated based on a signal from a specific area, and then a signal from another area is captured. The presence or absence of column random noise is determined by analyzing the degree of signal level variation from the reference value (average value). As shown in FIG. 4, an area from the first line (ave_start) to a predetermined line (ave_end) of the dummy area 160 is an average calculation detection area 171 for calculating an average value, and the next area in the dummy area 160. The region from the line (detect_start) to the last line (detect_end) of the VOPB region 152 is a noise determination detection region 172 for determining the presence or absence of column random noise for each column.

FIG. 5 is a diagram illustrating an example of a frequency distribution of pixel signal levels in the dummy area and the VOPB area.
FIG. 5 shows a frequency distribution when the horizontal axis represents the pixel signal level class and the vertical axis represents the number of pixels based on the output signals from the dummy area 160 and the VOPB area 152. The upper graph is an example of frequency distribution when column random noise is not generated, and the lower graph is an example of frequency distribution when column random noise is generated.

  The output signal from the dummy area 160 has substantially the same specification as the output signal from the VOPB area 152. Therefore, when column random noise is not generated, the levels of the output signals from the dummy area 160 and the VOPB area 152 are close to the black level. In the frequency distribution in the upper stage of FIG. 5, the position of the center peak substantially matches the average value of these output signal levels, and this value usually corresponds to the black level.

  On the other hand, when column random noise is generated, it is known that the frequency distribution is changed, and as shown in the lower part of FIG. The frequency distribution is such that sub-peaks appear on the (high level side) and below (low level side). Further, it is known that the level difference between the center peak and the upper and lower sub-peaks is almost the same.

  Here, the position and width of the center peak on the frequency distribution vary depending on the photographing conditions such as the gain, exposure time, temperature at that time, and the like given to the obtained image signal. For example, a region having a constant width from a center peak narrower than the width between sub-peaks includes an output signal level that varies due to random noise generated regardless of the column due to dark current or the like. On the other hand, column random noise occurs only in a small part of the column, and occurs almost equally on both the upper and lower sides of the center peak. For this reason, if the output from a sufficient number of pixels is detected, the signal level corresponding to the center peak of the output signal from those pixels is the same under the same shooting conditions regardless of whether or not column random noise is generated. It will be the same as the average level.

  Therefore, in the present embodiment, as a basic procedure for noise detection, first, the signal level corresponding to the center peak of the frequency distribution is obtained as the average value of the output signal from the average calculation detection region 171 in FIG. Thereafter, using this average value as a reference, the degree of variation in the output signal level for each column in the noise determination detection region 172 in FIG. 4 is detected, and a column with a large variation (that is, a pixel having a signal level separated from the average value). Column) is determined to have column random noise. More specifically, the absolute value of the difference between the output signal from each pixel of the noise determination detection region 172 and the average value is taken, the absolute values are added for each column, and the addition result is compared for each column. As a result, the column in which the column random noise is generated is detected.

  In addition, regardless of whether column random noise is generated or not, if detection data from a sufficient number of pixels is obtained at a certain photographing timing, the average value of the signal level in the dummy area 160 and the VOPB area 152 is not expected to change. Therefore, the average value of the pixel signals is calculated for each color component from the entire average calculation detection region 171. In recent years, the image pickup element 11a has a number of columns of about several thousand, and with such an image pickup element 11a, a sufficient number for calculating the average value can be obtained by securing several lines as the detection area 171 for average calculation. A pixel signal can be obtained.

  On the other hand, the noise determination detection region 172 for comparing pixel signals for each column with reference to the average value is, for example, 100 lines or more (in this example) so that this can be reliably detected when column random noise occurs. A larger number of lines such as 200 lines) is secured.

FIG. 6 is a block diagram illustrating a configuration of the noise removal processing unit according to the first embodiment.
As described above, the noise removal processing unit 20 a illustrated in FIG. 6 is provided inside the image processing unit 12. The noise removal processing unit 20 a includes an average value calculation unit 21, a difference average value calculation unit 22, a comparison determination unit 23, and a signal correction unit 24.

  The average value calculation unit 21 calculates the average value of the signal level for each color component based on the output signal from the average calculation detection region 171. As described above, several lines are secured as the average calculation detection area 171 for the calculation of the average value.

Here, as an example, if the average value of the R component is R_ave, this average value R_ave is obtained by the following equation (1). In Expression (1), R_sum represents an addition value obtained by adding output signals from R component pixels among the pixels in the average calculation detection region 171, and R_count_ave represents a number obtained by adding output signals.
R_ave = R_sum / R_count_ave (1)
However, in this average value calculation, if an output signal from a defective pixel is included in the calculation, an accurate average value may not be obtained. For this reason, for example, the threshold determination for the input signal is performed at the input stage to the average value calculation unit 21, and the pixel having a signal equal to or higher than a predetermined upper threshold and the signal equal to or lower than the predetermined lower threshold are included. Pixels are excluded from the average value calculation. Further, if the position of the defective pixel is known in advance, the output signal from the pixel at that position may be excluded from the calculation. As such defective pixel position information, information read into the defect correction unit may be shared.

  As described above, in the calculation of the average value, the pixels excluding the defective pixels among the same color component pixels in the average calculation detection region 171 are used. Therefore, the addition number R_count_ave in Expression (1) is used for the calculation. The number of pixels that have been applied will be applied. Note that the same calculation is performed for the Gr component, the Gb component, and the B component. Then, the average value for each color component is transferred to the difference average value calculation unit 22.

  When the calculation of the average value based on the output signal from the average calculation detection region 171 is completed, the output signal from the noise determination detection region 172 is then supplied to the difference average value calculation unit 22. The difference average value calculation unit 22 calculates the absolute value of the difference between the signal level of each pixel and the average value of the corresponding color component for each column and for each color component. Then, an average value of difference values (absolute values) in each column is calculated. As described above, about 200 lines are secured as the noise determination detection region 172 for the calculation of the difference average value.

Here, as an example, assuming that the difference average value of the R component in the column number xxxx is R_ave_xxxx, first, an integrated value R_sum_xxxx is obtained by integrating the difference value between the signal level of each pixel and the average value by the following equation (2). After that, the difference average value R_ave_xxxx is obtained by Expression (3). In Expressions (2) and (3), R_data indicates the signal level of each input pixel, and R_count_det indicates the number of pixels used in the calculation of Expression (2) (that is, Expression (2) The number of additions on the right side).
R_sum_xxxx = sum (| R_data−R_ave |) (2)
R_ave_xxxx = R_sum_xxxx / R_count_det (3)
Here, FIG. 7 and FIG. 8 are diagrams showing examples of signal level distributions in columns where column random noise is not generated and in columns where they are generated.

  When column random noise is not generated, as shown in FIG. 7, the output signal level from each pixel on one column is very centered on the average value (R_ave in the case of R component). Fits in a narrow area. On the other hand, when column random noise is generated, as shown in FIG. 8, the number of pixels having a signal level that is relatively far from the average value increases.

  For this reason, the difference average value calculation unit 22 repeats addition of the absolute value of the difference from the average value by a sufficient number of lines, such as 200 lines for each column, and thus occurs in a column where column random noise occurs. The added value is larger than that of the column that is not. However, since it is necessary to exclude the output signal from the defective pixel from this calculation, and the number of defective pixels differs for each column, the calculated added value is calculated by the above equation (3) as the number of effective pixels. Is divided by the number of additions R_count_det corresponding to.

  In the difference average value calculation unit 22, the difference average value for each color component in all columns is calculated by the same operation as described above, and is passed to the comparison determination unit 23. The comparison / determination unit 23 compares the obtained difference average value for each color component of each column, and determines that a column whose difference average value is larger than a predetermined threshold value has generated column random noise.

Here, as an example, it is determined that column random noise has occurred when the difference average value of the R component at the column number xxxx satisfies the condition of the following equation (4).
R_ave_xxxx-R_ave> R_judge_level (4)
When a plurality of (two in this example) color components are included in one column, the comparison / determination unit 23 compares each color component with a threshold value, and based on the comparison result, Finally, it is determined whether or not column random noise has occurred in the column. For example, it is determined that column random noise has occurred when the condition corresponding to the above equation (4) is satisfied for both of the two color components.

  Further, a different value may be set for each color component as a threshold value used for comparison. For example, for the Gr component and Gb component that are conspicuous in human visual characteristics, the threshold value may be set smaller than the R component and B component.

  Furthermore, the degree of signal level variation due to the influence of column random noise varies depending on the shooting conditions such as the gain applied to the pixel signal and the temperature at the time of shooting in accordance with the brightness of the subject. For this reason, it is possible to improve noise determination accuracy by changing some or all of the threshold values of each color component in accordance with such shooting conditions. In this case, for example, threshold values corresponding to the gain and temperature are stored in advance as a data table in the noise removal processing unit 20a, and the comparison determination is performed according to the gain and temperature values notified from the control unit 15 at the time of shooting. The threshold value applied to the processing of the unit 23 is read from the data table. As an example, the threshold is set lower as the ISO (International Organization for Standardization) sensitivity at the time of shooting is higher (that is, the gain is higher).

  In addition to the above-described comparison method, for example, by calculating the smallest difference average value for each column and comparing the difference between the minimum value and the difference average value of each column with a threshold value. Whether or not column random noise has occurred may be determined.

  Through the processing of the comparison determination unit 23 described above, the signal correction unit 24 is notified of the column determined to have generated column random noise. This notification is performed by, for example, a flag signal for each column. Thereafter, when an output signal from the effective image area 151 is input to the signal correction unit 24, the signal correction unit 24 outputs an output signal from a column in which column random noise is determined to occur in the vicinity thereof. Correction is performed based on the pixel signal.

  9 and 10 are diagrams for explaining pixel positions used for signal correction. FIG. 9 shows the case where the pixel to be corrected corresponds to the R component, and FIG. 10 shows the case corresponding to the Gb component.

  Although the column in which the column random noise is considered to be generated is specified by the above-described processing, it is not possible to specify how much column random noise is generated in which pixel of the column. For this reason, the signal correction unit 24 of the present embodiment replaces the data of all the pixels in the column with the interpolation data generated based on the signal levels of the surrounding same color component pixels. Various interpolation calculation methods are conceivable. For example, interpolation data is obtained by simply averaging the signal levels of pixels that are not defective among the nearest surrounding pixels having the same color component as the correction target pixel. The correction target pixel data is replaced with the interpolation data only when the signal level of the correction target pixel is not less than the maximum value or not more than the minimum value of the signals of the surrounding pixels. In order to obtain more accurate interpolation data, a correlation with surrounding pixels may be determined, and an interpolation calculation may be performed according to the determination result.

  9 and 10 show examples of how to select peripheral pixels used for interpolation calculation. When the correction target pixel is an R component pixel, as shown in FIG. 9, the closest pixel of the same color component exists at a position separated by two pixels in each horizontal / vertical direction. In addition, when the correction target pixel is a Gb component pixel, a Gr component pixel can be used for the interpolation calculation. Therefore, as shown in FIG. 10, it is possible to use the signal of the Gr component pixel adjacent to the correction target pixel in the oblique direction for the interpolation calculation.

  Note that, in a column in which it is determined that column random noise has occurred, only pixels of G component (Gb component or Gr component) that are considered to be more easily affected by noise generation may be replaced with interpolation data. .

  In addition, it has been found that pixels in which column random noise has occurred are conspicuous visually when the gain at the time of photographing is particularly high. For this reason, when the gain is lower than the predetermined value, the signal correction as described above is not performed for the column determined that the column random noise is generated, and the signal correction is performed only when the gain exceeds the predetermined value. You may do it.

  The noise removal processing unit 20a as described above is desirably provided as close to the camera block 11 as possible in the image signal transmission system of the image processing unit 12. In such a position, for example, a black correction unit that corrects the black level of the output signal from the effective image area 151 based on the output signal from the VOPB area 152 or a defective pixel on the image sensor 11a is generally detected. A defect correction unit or the like for correcting the pixel signal is provided.

  The noise removal processing unit 20a may be provided before the black correction unit. For example, the noise removal processing unit 20a is provided after the black correction unit, and column random noise is based on the image signal whose black level is corrected. By detecting this, column random noise can be more accurately detected as signal level variations that cannot be eliminated even by correcting the black level. On the other hand, the signal correction function (the signal correction unit 24) provided in the noise removal processing unit 20a can be shared with the signal correction function of the defect correction unit. In this case, a position close to the defect correction unit (for example, The noise detection function (the above-described average value calculation unit 21, difference average value calculation unit 22, and comparison determination unit 23) of the noise removal processing unit 20a may be provided in the preceding stage.

  According to the first embodiment described above, first, after calculating the average value of the signal when not affected by the incident light based on the output signal from the average calculation detection region 171, the noise determination detection is performed. Based on the output signals from the region 172, it is determined that column random noise has occurred in a column in which there are many pixels whose output signal levels are far from the average value. As a result, for example, random noise in a specific column that cannot be detected only from the average value of the output signal in the VOPB region 152 can be detected.

  In addition, by performing such noise detection every time shooting is performed, even when a signal level shifts according to the shooting conditions, column random noise can be reliably detected in consideration of the shift.

  Furthermore, by using not only the VOPB area 152 but also the output signal from the dummy area 160, a sufficient number of lines necessary for noise detection for each column can be secured, and only the first few lines are common to the columns. The number of lines in the dummy area 160 is suppressed by using it for calculating the average value. Accordingly, it is possible to obtain sufficient noise detection accuracy while shortening the time required for photographing one frame as much as possible.

[Second Embodiment]
By the way, in the configuration of the noise removal processing unit 20a shown in the first embodiment, the difference average value calculation unit 22 requires an arithmetic circuit for subtraction, addition, and division. On the other hand, the second embodiment below is the same as the first embodiment in that the presence / absence of column random noise is determined based on the degree of variation in the signal level for each column with respect to the average value. However, the arithmetic processing procedure for the determination is greatly reduced to suppress the circuit scale and arithmetic processing load.

FIG. 11 is a diagram for explaining the outline of the noise detection method according to the second embodiment.
FIG. 11 shows an example of the frequency distribution of the output signal level from the column in which column random noise is generated among the columns of the noise determination detection region 172. As described above, in the frequency distribution in the case where column random noise is generated, it is known that sub-peaks appear at positions approximately the same distance above and below the center peak. In the present embodiment, basically, the number of pixels corresponding to the sub-peak is counted to determine whether or not column random noise is generated.

  Specifically, a noise detection range having a predetermined width above and below the center peak is set, the number of pixels whose output signal level is within this noise detection range is counted for each column, and the number is a predetermined value. If it is above, it is determined that column random noise has occurred. In FIG. 12, the threshold value thresh_LL to the threshold value thresh_LH is the lower noise detection range, and the threshold value thresh_HL to the threshold value thresh_HH is the upper noise detection range. In practice, pixel counting using such a noise detection range is executed for each color component.

  Incidentally, as described above, the position of the center peak can vary depending on the photographing conditions. Therefore, in the present embodiment, the center peak position (signal level) is calculated as an average value based on the output signal from the average calculation detection region 171 in the same procedure as in the second embodiment. Based on this average value, the upper and lower thresholds of the noise detection range are set each time. That is, in FIG. 12, level differences band_minus0, band_minus1, band_plus0, and band_plus1 from the average value ave to each threshold value of the noise detection range are determined in advance, and these level differences are added to or subtracted from the calculation result of the average value ave. By doing so, each threshold value is calculated.

FIG. 12 is a block diagram illustrating a configuration of a noise removal processing unit according to the second embodiment.
The noise removal processing unit 20b illustrated in FIG. 12 includes an average value calculation unit 21, a detection range setting unit 25, a counter 26, a comparison determination unit 27, and a signal correction unit 24. Among these, the average value calculation unit 21 and the signal correction unit 24 have the same functions as the corresponding blocks described with reference to FIG.

  In the same procedure as in the first embodiment, the average value calculation unit 21 calculates the average value of the signal level for each color component based on the output signal from the detection area 171 for average calculation, and then averages these values. The value is passed to the detection range setting unit 25. Based on these average values, the detection range setting unit 25 calculates threshold values thresh_LL, thresh_LH, thresh_HL, and thresh_HH that define the noise detection range.

Here, as an example, the average value of the R component is R_ave, the lower limit of the lower noise detection range for the R component, the level difference for calculating the upper limit is R_band_minus1, R_band_minus0, and the lower limit and upper limit of the upper noise detection range are Assuming that the level difference for calculation is R_band_plus0 and R_band_plus1, respectively, the lower limit and upper threshold values R_thresh_LL and R_thresh_LH of the lower noise detection range for the R component, and the lower limit and upper limit of the upper noise detection range, respectively. The threshold values R_thresh_HL and R_thresh_HH are obtained by the following equations (5) to (8), respectively.
R_thresh_LL = R_ave−R_band_minus1 (5)
R_thresh_LH = R_ave−R_band_minus0 (6)
R_thresh_HL = R_ave + R_band_plus0 (7)
R_thresh_HH = R_ave + R_band_plus1 (8)
Through the above processing, the detection range setting unit 25 calculates the threshold value of the noise detection range for each color component and sets it in the counter 26. Next, an output signal from the noise determination detection region 172 is supplied to the counter 26. The counter 26 determines whether or not the level of the input signal of each color component is within the corresponding noise detection range, and determines the number of pixels having the signal level included in the noise detection range for each column and color component. Count every time.

  When all output signals from the noise determination detection region 172 are supplied, the counter 26 passes the count value for each color component in each column to the comparison determination unit 27. The comparison determination unit 27 compares these count values with a predetermined determination threshold value, and determines that column random noise has occurred in the column when the count value exceeds the determination threshold value. Then, the signal correction unit 24 is notified of the column determined to generate noise.

  Here, FIG. 13 and FIG. 14 are diagrams for explaining the pixel counting operation in each of the column in which the column random noise is not generated and the column in which the column random noise is generated.

  FIG. 13 shows an example of the level distribution of the output signal in the column where no column random noise occurs, and FIG. 14 shows an example of the level distribution of the output signal in the column where column random noise occurs. Yes. In the figure, the region from the threshold value thresh_LL to the threshold value thresh_LH is the lower noise detection range, and the region from the threshold value thresh_HL to thresh_HH is the upper noise detection range. The counter 26 counts the number of pixels having the signal level within the noise detection range set in this way for each column.

  As shown in FIG. 13, if no column random noise occurs, the signal level of each pixel gathers in the vicinity of the average value ave. On the other hand, as shown in FIG. 14, when column random noise is generated, pixels having a signal level far from the average value ave exist. As described with reference to FIG. 11, when column random noise occurs, a sub peak appears on the frequency distribution of the signal level. Therefore, by setting the noise detection range to include the vicinity range of the sub peak, noise It can be determined that the more pixels having a signal level within the detection range, the higher the possibility that column random noise is generated.

  In order to count such pixels, as in the first embodiment, the noise determination detection area 172 includes a hundred lines or more so that a sufficient number of column random noise generation pixels are included. It is necessary to secure a certain number of lines. On the other hand, as in the first embodiment, it is only necessary to secure about several lines for the detection area 171 for average calculation.

  By the way, for example, after comparing the count values in the upper and lower noise detection ranges for the same column and the same color component, the comparison determination unit 27 compares these count addition values with a predetermined determination threshold value. When the count addition value exceeds the determination threshold value, it is determined that column random noise has occurred. The circuit configuration in this case is illustrated in FIG.

FIG. 15 is a diagram illustrating a configuration example of a noise determination circuit for each color component in the comparison determination unit.
FIG. 15 shows a configuration example of a circuit for column number xxx for determining whether or not noise has occurred for the R component as an example. In FIG. 15, comparators 201 and 202 compare R component input data R_data from noise determination detection region 172 with threshold values R_thresh_LL and R_thresh_HH, respectively. The AND gate 203 is a circuit that takes the logical product of the outputs of the comparators 201 and 202, and the selector 204 selectively selects one of the input data R_data and “0” according to the output value of the AND gate 203. Output. With such a configuration, when the input data R_data takes a value larger than the lower limit of the lower noise detection range and smaller than the upper limit of the upper noise detection range, the output of the AND gate 203 becomes H level, and the selector 204 To input data R_data.

  The comparator 205 outputs a value of “1” when the output value of the selector 204 is smaller than the threshold value R_thresh_LH, and the counter 206 counts the output value of the selector 204. As a result, the number of pixels having a signal level within the lower noise detection range is counted as the count value R_count_L_xxxx. The comparator 207 outputs a value of “1” when the output value of the selector 204 is greater than the threshold value R_thresh_HL, and the counter 208 counts the output value of the comparator 207. As a result, the number of pixels having a signal level within the upper noise detection range is counted as the count number R_count_H_xxxx.

  The adder 209 adds the count numbers R_count_L_xxxx and R_count_H_xxxx output from the counters 206 and 208, respectively, when detection of all the pixels in the noise determination detection region 172 is completed. The comparator 210 indicates that column random noise may be generated in this column when the count addition value from the adder 209 is larger than the determination threshold value R_judge_level for determining the occurrence of noise. The noise flag R_noise_flag_xxxx is set to “1”.

  As described above, after adding the count values in the upper and lower noise detection ranges, the count addition value is compared with the determination threshold value, for example, for the same column and the same color component. When the count values in the upper and lower noise detection ranges are individually compared with the judgment threshold value, and both count values exceed the judgment threshold value, column random noise occurs in that column. May be determined.

  When a plurality of color components are included in one column, it is necessary to output a final determination result based on the determination result for each color component (corresponding to the noise flag). FIG. 16 illustrates the configuration of the final determination circuit for this purpose.

FIG. 16 is a diagram illustrating a configuration example of a final determination circuit in the comparison determination unit.
16, noise flags R_noise_flag_xxxx, Gb_noise_flag_xxxx, Gr_noise_flag_xxxx, and B_noise_flag_xxx indicate the determination results for the R component, Gb component, Gr component, and B component, respectively. That is, when these flags are “1”, it is determined that column random noise is generated.

  In the case of the Bayer array, for example, if the odd column includes the R component and the Gb component, the even column includes the Gr component and the B component. In the example of FIG. 16, AND gates 221 and 222 are provided for each of the odd-numbered column and the even-numbered column, and if the flag value is “1” for both components on the column, the column random noise is finally generated in that column. It is determined that the noise has occurred, and the noise flag noise_flag_xxxx for each final column is set to “1”.

  As another configuration of the final determination circuit, for example, the noise flag value of each color component in one column is simply added, and column random noise occurs when the added value exceeds a predetermined reference value. It is good also as a structure which determines with having.

  In the detection range setting unit 25, the value of the level difference (band_minus0, band_minus1, band_plus0, band_plus1) for determining the lower limit and the upper limit of the noise detection range may be set to a different value for each color component. For example, for the G component (Gb component and Gb component) that is considered to be relatively easily affected by column random noise, the noise detection range may be set wider by reducing the level difference band_minus1, band_plus0. .

  Also, depending on the shooting conditions such as the gain applied to the pixel signal according to the brightness of the subject and the temperature at the time of shooting, the value of some or all of the level differences (band_minus0, band_minus1, band_plus0, band_plus1) for each color component May be changed. For example, the higher the ISO sensitivity at the time of shooting (that is, the higher the gain), the more noticeable the influence of noise generation. Therefore, the noise detection range may be expanded by reducing the level difference band_minus1, band_plus0. In this case, for example, each parameter corresponding to various conditions such as temperature and exposure time may be held as table data, and the parameter corresponding to the condition at that time may be set for each photographing.

  Through the above processing of the comparison / determination unit 27, the column in which the column random noise has occurred is notified to the signal correction unit 24. Thereafter, as in the first embodiment, when an output signal from the effective image area 151 is input to the signal correction unit 24, the signal correction unit 24 is determined to have generated column random noise. The output signal from the column is corrected based on the signal of the neighboring pixels. The correction method in the signal correction unit 24 is as described in the first embodiment.

  The signal correction unit 24 may be shared with a defective pixel correction function. Furthermore, in the present embodiment, the threshold thresh_LL indicating the lower limit of the lower noise detection range and the threshold thresh_HH indicating the upper limit of the upper noise detection range are thresholds for detecting defective pixels. It can also be used as a value. Therefore, the circuit scale and the manufacturing cost can be suppressed by combining the function of detecting defective pixels generated after product shipment with the noise removal processing unit 20b.

  According to the second embodiment described above, after setting the noise detection range based on the average value of the signal when not affected by the incident light, the signal level of the pixel in the noise determination detection region 172 is noise. By counting those within the detection range, it is possible to reliably detect the occurrence of column random noise.

  In addition, in the noise detection process after calculating the average value, after adding and subtracting to determine the noise detection range for each color component, basically only the signal level comparison and counting operation is performed for each column. Is called. Therefore, the circuit scale, processing load, processing time, and power consumption can be significantly suppressed as compared with the case where subtraction, addition, and division are performed for each column as in the first embodiment.

  In addition, since the average value is calculated each time shooting is performed, and the noise detection range is set based on the average value, even if a signal level shifts according to the shooting conditions, an error in column random noise due to the shift occurs. The possibility of occurrence of detection can be reduced.

  In each of the above embodiments, the calculation of the average value and the subsequent processing such as noise detection for each column are performed for each color component. However, these processes are not necessarily performed for each color component. . In other words, detection for these processes is performed from a signal that does not include a component corresponding to incident light, and therefore the sensitivity difference for each color component is considered to be very small.

  However, for example, the effect of column random noise on the output image is considered to be different for each color, so by performing the above processing for each color component, the threshold used for noise determination changes for each color component. For example, it is possible to perform noise detection processing according to the characteristics of each color component, and as a result, noise detection accuracy can be improved.

  Also, depending on the structure of the image sensor, for example, the column output may be independent for each color component. Alternatively, there is a configuration in which output signals from a plurality of same-color component pixels can be added (mixed) and output. As described above, since various output methods can be considered for output for each color component, such an output structure can be obtained by performing processing for each color component and outputting a noise determination result as described above. It is possible to give versatility to various configurations such as changing the final determination method according to the difference.

  In addition, at least a part of the functions of the noise removal processing units 20a and 20b described in the above embodiments is realized not by hardware processing in the image processing unit 12 but by software processing in the control unit 15 or the like, for example. Also good.

  In addition, as for the average value used as a reference for noise detection, if it is considered that the average value does not fluctuate greatly, it is not calculated and calculated for each shooting as described above, but a preset value is read and used. You may do it. In particular, when detecting column random noise after correcting the black level, the black level correction suppresses the signal level variation due to dark current, etc., so noise detection accuracy can be improved even if the average value is fixed. May be less affected. In addition, instead of obtaining an average value every time a picture is taken (for example, every time an image is recorded in response to a shutter operation), the average value is obtained at regular time intervals or at power-on and stored in a memory or the like. It may be configured to read this at the time of shooting.

  In each of the above embodiments, the average value is calculated based on the pixel signal taken in regardless of the column. This is because the position of the center peak on the frequency distribution shown in FIG. 5 etc. is determined mainly by the cause of the dark current, which is not so related to the position of the column. Therefore, priority is given to suppressing the processing load, processing time, circuit scale, and the like. For example, even if column random noise occurs in the first few lines of the dummy area 160 in some columns, the component is only included in several of the thousands of pixels, so when calculating the average value Therefore, it is considered that the influence of the component of the column random noise is negligibly small as compared with the noise component not related to the column such as dark current. On the other hand, regardless of the column, the average value calculation is performed based on the output signals from as few lines as possible, thereby reducing the processing load of the average value calculation, reducing the processing time, and temporarily storing the calculation results. Thus, the effect of reducing the circuit scale can be obtained.

  However, strictly speaking, it is considered that the position of the center peak is slightly different for each column due to the influence of variations in the characteristics of the output stage circuit of the image sensor 11a. In particular, in the method described in the second embodiment, the noise detection range to be counted for determining the occurrence of column random noise is a limited range based on the position of the center peak. For this reason, there is a possibility that the influence of the position shift of the center peak on the noise detection accuracy becomes relatively large. On the other hand, the reference value for column random noise detection is optimized by increasing the number of lines in the detection area 171 for average calculation and calculating the average value based on the output signal for each column. The detection accuracy can be increased.

It is a block diagram which shows the principal part structure of the imaging device which concerns on embodiment. It is a figure which shows the structural example of the image pick-up element in a camera block, and its peripheral circuit. It is a figure which shows roughly the structure of the light receiving element area | region of an image pick-up element. It is a figure which shows typically the detection area in an image sensor. It is a figure which shows the example of the frequency distribution of the pixel signal level in a dummy area | region and a VOPB area | region. It is a block diagram which shows the structure of the noise removal process part which concerns on 1st Embodiment. It is a figure which shows the example of the signal level distribution in the column in which the column random noise has not generate | occur | produced. It is a figure which shows the example of the signal level distribution in the column in which the column random noise has generate | occur | produced. It is a figure for demonstrating the position of the pixel used for signal correction | amendment, when the pixel of correction object respond | corresponds to R component. It is a figure for demonstrating the position of the pixel used for signal correction | amendment, when the pixel of correction object respond | corresponds to Gb component. It is a figure for demonstrating the outline | summary of the noise detection method in 2nd Embodiment. It is a block diagram which shows the structure of the noise removal process part which concerns on 2nd Embodiment. It is a figure for demonstrating the count operation of the pixel in the column in which the column random noise has not generate | occur | produced. It is a figure for demonstrating the count operation of the pixel in the column in which the column random noise has generate | occur | produced. It is a figure which shows the structural example of the noise determination circuit for every color component in a comparison determination part. It is a figure which shows the structural example of the final determination circuit in a comparison determination part.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 11 ... Camera block, 11a ... Image sensor, 12 ... Image processing part, 13 ... SDRAM, 14 ... Display part, 15 ... Control part, 16 ... Flash memory, 17 ... External memory controller, 17a ...... External memory, 18... Operation unit, 20a, 20b... Noise removal processing unit, 21... Average value calculation unit, 22... Difference average value calculation unit, 23, 27. Signal correction unit, 25 …… Detection range setting unit, 26 …… Counter

Claims (18)

In a noise detection apparatus for detecting noise from an output signal from an XY address scanning type solid-state imaging device,
Before capturing an output signal from an effective image area that receives incident light on the solid-state image sensor, an output signal system for each column of the solid-state image sensor includes a signal component corresponding to the incident light to the solid-state image sensor. A signal capturing unit that captures the black dummy signal that is output in a line unit on the solid-state image sensor,
A reference value acquisition unit for acquiring a reference value corresponding to an average value of the signal level of the black dummy signal;
Accepts the input of the black dummy signal for a predetermined number of lines from the signal capturing unit, and generates random noise based on the signal level for each column of the black dummy signal and the reference value from the reference value acquisition unit A noise determination unit for determining a column that is
A noise detection apparatus comprising:   2. The noise according to claim 1, wherein the noise determination unit determines that the random noise is generated in a column having a large variation degree of the signal level of the input black dummy signal with respect to the reference value. Detection device.   The noise determination unit compares the signal level for each column of the black dummy signal for the predetermined number of lines with a threshold value defined according to the reference value from the reference value acquisition unit, The noise detection apparatus according to claim 2, wherein a column in which the random noise is generated is determined based on a comparison result.   The noise determination unit sets a noise detection range defined by the threshold value at a position separated by a predetermined signal level above and below the reference value from the reference value acquisition unit, and inputs Based on the black dummy signal, the pixels having the output signal level within the noise detection range are counted, and the random noise is generated in the column whose count value exceeds a predetermined determination threshold value. The noise detection device according to claim 3, wherein the noise detection device is determined.   The noise detection device according to claim 4, wherein the noise determination unit changes a relative level difference of at least one of an upper limit or a lower limit of the noise detection range with respect to the reference value according to a shooting condition. . The reference value acquisition unit acquires the reference value for each color component,
The noise determination unit executes a determination process based on the signal level of the input black dummy signal and the reference value for each color component, and sets the noise detection range to a different range for each color component. The noise detection device according to claim 4, wherein   The noise determination unit calculates a difference value between the signal level of the black dummy signal for the predetermined number of lines and the reference value, and determines a column in which the random noise is generated based on the difference value for each column. The noise detection device according to claim 2, wherein the determination is performed.   The noise determination unit calculates a difference average value that is an average of the difference values for each column, and determines that the random noise is generated in a column in which the difference average value exceeds a predetermined determination threshold value. The noise detection device according to claim 7, wherein   The noise detection apparatus according to claim 8, wherein the noise determination unit changes the determination threshold according to a shooting condition. The reference value acquisition unit acquires the reference value for each color component,
The noise determination unit performs a determination process based on the signal level of the input black dummy signal and the reference value for each color component, and sets the determination threshold to a different value for each color component. The noise detection device according to claim 8.   The reference value acquisition unit accepts the black dummy signal having a smaller number of lines than the number of lines input to the noise determination unit from the signal capturing unit immediately before the black dummy signal is input to the noise determination unit, The noise detection apparatus according to claim 1, wherein the reference value is obtained by calculating the average value of the signal level of the black dummy signal.   The noise detection apparatus according to claim 1, wherein a digital output circuit is individually provided for each column in an output stage of the solid-state imaging device.   13. The noise detection apparatus according to claim 12, wherein the signal capturing unit captures at least a part of the black dummy signal by causing each digital output circuit to output in a state without an input signal.   The black dummy signal captured by the signal capturing unit includes the first signal captured by outputting each digital output circuit without an input signal, and the solid-state imaging immediately after capturing the first signal. The noise detection apparatus according to claim 12, further comprising: a second signal output from a pixel in a light shielding area adjacent to the effective image area in a vertical direction in the element. The reference value acquisition unit receives, from the signal capturing unit, the black dummy signal having a smaller number of lines than the number of lines input to the noise determination unit immediately before the black dummy signal is input to the noise determination unit. The reference value is obtained by calculating the average value of the signal level of the black dummy signal,
The first signal is input as the black dummy signal to the reference value acquisition unit, and then the first signal and the second signal as the black dummy signal to the noise determination unit. The noise detection device according to claim 14, wherein: is input.   Of the signals output from the effective image area of the solid-state image sensor after the black dummy signal is input to the noise determination unit, the random determination unit determines that the random noise is generated by the noise determination unit. 2. The noise detection apparatus according to claim 1, further comprising a signal correction unit that corrects the signal based on an output signal from a pixel of the same color component in the vicinity thereof. In an imaging apparatus that captures an image using an XY address scanning type solid-state imaging device,
Before capturing an output signal from an effective image area that receives incident light on the solid-state image sensor, an output signal system for each column of the solid-state image sensor includes a signal component corresponding to the incident light to the solid-state image sensor. A signal capturing unit that captures the black dummy signal that is output in a line unit on the solid-state image sensor,
A reference value acquisition unit for acquiring a reference value corresponding to an average value of the signal level of the black dummy signal;
Accepts the input of the black dummy signal for a predetermined number of lines from the signal capturing unit, and generates random noise based on the signal level for each column of the black dummy signal and the reference value from the reference value acquisition unit A noise determination unit for determining a column that is
An imaging device comprising: In a noise detection method for detecting noise from an output signal from an XY address scanning type solid-state imaging device,
Before the signal capturing unit captures an output signal from an effective image area that receives incident light on the solid-state image sensor, the output signal system for each column of the solid-state image sensor responds to the incident light to the solid-state image sensor. The black dummy signal output so as not to include the received signal component is taken in line units on the solid-state image sensor,
The reference value acquisition unit acquires a reference value corresponding to the average value of the signal level of the black dummy signal,
A noise determination unit receives an input of the black dummy signal for a predetermined number of lines from the signal capturing unit, and based on the signal level for each column of the black dummy signal and the reference value from the reference value acquisition unit , Determine the column where random noise occurs,
The noise detection method characterized by the above-mentioned.

Images