JPWO2015045486A1 - Imaging device - Google Patents

Abstract

An object of the present invention is to reduce an extreme value exclusion average circuit that follows an environment by performing an operation of shifting to an appropriate threshold without rearranging data while excluding extreme values that should be excluded from the average. It is to be realized by logic. An imaging apparatus according to the present invention is an imaging apparatus including an imaging unit, a correction value output unit, and a subtraction unit, and the correction value output unit is a signal level for each pixel output from a light-shielded pixel region of the imaging unit. A pixel signal having a signal level within a predetermined level is selected based on a plurality of threshold values, the selected pixel signal is averaged, and the averaged signal is output as a correction value signal. A signal output from the pixel region is subtracted by a correction value signal.

Description

  The present invention relates to an imaging apparatus, and more particularly to reduction of noise included in an image signal output from an imaging element.

  The image sensor has a minute output difference for each pixel, and is known as fixed pattern noise. In addition, some of the pixels have a defect, and it is also known that there is a pixel that always shows a protruding output as a pixel defect. Further, it is common to offset subtract the optical black pixel from the video pixel on the circuit side of the imaging device using a reference optical black (light-shielding) pixel in order to set the output at the time of light shielding to 0 output.

  Conventionally, after correcting dark current unevenness in the vertical direction of the screen of a signal acquired from four lines of vertical light-shielded pixels read out before effective pixels on the light-receiving surface of the solid-state imaging device, each vertical pixel signal of the four lines of vertical light-shielded image is corrected. The second value is calculated from the minimum value, stored as a vertical smear correction signal, and the gain is varied in accordance with the image AGC, and is subtracted from the image signal after AGC output from the effective pixel on the light receiving surface of the solid-state imaging device. . The signal output from the solid-state image sensor is A / D converted to 14 bits to calculate the representative value signal, attenuated to 15/16, and output from the effective pixels on the light receiving surface of the solid-state image sensor. A technique for subtracting from an image signal is disclosed (for example, see Patent Document 1).

JP 2008-109639 A

  According to the conventional general method, which is an optical black (OB) region and simple averaging over the same pixel, a pixel having a defect such as a white scratch or a pixel having a defect that behaves by excitation or blinking due to radiation. There is a problem that values that should not be added to the averaging due to the above are averaged together.

  An object of the present invention is to find a value that should be averaged following an environmental change without adding a value that should not be averaged to an average operation in a correction logic circuit for offset output fluctuation or fixed pattern noise of an image sensor, and Alternatively, a logic circuit that automatically performs an average operation without depending on an external calculation circuit is to be realized with a small circuit configuration in accordance with the purpose.

  The imaging device of the present invention is an imaging device including an imaging unit, a correction value output unit, and a subtraction unit, and the correction value output unit outputs a plurality of signal levels for each pixel output from the light-shielded pixel region of the imaging unit. A pixel signal having a signal level within a predetermined level is selected based on the threshold value of the pixel, the selected pixel signal is averaged, the averaged signal is output as a correction value signal, and the subtraction unit is a photosensitive pixel region of the imaging unit The signal output from the subtractor is subtracted by a correction value signal.

  The imaging device according to the present invention is the above-described imaging device, and holds a correction value holding register unit that holds a correction value signal as a comparison reference value, a control unit that controls the start and end of averaging, and a plurality of threshold values. A setting unit, a comparison reference value, a difference detection unit that calculates a difference absolute value of a new correction value signal, and a threshold detection that determines whether or not the magnitude of the difference absolute value exceeds a plurality of thresholds A determination unit, an addition unit that adds the output of the threshold determination unit and the pixel signal, a threshold value pixel data appearance count determination unit that increments a counter when it is determined that the threshold determination is within the threshold for a plurality of thresholds, and a threshold A selection unit that selects a minimum value that exceeds an output value of the inner pixel data appearance number determination unit that exceeds a predetermined value and selects an output of the addition unit that uses the same threshold, and an output of the selected addition unit Appearance of pixel data within the selected threshold In the output of the number determination unit, the normalization unit that normalizes the addition value based on the threshold value that is divided and used to obtain an average value, and the output of the normalization unit is held for each area that is addressed by the control unit. The correction value holding memory unit that can output the value and the output of the correction value holding memory unit are held by the correction value holding register unit at the timing specified by the control unit, and used as the comparison reference value and the averaged output to the subordinate circuit It has an averaging circuit configuration.

  The imaging apparatus of the present invention is the above-described imaging apparatus, wherein the selection unit that selects the output of the addition unit and the output of the in-threshold pixel data appearance frequency determination unit is switched according to the threshold value.

  The imaging device according to the present invention is the above-described imaging device, wherein the correction value output unit determines a signal level for each pixel of the horizontal line output from the light-shielded pixel region of the imaging unit based on a plurality of thresholds, and a signal within a predetermined range A level pixel signal is selected, the selected pixel signal is averaged, and the averaged signal is output as a correction value signal.

  The imaging device of the present invention is the above-described imaging device, wherein the correction value output unit further includes a correction value holding memory unit, and the correction value holding memory unit is output from the correction value signal and the light-shielded pixel region of the imaging unit. The signal level for each pixel is stored.

  The imaging apparatus according to the present invention is the above-described imaging apparatus, further including a plurality of frame memories, wherein the frame averaging circuit of the pixel signal is configured using a plurality of frame memories.

  The imaging apparatus according to the present invention is the above-described imaging apparatus, wherein the correction value output unit outputs a correction value signal, and again determines a signal level for each pixel output from the light-shielded pixel region of the imaging unit with a plurality of threshold values. Then, a pixel signal having a signal level within a predetermined level is selected, the selected pixel signal is averaged, and the averaged signal is output as a correction value signal.

  An imaging apparatus according to the present invention is the above-described imaging apparatus, wherein the correction value signal is generated by multiplying the information stored in the frame memory by a predetermined coefficient.

  The imaging apparatus of the present invention is the above-described imaging apparatus, further including a clip unit, wherein the clip unit suppresses a signal output from the light-shielding pixel region of the imaging unit with a predetermined value.

  The imaging device of the present invention is the above-described imaging device, and selects an output value of the in-threshold pixel data appearance number determination unit that shows a minimum value exceeding a predetermined value, and an addition unit using the same threshold value A plurality of threshold values are output from a selection unit that selects the information on the selected threshold value and a threshold value determination unit that detects whether the absolute value of the difference exceeds the threshold value. More than the number of threshold determination results, it has a system for inputting threshold information to be output to the threshold determination results.

According to the present invention, in order to reduce noise, when averaging image signals obtained from pixels in the optical black region of the image sensor, it is appropriate to exclude the extreme values that should be excluded from the average, without rearranging the data. By performing an operation to shift to a threshold value, an extreme value exclusion average circuit that follows the environment can be realized with a small-scale logic.
Furthermore, the method of calculating an average from near-difference values using a difference between an expected value, that is, a correction value up to now and new correction source data according to the present invention, can follow a change in environment such as temperature.

It is a block diagram for demonstrating the imaging device which is one Example of this invention. It is a block diagram for demonstrating operation | movement of the correction value output part which is one Example of this invention. It is a figure for demonstrating the light-shielding pixel area | region and photosensitive pixel area | region of an image pick-up element. It is a figure for demonstrating operation | movement of the correction value output part which is one Example of this invention. It is a figure for demonstrating the operation | movement of the threshold value transition control of the correction value output part which is one Example of this invention. It is a figure for demonstrating the data storage to a frame memory. It is a figure for demonstrating the operation | movement of several frame same pixel averaging which is one Example of this invention. It is a figure for demonstrating the operation | movement of the re-averaging of the correction value output part which is one Example of this invention. It is a figure for demonstrating the operation | movement of the environmental change tracking fixed noise removal of the correction value output part which is one Example of this invention.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram for explaining an imaging apparatus according to an embodiment of the present invention.
In FIG. 1, the imaging apparatus 100 includes a SENS unit 1, a subtraction unit 2, a correction value output unit 3, a subsequent stage unit 4, and a control unit 5.
The SENS unit 1 includes an image pickup unit 101 having an image pickup device as an image pickup unit and an A / D conversion unit (Analog-to-Digital Converter) 102, and light input to the image pickup unit through an optical component (not shown). Is converted into an electrical signal, and a digital output signal 1a is output in a series of operation units until the electrical signal is further converted into a digital signal that can be handled by the logic circuit in the A / D conversion unit. The SENS unit 1 is controlled by a control signal 5 a output from the control unit 5.

The subtraction unit 2 receives the digital output signal 1a from the SENS unit 1, performs subtraction processing on the value of the correction value output signal 3a, which is another input, and outputs a corrected video signal 2a.
The correction value output unit 3 receives the digital output signal 1a from the SENS unit 1 and outputs a correction value output signal 3a. The correction value output unit 3 includes, for example, a clip unit 30, an expected value difference determination average circuit 300, a correction value holding memory unit 38, and a correction value holding register unit 39. The frame memory 40 is included in the correction value holding memory unit 38.
The correction value output signal 3a is also used by the correction value output unit 3 to input a reference value in an expected value difference determination averaging circuit 300 that is an averaging circuit using expected value difference determination that operates internally.
Further, in order to determine the operation condition, the correction value output unit 3 receives the adjustment data from the control unit 5 as the camera adjustment value setting output and the operation switching timing as the camera timing output 4c. The adjustment data is the average range%, the threshold, and the like, and the operation switching timing is timing information for address switching of the internal memory and rewriting of the contents of the data register.

The rear stage unit 4 receives the corrected video signal 2a and outputs a camera video output 4a.
The video output 4a outputs a camera video that can be connected to another video device based on the corrected video signal.
The control unit 5 controls the entire imaging apparatus 100. For example, the control unit 5 outputs control signals 5a, 5b, 5c, and the like by inputting 3b, 4b, and the like.

FIG. 2 is a block diagram for explaining the operation of the correction value output unit 3 according to an embodiment of the present invention.
The correction value output unit 3 includes a clip unit 30, a difference detection unit 31, a threshold determination unit 32, an addition unit 33, an intra-threshold pixel data appearance frequency detection unit 34, a setting unit 35, a selection unit 36, a normalization unit 37, and a correction value. A holding memory unit 38 and a correction value holding register unit 39 are included.
The expected value difference determination average circuit 300 in FIG. 1 includes, for example, the difference detection unit 31, the threshold determination unit 32, the addition unit 33, the in-threshold pixel data appearance frequency detection unit 34, the setting unit 35, and the selection unit 36 in FIG. , A normalizing unit 37 and a setting unit 35.

  The clip unit 30 clips the digital output signal 1a from the SENS unit 1 to the bit width used for correction, limits the bit width processed by the correction value output unit 3, and outputs the digital output signal 30a. Note that the processing bit range need not be limited as long as it is not necessary.

The difference detector 31 obtains the absolute value of the difference from the reference value input 3a and the pixel data input 30a, and outputs an absolute difference output 31a.
The threshold value determination unit 32 has a plurality of threshold values therein, and outputs a threshold value determination output group 32a that indicates by digital values whether or not the difference absolute value output 31a exceeds each threshold value. The threshold determination unit 32 may input the threshold setting data group 35b so that the user can set the threshold, or may input the threshold transition control signal 38 to automatically change the threshold. There may be no fixed thresholds.
For example, when the threshold setting data group 35b has three types of thresholds abc, a threshold a of 35ba, a threshold b of 35bb, and a threshold c of 35bc are input.

The adding unit 33 is an adding circuit that receives the pixel data input 30a and the threshold determination output group 32a and adds the value 30a of the pixel data input to each threshold only when the threshold is not exceeded. After the reset, the addition result for each threshold is output as an addition output group 33a after a period in which the addition operation is repeatedly averaged.
For example, when there are three types of threshold values abc, the addition output group 33a outputs an addition result for the threshold value a of 33aa, an addition result for the threshold value b of 33ab, and an addition result for the threshold value c of 33ac.

The threshold value pixel data appearance count determination unit 34 is an addition circuit that counts only when the threshold determination output group 32a is input and does not exceed the threshold value, and is reset by the camera timing output 4c. The number of times that the threshold value with respect to each threshold value has fallen after the period in which the operations are repeatedly averaged is output as the appearance number group 34a.
For example, in the appearance number group 34a, when there are three types of threshold values abc, the number of times that the threshold value a of 34aa is below, the number of times that the threshold value b of 34ab is below, and the number of times that the threshold value c of 34ac is below is output.

The setting unit 35 receives the camera adjustment value setting output 4b as input, holds the setting for the correction value output unit 3, and gives it to the internal circuit.
The usage average range% setting 35a is a setting value for designating how many times the number of occurrences of the threshold among the parameters used for averaging exceeds the percentage. The threshold setting data group 35b is a setting value that gives each threshold used in the threshold determination unit 32, and may not be present depending on the configuration. In addition, the parameter range used for averaging and the start and end values of timing are held.

  The selection unit 36 selects the one having the lowest threshold value from among the appearance number group 34a that exceeds the number of appearances by the use average range% setting 35a, and outputs the appearance number 36a of the selected threshold value 36a. .

The selection unit 36 selects an addition result for the same threshold as the number of appearances of the threshold selected in the addition output group 33a, and outputs the result as an addition result 36b for the selected threshold.
For example, there are three threshold values abc, the addition result for 33aa threshold a, the addition result for 33ab threshold b, the addition result for 33ac threshold c, the number of times below 34aa threshold a, and the 34ab threshold b. If the number of times below the threshold c of 34ac is selected as the number of appearances of the selected threshold of 36a, the addition result for the selected threshold of 36b The addition result for the threshold value c of 33ac is selected.

Further, the selection unit 36 may output a threshold value transition control signal 36c in order to automatically control and change the threshold value in the threshold value determination unit 32.
For example, although the threshold determination unit 32 has eight thresholds, the threshold determination output group 32a outputs only the results of five thresholds, and the addition unit 33 and the in-threshold pixel data appearance frequency determination group unit 34 The threshold value transition control signal 38b is used when the number of circuits is reduced so that there are only five adders. When the number of appearances 36a of the selected threshold value is not the midpoint of the five currently used threshold values, the threshold value to be used is changed up and down among the eight threshold values. This threshold value transition control signal 36c is stored in the correction value holding memory unit 38 together with threshold control for the average of the same coordinates, and the next time the same place is averaged, the threshold value transition control signal 38b is sent from the correction value holding memory unit 38. Used as

  The normalizing unit 37 divides the addition result 36b for the selected threshold value by the number of appearances 36a of the selected threshold value, and outputs an extreme value exclusion averaged output 37a.

The correction value holding memory unit 38 receives the extreme value exclusion averaged output 37a and the camera timing output 4c, and at the place where the address is designated based on the timing when the averaging is completed and the value to be held as the correction value is determined. Store the data. The frame memory 40 is included in the correction value holding memory unit 38.
Further, at a timing other than data storage, the data at the location designated by the address is output as the correction value memory output 38a. When the threshold transition control signal 36c is used, it may be stored as an additional signal at the same time and output as the threshold transition control signal 38b.

The correction value holding register unit 39 receives the correction value memory output 38a and the camera timing output 4c, and stores the correction value memory output 38a, which is the average result at the same previous address, at the timing when the averaging is started. The correction value output signal 3a is output. The correction value output signal at this time serves as a reference value input for the difference detector 31.
In addition, the correction value holding register unit 39 finishes averaging and stores the extreme value exclusion averaged output 37a as a correction value in the correction value holding memory unit 38, that is, the correction value holding memory unit 38 performs a new correction. When the value is output to the correction value memory output 38a, it is stored again in the register and the correction value output signal 3a is output.

FIG. 3 is a diagram for explaining the light-shielded pixel region and the photosensitive pixel region of the image sensor.
The photosensitive pixel area 10 indicates a photosensitive pixel area of the image sensor, and the image sensor can output an image based on an optical image such as a lens.
The light-shielded pixel region 11 indicates a light-shielded pixel region of the image sensor, and is a region where pixels having the same characteristics as the photosensitive pixel in the photosensitive pixel region 10 are always shielded from light.
The all pixel area 12 represents all the pixel areas of the image sensor, and behaves as pixel coordinates in terms of external driving operation. What are the photosensitive pixels in the photosensitive pixel area 10 and the light-shielded pixels in the light-shielded pixel area 11 of the image sensor? Including parts with different characteristics.

The arbitrary line 13 is an enlarged view of the leading portion of the arbitrary line of the image sensor.
A light shielding pixel 14 indicates a light shielding pixel of the image sensor. In this example, 24 pixels are arranged in the horizontal left portion.
A photosensitive pixel 15 indicates a photosensitive pixel of the image sensor.

FIG. 4 is a diagram for explaining the operation of the correction value output unit according to the embodiment of the present invention.
FIG. 4 shows the effect of averaging the 24 pixels by the light-shielded pixels 14 in FIG. 3 with the block configuration of FIG.

14aa shows the output level of the light-shielded pixel at the start of processing. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels.
14ab indicates the output level of the absolute difference value when the reference value is indefinite. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The vertical direction is expressed in a different scale by matching the position of the pixel in the horizontal direction with respect to the output level of the light-shielded pixel at the start of processing 14aa. Since no reference value exists at the start of processing, it is undefined.
In this example, the case where 0 is cleared is shown. However, regardless of the indefinite value, the output of the difference is increased as a whole, and the range having a large threshold value shown below is automatically selected, and all the target pixels are selected. There is no problem because it is only an average.

  14t indicates a threshold level scale. The scale in the vertical direction is matched to the output of the absolute difference value such as 14ab, and the threshold scales of TH0, TH1, TH2, TH3, TH4, and TH5 are shown. Here, an example in which each threshold is doubled is shown.

  14ta indicates the threshold level selected in 14aa. It expresses that TH6 is double TH5. In this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all the target pixels is included within the threshold. Even when random noise is superimposed on the output of light-shielded pixels due to changes in the temperature environment, or when a DC addition occurs and the absolute value of the difference becomes large, all objects automatically Since the threshold level including the absolute difference value of 16 pixels corresponding to two-thirds of the pixels is selected, the convergence operation can be performed following the threshold level.

14ac indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time.
14ae is an output level obtained by simply averaging 14aa. It includes a 14ac synchronic flaw, so the level is lower than the average that can be expected by visual inspection.

  Reference numeral 14af indicates an output obtained by averaging 14aa according to the present invention and a reference value level for 14ba. Since the reference value used for determination of the absolute difference value at the start of processing is indefinite and the differential absolute value of all target pixels is below the threshold level of 14 ta, there is no pixel that is not subject to averaging. The level is no different from the simple average output level of 14ae. However, the effect of the present invention appears in the subsequent frame. At the video frame rate for outputting several tens of images per second, there is no problem because the image is converged in several frames from the start of processing.

14ba indicates the output of the light-shielded pixel in the next frame of 14aa. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14bb indicates an output level of an absolute difference value from 14ba using 14af as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched with the output level of 14ba, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is expressed with a scale that matches the vertical direction. Since 14af exists as a reference value, it can be seen that the difference absolute value is smaller than 14ab and converged.

14tb indicates the threshold level selected in 14ba. It expresses that it becomes TH2. As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.
14bc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
Reference numeral 14bd indicates a flaw due to radiation or a blinking pixel defect. If there is a radiation collision or a blinking pixel defect, a steep noise level is generated in local time and space, which becomes a hotbed of average error.

14be is an output level obtained by simply averaging 14ba. It includes 14 bc synchronous flaws, but also includes more prominent 14 bd radiation and flashing pixel defect flaws, which makes the level much higher than it should be visually expected.
Reference numeral 14bf indicates an output obtained by averaging 14ba according to the present invention and a reference value level for 14ca. Since the operation is set to leave at least two thirds of all the target pixels in this example using the absolute difference value of 14 bb, a value greatly deviating exceeding the threshold level of 14 tb is excluded and averaged. It can be seen that, unlike the simple average, the value is closer to the average value that should be more than the previous reference value of 14af.

14ca indicates the output of the light-shielded pixel of the next frame of 14ba. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14cb indicates an output level of an absolute difference value from 14ca using 14bf as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched with the output level of 14ca, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is expressed with a scale that matches the vertical direction. As a reference value, 14bf having an average accuracy higher than that of 14af exists, so that it can be seen that the absolute value of the difference becomes smaller and converged than 14bb.

14tc indicates the threshold level selected in 14ca. It expresses that it becomes TH1. As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.
14cc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
14ce is an output level obtained by simply averaging 14ca. Since it includes a 14 cc synchronism flaw, the level is lower than the average that can be expected visually as in the case of 14ae.

  14cf indicates the output averaged by 14ca according to the present invention and the level of the reference value for 14da. Since the operation is set to leave at least two thirds of all the target pixels in this example using the absolute difference value of 14 cb, a value greatly deviating beyond the threshold level of 14 tc is excluded and averaged. Unlike the previous average of 14 cf, the value is closer to the average value that should be present with the accuracy of the portion that is not noticed by visual inspection of the figure, even more than the previous 14 cf reference value.

14da shows the output of the shading pixel of the next frame of 14ca. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14db indicates an output level of an absolute difference value from 14da using 14cf as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched to the output level of 14da, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is represented by a scale that matches the vertical direction. Although 14cf whose average accuracy is higher than 14bf is used as a reference value, in this example, since it is almost converged, the difference in absolute value of difference cannot be seen from the appearance of 14cb.

14td indicates the threshold level selected at 14da. It expresses that it becomes TH1. As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.
14 dc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
14de is an output level obtained by simply averaging 14da. Since it includes 14 dc synchronous flaws, the level is lower than the average that can be expected visually as in 14 ae.

  Reference numeral 14df denotes an output obtained by averaging 14da according to the present invention and a reference value level in the next frame. Since the operation is set to leave at least two-thirds of all target pixels in this example using the absolute difference value of 14 db, a value greatly deviating beyond the threshold level of 14 td is removed and averaged. Unlike the simple average of, like the previous 14cf reference value, the average value that should be present is shown. In this example, the previous reference value of 14cf and the reference value of 14df are the same value, and are converged to an average value that should be present.

While the simple average of 14ae, 14be, 14ce, and 14de is always affected by the synchronous noise and the influence of external noise such as radiation, the average of 14af, 14bf, 14cf, and 14df according to the present invention and It can be seen that the reference value converges with increasing average accuracy each time the frame is repeated.
In addition, as described above, even when the reference value and the input value are shifted due to environmental changes, the absolute difference value is obtained and the operation setting is performed so that two-thirds or more of all target pixels are left. The range can be automatically expanded, and the convergence operation can be continued from a condition similar to 14af.
This automatic tracking is a feature of the present invention that cannot be obtained by a method using a fixed threshold value using only a conventional input value.

FIG. 5 is a diagram for explaining the operation of threshold value transition control of the correction value output unit according to the embodiment of the present invention.
FIG. 5 shows the effect of averaging the 24 pixels by the light-shielding pixels 14 in FIG. 3 with the block configuration of FIG. In particular, the threshold transition control signal 36c is used to reduce the circuit scale, and the threshold determination output group 32a is dynamically changed by the threshold transition control signal 38b. In this embodiment, the number of internal circuits in the selector 36 can be reduced.

14aa shows the output level of the light-shielded pixel at the start of processing. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels.
14ab indicates the output level of the absolute difference value when the reference value is indefinite. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The vertical direction is expressed in a different scale by matching the position of the pixel in the horizontal direction with respect to the output level of the light-shielded pixel at the start of processing 14aa. Since no reference value exists at the start of processing, it is undefined. In this example, the case where 0 is cleared is shown. However, regardless of the indefinite value, the output of the difference is increased as a whole, and the range having a large threshold value shown below is automatically selected, and all the target pixels are selected. There is no problem because it is only an average.

14ts shows the level scale of the variable threshold output. The scale in the vertical direction is matched to the output of the absolute difference value such as 14ab, and four dynamic threshold scales THs0, THs1, THs2, and THs3 are shown. Here, each threshold value has a double relationship, and indicates that a rough threshold value is started when the operation starts or when the environment changes.
14tsa indicates the threshold level selected in 14aa. Since the reference value and the threshold value transition control signal are indefinite, the threshold value is started from the coarsest state, and the smallest threshold value THs0 is out of scale. Therefore, all pixels are averaged while THs0 is selected.

Since THs0 has been selected, the threshold value transition control of the absolute value of the threshold value assigned to THs0 is made THs2, which is the center of the threshold judgment output group 32a in this example.
In this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all the target pixels is included within the threshold. Even when random noise is superimposed on the output of light-shielded pixels due to changes in the temperature environment, or when a DC addition occurs and the absolute value of the difference becomes large, all objects automatically Since the threshold level including the absolute difference value of 16 pixels corresponding to two-thirds of the pixels is selected, the convergence operation can be performed following the threshold level.

14ac indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time.
14ae is an output level obtained by simply averaging 14aa. It includes a 14ac synchronic flaw, so the level is lower than the average that can be expected by visual inspection.
Reference numeral 14af indicates an output obtained by averaging 14aa according to the present invention and a reference value level for 14ba. Since the reference value used for determination of the absolute difference value at the start of processing is indefinite and the differential absolute value of all target pixels is below the threshold level of 14 tsa, there are no pixels that are not subject to averaging. The level is no different from the simple average output level of 14ae. However, the effect of the present invention appears in the subsequent frame. At the video frame rate for outputting several tens of images per second, there is no problem because the image is converged in several frames from the start of processing.

14ba indicates the output of the light-shielded pixel in the next frame of 14aa. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14bb 'indicates the output level of the absolute value of the difference from 14ba using 14af as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched with the output level of 14ba, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is expressed with a scale that matches the vertical direction. Since 14af exists as a reference value, it can be seen that the difference absolute value is smaller than 14ab and converged.

14tsb indicates the threshold level selected in 14ba. It represents that THs0 after the threshold transition control is performed. Since THs0 has been selected, the threshold value transition control of the absolute value of the threshold value assigned to THs0 is made THs2, which is the center of the threshold judgment output group 32a in this example.
As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.

14bc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
Reference numeral 14bd indicates a flaw due to radiation or a blinking pixel defect. If there is a radiation collision or a blinking pixel defect, a steep noise level is generated in local time and space, which becomes a hotbed of average error.

14be is an output level obtained by simply averaging 14ba. It includes 14 bc synchronous flaws, but also includes more prominent 14 bd radiation and flashing pixel defect flaws, which makes the level much higher than it should be visually expected.
14bf 'indicates the output averaged by 14ba according to the present invention and the level of the reference value for 14ca. Since the operation is set to leave at least two-thirds of all target pixels in this example using the absolute difference value of 14bb ′, a value greatly deviating beyond the threshold level of 14tsb is removed and averaged. It can be seen that, unlike the simple average of 14be, the value is closer to the average value that should be more than the reference value of the previous 14af.

14ca indicates the output of the light-shielded pixel of the next frame of 14ba. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14cb ′ indicates an output level of an absolute difference value from 14ca using 14bf ′ as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched with the output level of 14ca, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is expressed with a scale that matches the vertical direction. As a reference value, 14bf having an average accuracy higher than that of 14af exists, so that it can be seen that the absolute value of the difference becomes smaller and converged than 14bb.

  14tsc indicates the threshold level selected at 14ca. It represents that THs0 after the threshold transition control is performed. Since THs0 has been selected, the threshold value transition control of the absolute value of the threshold value assigned to THs0 is made THs2, which is the center of the threshold judgment output group 32a in this example. As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.

14cc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
14ce is an output level obtained by simply averaging 14ca. Since it includes a 14 cc synchronism flaw, the level is lower than the average that can be expected visually as in the case of 14ae.

  14cf indicates the output averaged by 14ca according to the present invention and the level of the reference value for 14da. Since the operation is set to leave at least two thirds of all the target pixels in this example using the absolute difference value of 14 cb, a value greatly deviating exceeding the threshold level of 14 tsc is excluded and averaged. Unlike the previous average of 14 cf, the value is closer to the average value that should be present with the accuracy of the portion that is not noticed by visual inspection of the figure, even more than the previous 14 cf reference value.

14da shows the output of the shading pixel of the next frame of 14ca. The position of the vertical direction level is aligned with the output level of the light-shielding pixel at the start of the process 14aa and is listed in the horizontal direction.
14db indicates an output level of an absolute difference value from 14da using 14cf as a reference value. The vertical direction corresponds to the output level and the horizontal direction corresponds to 24 pixels. The position of the pixel in the horizontal direction is matched to the output level of 14da, and the output level of the absolute difference value in the state where the reference value of 14ab is indefinite is represented by a scale that matches the vertical direction. Although 14cf whose average accuracy is higher than 14bf is used as a reference value, in this example, since it is almost converged, the difference in absolute value of difference cannot be seen from the appearance of 14cb.

  14tsd indicates the threshold level selected at 14da. It expresses that it becomes THs1. Since THs1 is selected, the threshold value transition control is performed on the absolute value of the threshold value assigned to THs1 to THs2, which is the center of the threshold determination output group 32a in this example. However, if the threshold is the lower limit, the values from THs0 to THs3 are retained. As described above, in this example, the threshold level is selected so that the absolute difference value of 16 pixels corresponding to two-thirds of all target pixels is included within the threshold.

14 dc indicates a synchronous flaw. The example illustrates that a synchronous flaw exists at a corresponding position in 24 pixels every time. It is the same as 14ac.
14de is an output level obtained by simply averaging 14da. Since it includes 14 dc synchronous flaws, the level is lower than the average that can be expected visually as in 14 ae.
Reference numeral 14df denotes an output obtained by averaging 14da according to the present invention and a reference value level in the next frame. Since the operation is set to leave at least two-thirds of all the target pixels in this example using the absolute difference value of 14 db, a value greatly deviating exceeding the threshold level of 14 tsd is removed and averaged, so that 14 de Unlike the simple average, it shows the average value that should be.

  While the simple average of 14ae, 14be, 14ce, and 14de is always affected by the synchronous noise and the influence of external noise such as radiation, the average of 14af, 14bf, 14cf, and 14df according to the present invention and It can be seen that the reference value converges with increasing average accuracy each time the frame is repeated.

In this embodiment, for the variable thresholds THs0, THs1, THs2, THs3 and the total average of five, there are 33 pixel data adding circuit groups, 34 in-threshold pixel data appearance frequency determination groups, and 36 selection circuits. It is sufficient to have an internal circuit. The total average of TH0 to TH6 or more according to the first embodiment is at least 8 circuits, and the circuit can be simplified.
In addition, as described above, even when the reference value and the input value are shifted due to environmental changes, the absolute difference value is obtained and the operation setting is performed so that two-thirds or more of all target pixels are left. The range can be automatically expanded, and the convergence operation can be continued from a condition similar to 14af. By having an overall average in addition to the four variable threshold values, the convergence operation can be continued from the beginning even if the value is greatly shifted due to environmental changes.

FIG. 6 is a diagram for explaining data storage in the frame memory.
The expected value difference determination of FIG. 2 is also effective for averaging a plurality of frames of images.
FIG. 6 shows an example of data storage in the frame memory. Reference numeral 40 denotes a frame memory. It shows that the vertical and horizontal addresses of the video are stored with correlation.

In a normal case, as shown in FIG. 6A, photosensitive region data 10a based on the photosensitive pixel region 10 of the image sensor is stored in the frame memory 40 for all the pixel regions 12 of the image sensor.
Conventionally, when it is desired to use a light-shielded image region for image correction, as shown in FIG. 6B, for all pixel regions 12 of the image sensor, photosensitive region data 10a based on the photosensitive pixel region 10 of the image sensor in the frame memory 40. In addition, the light shielding area data 11a by the light shielding pixel area 11 of the image sensor is stored. At this time, the vertical and horizontal pixel coordinates are stored so as to be correlated.
In one embodiment of the present invention, in addition to storing data in the frame memory of FIGS. 6A and 6B, in order to perform averaging more efficiently and accurately, FIG. 6C and FIG. Data is stored in the frame memory D).

FIG. 6C shows a format in which the photosensitive area data 10a by the photosensitive pixel area 10 of the image sensor and the light-shielded area average data 11fa obtained by averaging the data from the light-shielded pixel area 11 of the image sensor are stored in the data stored in the frame memory. Show.
At this time, the vertical and horizontal pixel coordinates are stored so as to be correlated. The data obtained by averaging the light-shielded areas is stored in the frame memory, and the data stored in the frame memory and the current image are then captured by completing the averaging of the light-shielded areas of the corresponding image data. When comparing data that is currently being shot, if the light-shielded area of the currently captured data is averaged, there is an advantage in that it is possible to compare without having to perform averaging on the storage data side of the frame memory at the same time. In addition, redundancy of circuit overlap and reduction of operating power can be achieved.

The light shielding area average data 11fa will be described with reference to FIG.
11L is the left part of the light shielding area. 11 is a part extracted for each portion where the light-shielding pixel regions of the image pickup elements are averaged. 11Lf is the averaged data of the left part of the light shielding area. Data averaging is completed in the horizontal direction with respect to the left portion of the light shielding area shown in 11L. Similarly, 11R is the right part of the light shielding region, 11Rf is the averaged data of the right part of the light shielding region, and the data averaging in the horizontal direction is completed for the right part of the light shielding region shown in 11R.

Similarly, as an example in which the light-shielding pixel region 11 of the image sensor is taken out for each part to be averaged, 11 LU is the upper left part of the light shielding region, 11 LfU is the upper part of the averaged data on the left of the light shielding region, and 11 U is the light shielding region. 11RU is the upper right part of the light shielding area, and 11RfU is the upper part of the averaged data on the right side of the light shielding area.
With respect to each of the corresponding areas and data, 11LUf is vertical averaged data at the upper left part of the light shielding area, 11LfUf is vertical averaged data at the upper part of the left averaged data of the light shielding area, and 11Uf. Denotes vertical averaged data at the center of the light shielding area, 11RUf denotes vertical averaged data at the upper right part of the light shielding area, and 11RfUf denotes vertical averaged data at the upper part of the right averaged data of the light shielding area.

Similarly, as an example in which the light-shielding pixel area 11 of the image pickup device is taken out for each part to be averaged, 11LD is the lower left part of the light-shielding area, 11LfD is the lower part of the average data left of the light-shielding area, and 11D is the light-shielding area. , 11RD indicates the lower right portion of the light shielding area, and 11RfD indicates the lower portion of the averaged data on the right side of the light shielding area.
Regarding the corresponding areas and data, 11LDf is vertical averaged data at the lower left part of the light shielding area, 11LfDf is vertical averaged data at the lower part of the left averaged data of the light shielding area, and 11Df as data obtained by averaging the lower part of the light shielding area in the vertical direction. Is the vertical averaged data at the lower central part of the light shielding area, 11RDf is the vertical averaged data at the lower right part of the light shielding area, and 11RfDf is the vertical averaged data at the lower part of the right averaged data of the light shielding area.

In the shaded area average data 11fa in FIG. 6C, vertical averaged data above 11LfUf shaded area left averaged data corresponding to four corners, vertical averaged data above 11RfUf shaded area right averaged data, and shaded light of 11LfDf. Correlation part between the vertical averaged data at the lower part of the area left averaged data, the vertical averaged data at the lower part of the right averaged data of 11RfDf and the averaged data of the left part of the 11Lf lighted area corresponding to the left and right, and the photosensitive area data of 10a 11Rf is a correlation part between the average data of the right part of the light shielding area and the photosensitive area data of 10a, vertical average data of the upper central part of the 11 Uf light shielding area above and below, and vertical average of the lower central part of the light shielding area of 11Df. In one embodiment, the digitized data is stored such that the vertical and horizontal pixel coordinates correlate with the photosensitive area data of 10a. It is.
In the case of the configuration of FIG. 6C, if 11fa shading area average data correlating with the photosensitive area data 10a is stored, there are various formats for the storage portion.

FIG. 6D shows the data stored in the frame memory in addition to the light-sensitive area data 10a by the light-sensitive pixel area 10 of the ten image sensors and the light-shielded area data 11a by the light-shielded pixel area 11 of the image sensor. A format for storing 11fa shading area average data obtained by averaging data from the area 11 is shown. At this time, the vertical and horizontal pixel coordinates are stored so as to be correlated.
The data storage in which the shading area is averaged is performed by completing the averaging of the shading area of the corresponding image data when the data is stored in the frame memory 40, and the data stored in the frame memory thereafter and When comparing the data being shot, averaging the light-shielded area of the data currently being shot has the advantage that it can be compared without having to perform averaging on the frame memory storage data side at the same time. , Redundancy of circuit overlap, and reduction of operating power.

Further, by storing the light-shielding area data 11a that is the original data before averaging, it is possible to compare the stored data with the current data for the data before averaging and the data after averaging, such as the ambient temperature. It can be used to confirm environmental changes.
In FIG. 6D, an example in which the vertical and horizontal pixel coordinates are stored so as to correlate with the photosensitive area data of 10a for all the data regarding the light shielding area data 11a and the light shielding area average data 11fa. It is shown.
In the case of the configuration of FIG. 6D, if the light shielding area data 11a correlated with the photosensitive area data 10a and the light shielding area average data 11fa are stored, there are various formats for the storage part.

Next, with reference to FIGS. 1 and 7, the operation for averaging the same pixels in a plurality of frames according to an embodiment of the present invention will be described.
FIG. 7 is a diagram for explaining the operation of averaging the same pixels in a plurality of frames according to an embodiment of the present invention.
Reference numeral 40 denotes a frame memory. The left frame memory 40 shows before the processing of the present invention, and the right frame memory 40 shows after the processing of the present invention.
When each image is photographed, it is assumed that there is an optical light shielding state, and the output data when the data of the photosensitive pixel area is also shielded is stored in the memory.

  10a1 is data of the photosensitive pixel area of the first frame, and 11a1 is data of the light-shielded pixel area of the first frame. 10a2 is data of the photosensitive pixel area of the second frame, and 11a2 is data of the light-shielded pixel area of the second frame. 10a3 is data of the photosensitive pixel area of the third frame, and 11a3 is data of the light-shielded pixel area of the third frame. 10a4 is data of the photosensitive pixel area of the fourth frame, and 11a4 is data of the light-shielded pixel area of the fourth frame. 10a5 is the data of the photosensitive pixel area of the fifth frame, and 11a5 is the data of the light-shielded pixel area of the fifth frame. 10a6 is the data of the photosensitive pixel area of the sixth frame, and 11a6 is the data of the light-shielded pixel area of the sixth frame. 10a7 is data of the photosensitive pixel area of the seventh frame, and 11a7 is data of the light-shielded pixel area of the seventh frame. 10a8 is data of the photosensitive pixel area of the eighth frame, and 11a8 is data of the light-shielded pixel area of the eighth frame.

  10a0 is simple average data of the photosensitive pixel area, and 11a0 is simple average data of the light-shielded pixel area. This shows that data obtained by simply averaging the outputs from the first frame to the eighth frame for each pixel is stored.

Reference numeral 10a1a denotes processing pixel data in the photosensitive pixel area of the first frame.
Reference numeral 10a2a denotes processing pixel data in the photosensitive pixel area of the second frame.
10a3a is processing pixel data in the photosensitive pixel region of the third frame, and indicates that a steep noise level is generated due to radiation or a blinking pixel defect.
Reference numeral 10a4a denotes processing pixel data in the photosensitive pixel region of the fourth frame.
Reference numeral 10a5a denotes processing pixel data in the photosensitive pixel area of the fifth frame.
Reference numeral 10a6a denotes processing pixel data in the photosensitive pixel area of the sixth frame.
Reference numeral 10a7a denotes processing pixel data in the photosensitive pixel area of the seventh frame.
10a8a is data of the processing pixel in the photosensitive pixel area of the eighth frame.
10a0a is processing pixel data in the simple average of the photosensitive pixel area, and the averaging includes a steep noise level due to radiation or a blinking pixel defect from the processing pixel data in the photosensitive pixel area of the third frame of 10a3a. Shows things.

Reference numeral 300 denotes an expected value difference determination averaging circuit which is a central configuration of the present invention. FIG. 3 is a part where a block configuration diagram of an averaging circuit using expected value difference determination used in the embodiment of the present invention in FIG. 2 is used.
In averaging the horizontal light shielding area, pixels on the same horizontal line are sequentially input to the pixel data input 30a. When the present invention is applied to the same pixel averaging for a plurality of frames, pixel data of the same coordinates in a plurality of frames stored in the pixel data input 30a is used as an input. The data of the above processing pixels are sequentially input.
In averaging the horizontal light shielding area, the previous average value of the same line was used as the reference value input 3a. When the present invention is applied to the same pixel averaging for a plurality of frames, a value obtained by averaging pixel data of the same coordinates of a plurality of frames is used for the reference value input 3a. Data of the processed pixel in the simple average of the photosensitive pixel area 10a0a described above is input. In this example, a simple average is used, but feedback using one embodiment of the present invention may be repeated to improve accuracy.

10a0aa is the level of the processed pixel data in the simple average. The output level is expressed in the vertical direction.
10a1aa is the level of the processed pixel data in the first frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a2aa is the level of the processed pixel data in the second frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a3aa is the level of the processed pixel data in the third frame. The vertical output level is expressed on the same scale as 10a0aa, indicating that a steep noise level is generated due to radiation or blinking pixel defects.
10a4aa is the level of the processed pixel data in the fourth frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a5aa is the level of the processed pixel data in the fifth frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a6aa is the level of the processed pixel data in the sixth frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a7aa is the level of the processed pixel data in the seventh frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.
10a8aa is the level of the processed pixel data in the eighth frame. The output level in the vertical direction is expressed on the same scale as 10a0aa.

10bb is the level of the absolute difference value in the processing pixel. Along with the processing pixel data of each frame in the horizontal direction, the coordinates in the vertical direction are represented by shifted positions. A location corresponding to 10a3aa which is affected by radiation and blinking pixel defects protrudes.
10t is a threshold value selected in the processing pixel. In this embodiment, since a threshold value including 3/4 or more is selected from the pixels of all the processing target frames, the data corresponding to the protruding frame corresponding to the above 10a3aa is excluded from the averaging.
10b0aa is the level of the processed pixel data using the present invention. Data affected by radiation and blinking pixel defects are excluded, and the average value should be present. The processed pixel data level of 10b0aa using the present invention is output to the extreme value exclusion average output of 37a, and is stored in the frame memory as the coordinate data of the processed pixel.
10b0 is the average data of the photosensitive pixel region using one embodiment of the present invention, and 11b0 is the average data of the light-shielding pixel region using one embodiment of the present invention. The influence of radiation and blinking pixel defects is excluded from the simple average data of the light-sensitive pixel area 10a0 and the simple average data of the light-shielded pixel area 11a0, and the same pixel averaging of a plurality of frames can be performed with better accuracy.

Next, the re-averaging operation of the correction value output unit according to an embodiment of the present invention will be described with reference to FIGS.
FIG. 8 is a diagram for explaining the re-averaging operation of the correction value output unit according to the embodiment of the present invention.
The expected value difference determination of FIG. 2 is also effective for the case of re-averaging the data area stored in the frame memory.
In this embodiment, it is shown that the averaging in units of pixels is completed using the above-described expected value difference determination averaging circuit 300, and 10b0 is the average of the photosensitive pixel area using the embodiment of the present invention. 11b0 is the average data of the light-shielded pixel region using one embodiment of the present invention.

  Reference numeral 14 'denotes light-shielded pixel data stored in a memory on an arbitrary line. Instead of pixels from the image sensor, they are sequentially input to the pixel data input 30a. 14aa 'is the stored data level of the light-shielded pixel, the horizontal axis is aligned with 14', and the vertical axis indicates the output level. In this example, since the pixel unit is averaged, there is no influence of radiation or blinking defects. 14ac 'is a synchronous wound. In addition to this synchronization flaw, pixel-specific noise remains in the pixel-unit averaging, and re-averaging from the stored data is performed in order to accurately obtain the average value of the light-shielded pixel region.

  Since the reference value input of 3a is undefined for the first time, the first output of 37a is a simple average. 14af 'is the level of the simple average output. An average that includes data that should be excluded, such as synchronous flaws. For this reason, at least once, the data is fed back to the reference value input of 3a, and data in which the extreme value exclusion average output of 37a works effectively is generated. At this time, the pixel data input of 30a may be repeated using data on the same line. 14bf ″ is a level obtained by re-averaging the light-shielded area by using one embodiment of the present invention. The re-averaged level of 14bf ″ is stored in the 40 frame memory in correlation with the same line.

  11b0f is light-shielded area re-averaged data. 14bf ″ is stored and generated for each line. By using a method of generating an average value of the light shielding area by re-averaging, the central configuration of one embodiment of the present invention is used. The required average value can be obtained with high accuracy without mounting a plurality of expected value difference determination averaging circuits 300.

Next, with reference to FIG. 1 and FIG. 9, the operation for removing the environmental change tracking fixed noise of the correction value output unit according to the embodiment of the present invention will be described.
FIG. 9 is a diagram for explaining the operation of removing the environmental change tracking fixed noise of the correction value output unit according to the embodiment of the present invention.
Reference numeral 10c denotes data of the photosensitive pixel area being shot, and reference numeral 11c denotes data of the light-shielded pixel area being shot. In the expected value difference determination averaging circuit 300, which is the central configuration of the present invention, the data 11c averages the data of the light-shielded pixel area being photographed in the horizontal direction for each line.

Using the above-described embodiment, the frame memory 40 has already stored photosensitive area data at the time of shading of 10b0, shading area data of 11b0, and shading area re-averaged data of 11b0f. These data are used for a correction signal for removing fixed noise.
Reference numeral 51 denotes a read signal for re-averaged data. For the line corresponding to the extreme value exclusion average output of 37a, the light-blocking area re-averaged data of 11b0f is read.
Reference numeral 52 denotes a 5-unit coefficient multiplication circuit. A value obtained by multiplying the read signal of 51 re-averaged data is generated using the upper and lower two coefficients with respect to the current center coefficient k. For example, 51 re-averaged data read signals are multiplied by using k ± 0.1 as upper and lower coefficients with respect to k and k ± 0.2 as further upper and lower coefficients.

53 is a diagram of line correlation. In order to show that the operating line of the extreme value exclusion average output of 37a, the line of the read signal of 51 re-averaged data, and the line output of the 52 series of coefficient multiplier circuits are the same. It is described in.
Reference numeral 54 denotes a 5-unit difference absolute value generation circuit. An extreme value exclusion average output of 37a and 52 output the absolute difference value of each of the five outputs by five coefficient multiplication circuits.
Reference numeral 55 denotes a minimum value determination circuit. A flag is set to the coefficient having the smallest difference absolute value from the 54-fold difference absolute value generation circuit.
Reference numeral 56 denotes a 5-shift register. A flag output from the 55 minimum value determination circuit is stored in the shift register for a certain amount for each coefficient.

  Reference numeral 57 denotes a coefficient correction circuit. Monitor the amount of flags in each column of 56 five-shift registers. If necessary, the range of the shift register used in the camera adjustment value setting output of 4b may be set, and the correction coefficient switching may be managed by synchronizing with vertical synchronization or a plurality of frames with the camera timing output of 4c. When the frequency of occurrence other than the column of k becomes the maximum, the value of the coefficient k is replaced with the column having the maximum, and is output to the coefficient output of 57a. When the appearance frequencies of the two columns are substantially equal, the value of the coefficient k is replaced with the average value of the two and output to the coefficient output of 57a. When the coefficient output of 57a is replaced with other than k, a shift register clear signal of 57b is output, and 56 of the five-shift register is cleared. The newly set coefficient k is output to the coefficient output of 57a until a certain number of flags are accumulated in the 56 5-shift register.

58 is a coefficient multiplication circuit. Data for correction is generated by multiplying the photosensitive area data at the time of shading of 10b0 by the coefficient k. The coefficient for minimizing the difference between the light shielding area average during photographing and the light shielding area average for correction is calculated by the 55 minimum value determination circuit, and is changed to the coefficient by which the difference is minimized by the coefficient correction circuit 57. The coefficient k that minimizes the difference in environment between the stored data and the data currently being photographed is selected. By multiplying the coefficient k by the photosensitive area data at the time of shading of 10b0, a value equivalent to the fixed noise included in the photosensitive area data being photographed is generated.
Reference numeral 59 denotes a fixed noise subtraction circuit. The correction data is subtracted from the data of the photosensitive pixel area during the photographing 10c. By subtracting a value equivalent to the fixed noise included in the photosensitive area data being shot, it is possible to remove the fixed noise following the environmental change.

According to the above-described embodiment of the present invention, in order to reduce noise, when averaging image signals obtained from pixels in the optical black region of the image sensor, data is arranged while excluding extreme values that should be excluded from the average. By performing an operation of shifting to an appropriate threshold value without performing replacement, an extreme value exclusion average circuit that follows the environment can be realized with small-scale logic.
Further, according to the above-described embodiment of the present invention, an expected value, that is, a method for obtaining a new correction value by taking an average from near-difference values using a difference between a correction value up to the present and new correction source data is used for environmental fluctuations such as temperature. Can follow.

  Although the present invention has been described in detail above, the present invention is not limited to the single-plate type imaging device described herein, and can be widely applied to other multiple-plate type imaging devices. Needless to say.

  The present invention is not limited to the single-plate type imaging device described here, and can be widely applied to a multi-plate type imaging device other than the above.

  1: SENS unit, 2: subtraction unit, 3: correction value output unit, 4: subsequent stage unit, 5: control unit, 30: clip unit, 31: difference detection unit, 32: threshold value determination unit, 33: addition unit, 34 : In-threshold pixel data appearance number determination unit, 35: setting unit, 36: selection unit, 37: normalization unit, 38: correction value holding memory unit, 39: correction value holding register unit, 40: frame memory, 300: expectation Value difference judgment averaging circuit.

Claims (10)

In an imaging device including an imaging unit, a correction value output unit, and a subtraction unit,
The correction value output unit determines a signal level for each pixel output from the light-shielded pixel region of the imaging unit based on a plurality of thresholds, selects a pixel signal having a predetermined signal level, and averages the selected pixel signals. And output the averaged signal as a correction value signal,
The subtracting unit subtracts a signal output from the photosensitive pixel area of the imaging unit by the correction value signal. The imaging device according to claim 1,
A correction value holding register unit that holds the correction value signal as a comparison reference value;
A control unit that controls the start and end of averaging;
A setting unit that holds a plurality of threshold values;
A difference detection unit that calculates a difference absolute value of the comparison reference value and a new correction value signal;
A threshold detection determination unit that determines whether or not the magnitude of the difference absolute value exceeds a plurality of thresholds;
An adder for adding the output of the threshold determination unit and the pixel signal;
An in-threshold pixel data appearance frequency determination unit that increments a counter when it is determined that the threshold determination is within the threshold for a plurality of thresholds;
A selection unit that selects a minimum value that exceeds an output value of the pixel data appearance frequency determination unit within the threshold value, and that selects the output of the addition unit using the same threshold value; and
A normalization unit that normalizes an addition value based on a threshold value obtained by dividing and using the output of the selected addition unit, the output of the selected in-threshold pixel data appearance number determination unit, and an average value;
A correction value holding memory unit capable of holding the output of the normalization unit for each area addressed by the control unit, and outputting an addressed value;
The correction value holding memory unit holds the output of the correction value holding memory unit at a timing designated by the control unit, and has an averaging circuit configuration that is an averaged output to the comparison reference value and the subordinate circuit. An imaging device that is characterized. In the imaging device according to claim 1 and claim 2,
The image pickup apparatus, wherein the selection unit that selects the output of the addition unit and the output of the in-threshold pixel data appearance frequency determination unit is switched according to a threshold value. In the imaging device according to claim 1,
The correction value output unit determines a signal level for each pixel of a horizontal line output from the light-shielded pixel region of the imaging unit with a plurality of thresholds, selects a pixel signal having a predetermined signal level, and selects the selected pixel An imaging apparatus characterized by averaging signals and outputting the averaged signal as a correction value signal. In the imaging device according to claim 1,
The correction value output unit further includes a correction value holding memory unit,
The image pickup apparatus, wherein the correction value holding memory unit stores the correction value signal and a signal level for each pixel output from a light-shielded pixel region of the image pickup unit. In the imaging device according to claim 1,
Have multiple frame memories,
An image pickup apparatus comprising a frame averaging circuit for the pixel signals using the plurality of frame memories. In the imaging device according to claim 1, claim 2, and claim 6,
After the correction value signal is output, the correction value output unit again determines a signal level for each pixel output from the light-shielded pixel region of the imaging unit using a plurality of threshold values, and outputs a pixel signal having a predetermined signal level. An imaging apparatus comprising: selecting, averaging the selected pixel signals, and outputting the averaged signal as a correction value signal. The imaging apparatus according to claim 7,
An image pickup apparatus, wherein a correction value signal is generated by multiplying information stored in the frame memory by a predetermined coefficient. In the imaging device according to claim 1,
Furthermore, it has a clip part,
The image pickup apparatus, wherein the clip unit suppresses a signal output from a light-shielded pixel region of the image pickup unit with a predetermined value. The imaging device according to claim 2,
Information on the selected threshold value is selected from the selection unit that selects the addition unit that uses the same threshold value and selects the output value of the in-threshold pixel data appearance number determination unit indicating the minimum value that exceeds a predetermined value. System to output,
In the threshold determination unit that detects whether or not the magnitude of the difference absolute value exceeds a plurality of thresholds, a threshold value to be output to the threshold determination result is greater than the number of threshold determination results output by the plurality of thresholds An imaging apparatus characterized by having a system for inputting information.

Images