JP2004171156A - Image processing apparatus and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus and an image processing method capable of reducing only stripe noises caused by variations among channels that have occurred since they cannot be corrected by gain adjustment, while avoiding deterioration of necessary information such as the edges of a subject, and deterioration of picture quality. <P>SOLUTION: In the image processing apparatus which subjects input image data to signal processing, a fixed pattern detecting means 1 uses pixel data to be processed and data on ambient pixels to detect a predetermined fixed pattern. Fixed pattern correcting means 2, 3 carry out a correction process (i.e., gain adjustment) only for those pixels from which the fixed pattern was detected by the fixed pattern detecting means 1. Only the stripe noises (fixed patterns) caused by variations among, e.g., the channels, can be reduced without deteriorating necessary information such as the edges of the subject nor deteriorating picture quality. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing apparatus and an image processing method capable of reducing fringe noise that cannot be completely corrected by gain adjustment between channels when signal processing is performed on a captured image in which a subject image is read as a plurality of channel outputs.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, there is an image sensor having a plurality of channel outputs, such as an all-pixel readout type CCD (Charge Coupled Device, abbreviation for Charge Coupled Device). For example, a two-channel readout type CCD has two horizontal transfer registers in order to lower the driving frequency, and transfers the odd-numbered line signal charges and the even-numbered line signal charges simultaneously in different horizontal transfer registers. The signal charges for two lines are read out within one horizontal transfer period.
[0003]
In a CCD having a plurality of channel outputs, when a gain variation between channels occurs, the variation becomes a level difference of an output signal between channels, and appears as fringe noise in an image. Therefore, in order to correct the gain variation between channels, as described in Japanese Patent Application Laid-Open No. H10-276976, the output of each channel is sampled and the average value is compared, for example. The output gain is adjusted to match the gain between channels. However, when shading having different aspects occurs between a plurality of channels, variations between channels cannot be completely corrected, and as a result, stripe noise may remain in an image.
[0004]
Further, as another prior art, Japanese Patent Application Laid-Open No. 62-296668 discloses a method of correcting a target pixel signal based on pixel signals positioned before, after, right and left of the target pixel signal in an image signal. There is disclosed an image correction circuit that controls so as not to perform correction when a density difference between a signal and pixel signals located in front, rear, left and right is within a predetermined range, and controls to output a correction value at other times. .
[0005]
Japanese Patent Application Laid-Open No. 4-239886 discloses a pixel value group that outputs a pixel value of a pixel of interest and a group of adjacent pixels two-dimensionally adjacent to the pixel of interest from an image signal composed of a plurality of scanning lines. Extraction means; noise determination means for determining whether the pixel of interest includes noise by detecting whether the pixel value of the pixel of interest is different from a group of neighboring pixel values by a predetermined threshold or more; There is disclosed an image noise device having a pixel value selecting means for selecting and outputting any one of the pixel values of a pixel of interest and a group of adjacent pixels based on a determination result by the means.
[0006]
Further, Japanese Patent Application Laid-Open No. 9-218940 discloses a screen processing apparatus for interpolating a defective image in an image, the image input means for inputting an image in which a plurality of effective image areas including a plurality of pixels are scattered, Area specifying means for specifying an effective image area in which the defective image exists, and interpolation means for interpolating the defective image with an image obtained from a peripheral area including at least one effective image area around the effective image area in which the defective image exists There is disclosed an image processing apparatus having the following. Thus, a good output image can be obtained by interpolating the defective image in the image in which the defective image exists over a plurality of pixels.
[0007]
Further, Japanese Patent Application Laid-Open No. 11-308625 discloses an imaging apparatus including a camera signal processing circuit for forming a video signal from a digitally converted input signal, wherein the camera signal processing circuit detects an edge of the input signal. The image processing apparatus includes an edge detection unit for detecting, a luminance signal processing unit for extracting luminance information of the input signal, and a color difference signal processing unit for extracting color information of the input signal. The signal amount D1, D2, D3 of each of the pixels and the predetermined set value A are | D1-D2 |> A and | D1-D3 |> A or | D1-D3 |> A and | D2-D3 An image pickup apparatus is disclosed in which a color difference signal processing unit suppresses a color difference signal when the condition that |> A is satisfied by the edge detection unit.
[0008]
However, in each of the above-mentioned JP-A-62-296668, JP-A-4-239886, JP-A-9-218940 and JP-A-11-308625, an image pickup device having a plurality of channel outputs is disclosed. There is no description of means for solving the problem that, when signal processing is performed on an image of a plurality of channels from a plurality of channels, the gain cannot be completely corrected by the gain adjustment between channels, and stripe noise remains in the image.
[0009]
[Patent Document 1]
Japanese Patent Application Laid-Open No. Hei 10-276976 (page 1-2, FIG. 1)
[0010]
[Patent Document 2]
JP-A-62-296668 (page 1-2, FIG. 1)
[0011]
[Patent Document 3]
JP-A-4-239886 (pages 1-3, FIG. 1)
[0012]
[Patent Document 4]
JP-A-9-218940 (pages 1-3, FIG. 1)
[0013]
[Patent Document 5]
JP-A-11-308625 (pages 1-3, FIG. 1)
[0014]
[Problems to be solved by the invention]
As a method of correcting fringe noise described in Japanese Patent Application Laid-Open No. Hei 10-276976, there is a method of performing correction by performing a smoothing process on the entire image. However, necessary information such as an edge of a subject is also blurred. As a result, there is a problem that the image quality is deteriorated.
[0015]
The present invention has been made in view of the above problems, without deteriorating necessary information such as an edge of a subject, and removing fringe noise caused by variations between channels that cannot be completely corrected by gain adjustment, without deteriorating image quality. It is another object of the present invention to provide an image processing apparatus and an image processing method capable of reducing only stripe noise.
[0016]
[Means for Solving the Problems]
The image processing apparatus according to the first aspect of the present invention is an image processing apparatus for performing signal processing on input image data, wherein a predetermined fixed pattern is formed by using pixel data to be processed and data of peripheral pixels. It is characterized by comprising fixed pattern detecting means for detecting, and fixed pattern correcting means for performing correction processing only on pixels for which a fixed pattern has been detected by the fixed pattern detecting means.
[0017]
According to this configuration, the correction process (that is, the gain adjustment) is performed only on the pixel in which the predetermined fixed pattern corresponding to the stripe noise is detected using the data of the pixel to be processed and the data of the peripheral pixels. Therefore, the necessary information such as the edge of the subject is not degraded, and the fringe noise caused by the inter-channel variation that cannot be corrected by the gain adjustment of the output between the channels is reduced without deteriorating the image quality. Only reduction is possible.
[0018]
According to a second aspect of the present invention, in the image processing apparatus for performing signal processing on input image data, a predetermined fixed value is used by using pixel data to be processed and data of peripheral pixels. Fixed pattern detection means for detecting a pattern, and fixed pattern correction means for performing correction processing only on pixels for which a fixed pattern has been detected by the fixed pattern detection means; and It is characterized in that it is provided after the means.
[0019]
According to this configuration, since the variation between channels cannot be completely corrected, the generated fringe noise can be reduced.
[0020]
Further, according to the image processing apparatus of the present invention, in the image processing apparatus for performing signal processing on input image data, the image processing apparatus performs predetermined fixed processing using pixel data to be processed and data of peripheral pixels. A fixed pattern detection unit for detecting a pattern, and a fixed pattern correction unit for performing a correction process only on pixels for which a fixed pattern has been detected by the fixed pattern detection unit are provided at a stage preceding the enlargement processing unit. And
[0021]
According to this configuration, the noise generated for each dot or each line may be corrected, so that the configuration is simplified.
[0022]
An image processing method according to an eleventh aspect of the present invention is an image processing method for performing signal processing on input image data, wherein a predetermined fixed pattern is formed by using pixel data to be processed and data of peripheral pixels. The method includes a step of detecting and a step of performing a correction process on a pixel for which a predetermined fixed pattern has been detected in the step.
[0023]
According to this method, correction processing (that is, gain adjustment) is performed only on pixels for which a predetermined fixed pattern corresponding to fringe noise has been detected, using data of a pixel to be processed and data of surrounding pixels. Therefore, the necessary information such as the edge of the subject is not degraded, and the fringe noise caused by the inter-channel variation that cannot be corrected by the gain adjustment of the output between the channels is reduced without deteriorating the image quality. Only reduction is possible.
[0024]
According to a twelfth aspect of the present invention, in the image processing method for performing signal processing on input image data, a predetermined fixed value is used by using pixel data to be processed and data of peripheral pixels. A step of detecting a pattern and a step of performing a correction process on a pixel for which a predetermined fixed pattern is detected in the step, and a step of correcting a variation between channels of the image sensor, which is provided at a subsequent stage. I do.
[0025]
According to this method, since the inter-channel variation cannot be corrected, the generated stripe noise can be reduced.
[0026]
Further, according to the image processing method of the present invention, in the image processing method for performing signal processing on input image data, a predetermined fixed value is used by using pixel data to be processed and data of peripheral pixels. The step of detecting a pattern and the step of performing a correction process on a pixel for which a predetermined fixed pattern has been detected in the step are provided before the step of performing an enlargement process.
[0027]
According to this method, noise generated for each dot or each line may be corrected, so that the configuration is simplified.
[0028]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
1 to 12 relate to a first embodiment of the present invention, FIG. 1 is a block diagram showing a basic configuration of an image processing apparatus, and FIG. 2 is a diagram showing stripes generated without being able to completely correct variations between channels. FIG. 3 is a diagram showing a noise level, FIG. 3 is a diagram showing a signal level between A and B in FIG. 2, FIG. 4 is a flowchart showing a processing flow of the fixed pattern detection unit in FIG. 1, and FIG. FIGS. 6 and 7 show examples of conditions that are not detected in processing step S3 of FIG. 4. FIGS. 6 and 7 show examples of conditions detected in step S3. FIG. 8 shows detection of fringe noise in processing step S10 of FIG. FIG. 9 is an explanatory diagram for explaining an example of a condition for performing the processing, and FIG. 9 is an explanatory diagram for explaining an example of a condition in which no stripe noise is detected in the processing step S10 in FIG. 4. FIG. FIG. 11 is a flowchart showing the processing flow of FIG. Flowchart illustrating a process flow of correcting portion, FIG. 12 is a diagram showing a processing result by the apparatus or method of the present embodiment.
[0029]
In the present embodiment, an image processing apparatus that performs signal processing for effectively removing vertical stripe noise is shown. The vertical stripe noise is, for example, stripe noise generated without completely correcting variations between channels. Here, consider the case where vertical stripe noise occurs every other dot. When the captured image is enlarged, vertical stripe noise may occur every other dot or more. In this case, the detection condition may be adjusted to the stripe noise, and the basic concept remains the same. . In this embodiment, vertical stripe noise at every other dot will be described to simplify the description.
[0030]
Before describing the configuration of the image processing apparatus according to the present invention in FIG. 1, vertical stripe noise will be described with reference to FIGS.
FIG. 2 shows a state of vertical stripe noise (fixed pattern) of an image projected on a monitor screen (octagonal shape in the figure). When a uniform subject such as a white chart is imaged, the signal level of the portion where the vertical stripe noise occurs (between AB in FIG. 2) is high and low for every other dot as shown in FIG. Pixels are arranged alternately.
[0031]
If the gain variation of the signal level is corrected using a smoothing filter or the like as in the related art, a portion where no stripe noise is generated is also smoothed. As a result, the image quality deteriorates.
[0032]
Also, in the image portion where the stripe noise occurs, if there is an edge that is not the stripe noise, for example, when the subject is a resolution chart or the like, the edge that rapidly changes with a fine bite is blurred and deteriorated. There was a thing.
[0033]
In view of the above, the present invention provides an image processing apparatus capable of detecting only stripe noise and correcting the same.
[0034]
That is, the image processing apparatus according to the present invention detects a predetermined fixed pattern using pixel data to be processed and data of peripheral pixels in the image processing apparatus that performs signal processing on input image data. And a fixed pattern correction unit that performs a correction process only on pixels for which a fixed pattern has been detected by the fixed pattern detection unit.
[0035]
In the image processing apparatus of the present invention shown in FIG. 1, for example, two signals of an odd-numbered line and an even-numbered line obtained from an image sensor of a two-channel readout type are subjected to gain adjustment between channels, and at that time, variations between channels are corrected. In FIG. 1, two signals of odd-numbered lines and even-numbered lines are gain-adjusted between channels, and the signals of two channels after the gain adjustment are combined in FIG. A configuration in which fringe noise is corrected for an image signal is shown.
[0036]
The image processing apparatus shown in FIG. 1 includes a fixed pattern detection unit 1 that detects a fixed pattern for each pixel using pixel data to be processed from input image data and data of surrounding pixels, A correction amount calculation unit 2 that calculates a correction amount for each pixel from data, and a fringe noise correction unit that performs correction based on the correction amount calculated by the correction amount calculation unit 2 only for pixels for which a fixed pattern is detected. 3 is provided. The fixed pattern detection unit 1 constitutes a fixed pattern detection unit, and the correction amount calculation unit 2 and the stripe noise correction unit 3 constitute a fixed pattern correction unit.
[0037]
Next, the operation of the fixed pattern detection unit 1 of FIG. 1 will be described with reference to the flowchart of FIG.
As shown in S1 to S5 in FIG. 4, the input image is processed from data A that is not delayed, data B that is delayed by one dot (step S1), and data C that is delayed by two dots (step S2). By comparing the pixel data (corresponding to B) with the pixel data on both sides, it is determined whether the pixel data is convex or concave (step S3). In step S3, as shown in FIG. 5, the pixel at the position B has a higher signal level than the neighboring pixels, and if the signal level difference between the neighboring pixels is within 10%, it is determined that the pixel is convex. On the other hand, although not shown, if the signal levels are both low and the signal level difference between the adjacent pixels is within 10%, it is determined to be concave and 1 is substituted for the variable "Flag" (step S4). As shown in FIG. 6, when the difference between the adjacent signal levels is 10% or more, or when there is a stepwise change as shown in FIG. 7, 0 is substituted for the variable “Flag” (step S5). .
[0038]
In the present embodiment, for simplicity of description, the signal level difference between adjacent pixels caused by vertical stripe noise is assumed to be within 10% of the signal level. Since it varies depending on the characteristics and the method of correcting the variation between channels, it may be appropriately changed and determined in consideration of these factors, for example, within 2, 4, 8, 16% or 2 to 8% of the signal level. %, 4 to 16%, etc., are not limited to “within 10% of the signal level” of the conditions described in the present embodiment.
[0039]
Next, as shown in S6 to S12 in FIG. 4, the flag that is not delayed, the flag that is delayed by one dot (step S6), the flag that is delayed by two dots (step S7), and the delay of three dots (step S8) ), The flag E delayed by 4 dots (step S9), and it is determined whether the convex or concave shape is continuous in adjacent pixels (step S10). In step S10, in FIG. 8, when the Flag of the pixel at the position C is 1, the flags of the pixels A, B, and C are all 1, or the flags of the pixels C, 2, and E are all 1, or If the flags of the pixels b, c, and d are both 1, it is determined that a fixed pattern of vertical stripes has been detected, and 1 is substituted for the variable "Shima" (step S11). Otherwise, it is determined that a fixed pattern of vertical stripes has not been detected, and 0 is substituted for the variable “Sima” (step S12).
[0040]
In the case of FIG. 8, since the flags of the pixels C, D, and E are all 1 and the flags of the pixels B, C, and D are all 1, 1 is substituted for the variable "Shima". For example, in the case of FIG. 9, even if the variable “Flag” of any pixel is 1, detection is not performed as vertical stripe noise.
[0041]
According to the processing method of FIG. 4 described above, it is possible to detect only vertical stripe noise generated every other dot. Note that the detection condition of the vertical stripe noise is not limited to the above-described method. For example, when the number of memories is increased and four convex or concave types are continuously formed, or when the number of memories is reduced, the convex or concave type is used. It may be determined that only one exists, and that it is vertical stripe noise.
[0042]
Next, the operation of the correction amount calculation unit 2 of FIG. 1 will be described with reference to the flowchart of FIG.
The image data input to the correction amount calculation unit 2 includes B-A and B-C from data A that is not delayed, data B that is delayed by one dot (step S21), and data C that is delayed by two dots (step S22). Is calculated (steps S23 and S24). Then, in step S25, the larger value of BA and BC is selected (step S25) and substituted for the variable "H" (steps S26 and S27). Thereafter, the variable “H” is halved and output as a correction value (step S28).
[0043]
By setting the correction value to の, which is the larger of the signal level difference between adjacent pixels, for example, when a uniform subject such as a white chart is imaged, the stripe noise correction unit 3 in the latter stage of FIG. The corrected signal level becomes the same as that of the adjacent pixel as shown in FIG. 12, and the vertical stripe noise is corrected.
[0044]
Next, the operation of the fringe noise correction unit 3 of FIG. 1 will be described with reference to the flowchart of FIG.
It is determined whether or not the variable “Shima” of the image input from the fixed pattern detection unit 1 is 1 (step S31). For the pixel where the variable “Shima” = 1, the input image data and the correction amount Using the correction value H / 2, when the image is determined to be convex in the fixed pattern processing step S3 in FIG. 4, (image data) − (correction value H / 2), and when it is determined to be concave, , (Image data) + (correction value H / 2) are output (step S32).
[0045]
For the pixel for which the variable “Shima” = 0 in step S31, the input image data is output as it is (step S33).
[0046]
As described above, according to the image processing apparatus of the present embodiment, as shown in FIG. 12, a portion having a large difference in signal level from an adjacent pixel, such as a portion having an edge of a subject, or a vertical stripe No correction is performed for the part where the signal level difference with the adjacent pixel is stair-like, such as the part where no noise is generated, and the signal level of only the part where the vertical stripe noise is generated is Since correction is performed so that the difference from the signal level is small, it is possible to efficiently reduce only vertical stripe noise without deteriorating the edge information etc. of the subject compared to the method using the conventional smoothing filter. It becomes.
[0047]
In the present embodiment, an image processing apparatus applicable to vertical stripe noise has been described. However, in the case of horizontal stripe noise, the same applies if the one-dot delay processing in FIGS. 4 and 10 is changed to one-line delay processing. Accordingly, the image processing apparatus can be applied to horizontal stripe noise. Also, in the present embodiment, the noise of oblique stripes such as beat noise is detected by information in only the horizontal direction or the vertical direction, so that it can be applied to an image processing apparatus of exactly the same configuration. It is.
[0048]
FIG. 13 is a block diagram illustrating a configuration of an image processing apparatus according to the second embodiment of the present invention.
As shown in FIG. 13, the image processing apparatus according to the present embodiment includes a one-line memory 11a for delaying input image data by one line, and a one-line memory 11b for further delaying the output of the one-line memory 11a by one line. , A subtractor 12a for subtracting the output of the one-line memory 11a from the input image data, a subtractor 12b for subtracting the outputs of the one-line memory 11a and the one-line memory 11b, and each output of the subtractors 12a and 12b and the one-line memory. A condition detector 13 for detecting a pixel whose signal level is higher or lower than both adjacent pixels from the output of 11a, a one-line memory 14a for delaying the output of the condition detector 13 by one line, and an output of the one-line memory 14a. One line memory 14b for further delaying one line, and one line for further delaying the output of one line memory 14b by one line A 1-line memory 14d for further delaying the output of the in-memory 14c and the 1-line memory 14c by one line, a horizontal stripe noise detection unit 15 for detecting horizontal stripe noise from the output of the condition detection unit 13 and each output of the one-line memories 14a to 14d. A correction value calculation unit 16 for calculating a correction value from the outputs of the subtraction units 12a and 12b, an output of the horizontal stripe noise detection unit 15 and an output of the correction value calculation unit 16, and a horizontal stripe noise from the outputs of the one-line memories 11a and 14b. And a horizontal stripe noise correction unit 17 for correction.
[0049]
Next, the operation of FIG. 13 will be described.
The image processing apparatus according to the present embodiment is an image processing apparatus that operates in the same manner as the first embodiment and that can correct horizontal stripe noise.
[0050]
First, in the input image data, the difference between the signal levels of the upper and lower pixels, that is, the signal levels of the input image data and the output of the one-line memory 11b is calculated with the output of the one-line memory 11a as the center pixel. The subtractor 12a calculates (output of the one-line memory 11a)-(input image data), and the subtractor 12b calculates (output of the one-line memory 11a)-(output of the one-line memory 11b).
[0051]
The calculated signal level difference is input to the condition detection unit 13. The condition detector 13 determines whether the signal level of the central pixel is higher or lower than the signal levels of the upper and lower pixels based on the difference between the signal levels and the signal level of the central pixel. When the outputs of the subtractors 12a and 12b are both positive values and within 10% of the signal level of the central pixel, it is determined to be convex and +1 is output. When the outputs of the subtractors 12a and 12b are both negative values and within 10% of the signal level of the central pixel, it is determined to be concave and -1 is output.
[0052]
The horizontal stripe noise detector 15 detects horizontal stripe noise from the output of the condition detector 13 and the outputs of the one-line memories 14a to 14d. The detection method is basically the same as that of the first embodiment. In S6 to S9 of FIG. 4, horizontal line noise detection is enabled by performing one line delay instead of one dot delay. It is. "Condition detecting unit 13, one-line memory 14a, one-line memory 14b", "one-line memory 14a, one-line memory 14b, one-line memory 14c", "one-line memory 14b," including the output of one-line memory 14b. When the three patterns of the output of the one-line memory 14c and the one-line memory 14d apply to “+1, −1, +1” or “−1, +1, −1”, it is determined that the pixel has horizontal stripe noise. Note that, similarly to the first embodiment, the conditions for detecting horizontal stripe noise are not limited to the above conditions.
[0053]
In the same manner as in the first embodiment, the correction value calculation unit 16 selects the larger value from the outputs of the subtraction units 12a and 12b, halves the value, and outputs it as a correction value.
[0054]
When the output of the one-line memory 14b is +1 for a pixel determined to have horizontal stripe noise by the horizontal-stripe noise detecting unit 15, the horizontal-stripe noise correction unit 17 determines whether or not the original image data of the output of the one-line memory 11a The output of the correction value calculator 16 is subtracted. When the output of the one-line memory 14b is -1, the output of the correction value calculator 16 is added to the original image data output from the one-line memory 11a.
[0055]
As described above, similarly to the first embodiment, it is possible to efficiently reduce only horizontal stripe noise without deteriorating the edge information of the subject or the like. In the present embodiment, a memory for adjusting the timing of the arithmetic processing is not particularly shown, but using the memory as needed does not depart from the spirit of the present invention.
[0056]
14 and 15 relate to the third embodiment of the present invention. FIG. 14 is a block diagram showing the configuration of the imaging device 20. FIG. 15 is a vertical stripe noise included in the signal processing unit 25 of the imaging device 20. FIG. 3 is a block diagram illustrating a configuration of a processing unit 250.
[0057]
As shown in FIG. 14, an imaging device 20 according to the present embodiment includes a CCD 21, a process circuit 22, an A / D converter 23, a gain adjustment circuit 24 for correcting a variation between channels of the CCD 21, a vertical stripe noise processing. It comprises a signal processing circuit 25 having a section 250, an enlargement circuit 26, and a D / A converter 27. The CCD 21 is a two-channel readout type CCD, and the processing system from the CCD 21 to the gain adjustment circuit 24 has a two-system circuit configuration having two signal lines of odd-numbered dots and even-numbered dots. For simplicity, each of the CCD 21, the process circuit 22, the A / D converter 23, and the gain adjustment circuit 24 is represented by only one block.
[0058]
As shown in FIG. 15, the signal processing circuit 25 is provided with a vertical stripe noise processing unit 250 for correcting vertical stripe noise generated without being able to completely correct variations between channels in the gain adjustment circuit 24 in the preceding stage. Note that the enlargement circuit 26 performs an electronic enlargement process on the digital video signal divided into one system after the vertical stripe noise processing at an enlargement ratio according to the size of an image displayed on a monitor (not shown).
[0059]
Next, the configuration of the vertical stripe noise processing unit 250 provided in the signal processing circuit 25 of the imaging device 20 will be described. As shown in FIG. 15, the vertical stripe noise processing unit 250 includes a one-dot delay memory 251a for delaying the input image data by one dot, and a one-dot delay memory 251b for further delaying the output of the one-dot delay memory 251a by one dot. A subtractor 252a for subtracting the output of the one-dot delay memory 251a from the input image data, a subtractor 252b for subtracting the outputs of the one-dot delay memory 251a and the one-dot delay memory 251b, and one-dot delay for the subtractors 252a and 252b. A condition detection unit 253 for detecting a pixel having a signal level higher or lower than both adjacent pixels from the output of the memory 251a, a one-dot delay memory 254a for delaying the output of the condition detection unit 253 by one dot, and a one-dot delay memory 254a One-dot delay memory 254 that further delays the output by one dot A one-dot delay memory 254c for further delaying the output of the one-dot delay memory 254b by one dot; a one-dot delay memory 254d for further delaying the output of the one-dot delay memory 254c by one dot; A vertical stripe noise detection unit 255 that detects vertical stripe noise from the outputs of the dot delay memories 254a to 254d; a vertical stripe noise correction unit 256 that corrects vertical stripe noise from the output of the vertical stripe noise detection unit 255 and the outputs of the one-dot delay memories 251a and 254b; It is comprised including.
[0060]
Next, the operation of the imaging device 20 of FIG. 14 and the vertical stripe noise processing unit 250 of FIG. 15 will be described.
In the imaging device 20, a video signal imaged by a CCD 21, a process circuit 22 provided with a CCD driver, a preamplifier, and the like is converted from an analog signal to a digital signal by an A / D converter 23, and the gain is adjusted by a gain adjustment circuit 24. The variation between the channels of the CCD 21 is corrected. The signal processing circuit 25 is provided with a vertical stripe noise processing unit 250, and corrects vertical stripe noise caused by shading of the same channel that could not be corrected by the gain adjustment circuit 24. Thereafter, the image is enlarged by the enlargement circuit 26, and the video signal is converted into an analog signal by the D / A converter 27 and output, or directly output as a digital signal.
[0061]
In the vertical stripe noise processing unit 250, the input image data is input to the left and right pixels with the output of the one-dot delay memory 251a as the center pixel, that is, the input image data and the one-dot delay memory The difference between each signal level and the output of 251b is calculated. The subtractor 252a calculates (output of the one-dot delay memory 251a)-(input image data), and the subtractor 252b calculates (output of the one-dot delay memory 251a)-(output of the one-dot delay memory 251b).
[0062]
The calculated signal level difference is input to the condition detection unit 253. The condition detector 253 determines whether the signal level of the central pixel is higher or lower than the signal levels of the left and right pixels based on the difference between the signal levels and the signal level of the central pixel. When the outputs of the subtractors 252a and 252b are both positive values and within 10% of the signal level of the central pixel, it is determined to be convex and +1 is output. When the outputs of the subtractors 252a and 252b are both negative values and within 10% of the signal level of the central pixel, it is determined to be concave and -1 is output.
[0063]
The vertical stripe noise detector 255 detects vertical stripe noise from the output of the condition detector 253 and the outputs of the one-dot delay memories 254a to 254d. "Condition detector 253, one-dot delay memory 254a, one-dot delay memory 254b", "one-dot delay memory 254a, one-dot delay memory 254b, one-dot delay memory 254c" including the output of one-dot delay memory 254b, Vertical stripes when the three patterns output from the “one-dot delay memory 254b, one-dot delay memory 254c, and one-dot delay memory 254d” apply to “+1, −1, +1” or “−1, +1, −1” It is determined as a pixel having noise. Note that, similarly to the first and second embodiments, the conditions for detecting vertical stripe noise are not limited to the above-described conditions.
[0064]
When the output of the one-dot delay memory 254b is +1 for a pixel determined to have vertical stripe noise by the vertical-stripe noise detection unit 255, the vertical-stripe noise correction unit 256 outputs the output of the one-dot delay memory 254a (the original image). 5% of the signal level of (data) is subtracted. When the output of the one-dot delay memory 254b is -1, 5% of the signal level of the output (original image data) of the one-dot delay memory 254a is added.
[0065]
As described above, similarly to the first and second embodiments, it is possible to efficiently reduce only the vertical stripe noise without deteriorating the edge information or the like of the subject. In this embodiment, the memory for adjusting the timing of the arithmetic processing is not particularly shown, but the use of the memory as necessary deviates from the gist of the present invention as in the second embodiment. is not.
[0066]
In the present invention, as a method of correcting fringe noise, a method mainly based on gain adjustment has been described, but a smoothing process using a spatial filter such as 3 × 3, 1 × 3, 5 × 5 may be performed. .
[0067]
【The invention's effect】
As described above, according to the first aspect of the present invention, a fixed pattern such as a vertical stripe noise, a horizontal stripe noise, and a diagonal noise is detected, and the gain adjustment is performed only on the portion where the fixed pattern is detected. By performing correction processing such as image and smoothing processing, it is possible to effectively reduce noise without deteriorating important information such as edge information of a subject.
[0068]
According to the second aspect of the present invention, it is not possible to completely correct the variation between channels, so that it is possible to reduce the generated fringe noise.
[0069]
According to the third aspect of the present invention, since the noise generated for each dot or each line may be corrected, the configuration is simplified.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a basic configuration of an image processing apparatus according to a first embodiment of the present invention.
FIG. 2 is a diagram showing a state of fringe noise generated in the image displayed on the monitor screen without being able to completely correct the variation between channels.
FIG. 3 is a diagram showing signal levels between AB in FIG. 2;
FIG. 4 is a flowchart illustrating a processing flow of a fixed pattern detection unit in FIG. 1;
FIG. 5 is a diagram showing an example of conditions detected in processing step S3 in FIG. 4;
FIG. 6 is a diagram showing an example of a condition not detected in processing step S3 of FIG. 4;
FIG. 7 is a view showing an example of a condition not detected in processing step S3 of FIG. 4;
FIG. 8 is an explanatory diagram for explaining an example of a condition for detecting fringe noise in processing step S10 of FIG. 4;
FIG. 9 is an explanatory diagram for explaining an example of a condition in which fringe noise is not detected in processing step S10 of FIG. 4;
FIG. 10 is a flowchart illustrating a processing flow of a correction amount calculation unit in FIG. 1;
FIG. 11 is a flowchart showing a processing flow of a fringe noise correction unit in FIG. 1;
FIG. 12 is a diagram showing a processing result by the apparatus or the method according to the embodiment.
FIG. 13 is a block diagram illustrating a configuration of an image processing apparatus according to a second embodiment of the present invention.
FIG. 14 is a block diagram showing a configuration of an imaging device according to a third embodiment of the present invention.
FIG. 15 is a block diagram illustrating a configuration of a vertical stripe noise processing unit included in the signal processing circuit of the imaging device in FIG. 14;
[Explanation of symbols]
1: fixed pattern detection unit (fixed pattern detection means)
2. Correction amount calculation unit (fixed pattern correction means)
3. Stripe noise corrector (fixed pattern corrector)

Claims (20)

In an image processing device that performs signal processing on input image data,
Using fixed pixel pattern detection means for detecting a predetermined fixed pattern, using pixel data to be processed and data of peripheral pixels,
Only for pixels for which a fixed pattern has been detected by the fixed pattern detection means, a fixed pattern correction means for performing a correction process,
An image processing apparatus comprising: In an image processing device that performs signal processing on input image data,
Using fixed pixel pattern detection means for detecting a predetermined fixed pattern, using pixel data to be processed and data of peripheral pixels,
Only for pixels for which a fixed pattern has been detected by the fixed pattern detection means, a fixed pattern correction means for performing a correction process,
An image processing apparatus provided at a subsequent stage of a unit for correcting a variation between channels of an image sensor. In an image processing device that performs signal processing on input image data,
Using fixed pixel pattern detection means for detecting a predetermined fixed pattern, using pixel data to be processed and data of peripheral pixels,
Only for pixels for which a fixed pattern has been detected by the fixed pattern detection means, a fixed pattern correction means for performing a correction process,
An image processing apparatus characterized in that is provided before the enlargement processing means. The image processing apparatus according to claim 1, wherein the fixed pattern detection unit detects a fixed pattern based on signal level information of a target pixel and a pattern adjacent to the target pixel. . 5. The image processing apparatus according to claim 4, wherein the fixed pattern detection unit detects a fixed pattern based on a difference in signal level between a target pixel and a pixel adjacent to the target pixel. 6. The image processing apparatus according to claim 5, wherein the fixed pattern detection unit determines that the fixed pattern is a fixed pattern when the signal level of the target pixel is higher or lower than the signal level of a pixel adjacent to the target pixel. . The fixed pattern detection means determines that the signal level of the pixel of interest is higher or lower than the signal level of a pixel adjacent to the pixel of interest, and temporarily determines that the pixel is a fixed pattern, and temporarily fixes adjacent pixels in the same manner. 7. The image processing apparatus according to claim 6, wherein when it is determined that the pattern is a pattern, the image processing apparatus determines that the pattern is a fixed pattern. The image processing apparatus according to claim 1, wherein the fixed pattern correction unit performs the correction by adjusting a signal level of a pixel of interest. 9. The image processing apparatus according to claim 8, wherein the fixed pattern correction unit adjusts a signal level of the target pixel based on information of the target pixel and a pixel adjacent to the target pixel. 10. The image processing apparatus according to claim 9, wherein the fixed pattern correction unit adjusts a signal level of the target pixel based on a difference between signal levels of the target pixel and pixels adjacent to the target pixel. In an image processing method for performing signal processing on input image data,
A step of detecting a predetermined fixed pattern using pixel data to be processed and data of peripheral pixels,
Performing a correction process on the pixel for which the predetermined fixed pattern is detected in the step,
An image processing method comprising: In an image processing method for performing signal processing on input image data,
Detecting a predetermined fixed pattern using the pixel data to be processed and the data of the peripheral pixels,
Performing a correction process on the pixel for which the predetermined fixed pattern is detected in the step,
An image processing method provided after the step of correcting the variation between channels of the image sensor. In an image processing method for performing signal processing on input image data,
Detecting a predetermined fixed pattern using the pixel data to be processed and the data of the peripheral pixels,
Performing a correction process on the pixel for which the predetermined fixed pattern is detected in the step,
An image processing method provided before a step of performing an enlargement process. 14. The method according to claim 11, wherein the step of detecting the fixed pattern is a step of detecting the fixed pattern based on signal level information of a target pixel and a pixel adjacent to the target pixel. Image processing method. 15. The image processing method according to claim 14, wherein the step of detecting the fixed pattern is a step of detecting the fixed pattern based on a signal level difference between a target pixel and a pixel adjacent to the target pixel. 16. The image according to claim 15, wherein the step of detecting the fixed pattern determines that the fixed pattern is a fixed pattern when the signal level of the pixel of interest is higher or lower than the signal level of a pixel adjacent to the pixel of interest. Processing method. The step of detecting the fixed pattern, if the signal level of the pixel of interest is higher or lower than the signal level of the pixel adjacent to the pixel of interest, temporarily determines that the pattern is a fixed pattern, and the adjacent pixels in the same way 17. The image processing method according to claim 16, wherein if it is determined that the pattern is a fixed pattern, the pattern is determined to be a fixed pattern. 18. The image processing method according to claim 11, wherein in the step of performing the correction processing, the correction is performed by adjusting a signal level of a pixel of interest. 19. The image processing method according to claim 18, wherein the step of performing the correction process adjusts a signal level of the target pixel based on information of the target pixel and a pixel adjacent to the target pixel. 20. The image processing method according to claim 19, wherein the step of performing the correction process adjusts a signal level of the target pixel based on a difference between signal levels of the target pixel and pixels adjacent to the target pixel.

Images