JPH1051693A - Defective picture element correction processing unit for image pickup element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To correct a signal from a defective picture element by classifying defective picture elements based on a defective picture element detection result, eliminating the defective picture elements, and interpolating the defective picture element image signals with signals of picture elements adjacent to the defective picture element. SOLUTION: A frame addition processing unit 23 of a defective picture element detection section 13 adds image signals of optional several frames received via a switch 11 to eliminate a noise component, the result is given to an LPF processing unit 25, in which a steep change resulting from a defective picture element is eliminated to obtain only a low frequency component. A difference between the low frequency component and an output of the adder 23 is obtained by a subtractor 27 to extract the defective picture element. A normal defective label denoting '1' for an extracted defective picture element and '0' for a normal picture element is generated by a separator 29, a classification device 31 classifies the picture elements of the label '1' furthermore and correction data labelled based on the classified kinds are stored in a ROM, the data are given to a correction coefficient generating section 17, in which a weight correction coefficient for interpolation by a defective picture element correction processing section 21 is generated and it is fed to the processing section 21. Then the same weight is given to all picture elements used for the interpolation. That is, the defective picture element is interpolated by using a mean value of adjacent normal picture elements.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a defective pixel correction processing device for an image pickup device for correcting a defective pixel of an image pickup device of a TV camera mounted on an X-ray diagnostic apparatus and an endoscope apparatus.

[0002]

2. Description of the Related Art In recent years, X-ray diagnostic apparatuses, endoscope apparatuses, and the like are equipped with a TV camera using an image sensor, and an image of a subject is captured by the TV camera. In the case of an image pickup device used in this TV camera, for example, a CCD, the smaller the number of defective pixels, the higher the price thereof. Therefore, it is necessary to use an element including some defective pixels. For this reason, a defective pixel correction processing device for correcting a defective pixel is used.

FIG. 18 shows an X-ray diagnostic apparatus equipped with such a TV camera and a defective pixel correction processing device. X-ray diagnostic apparatus 100 equipped with a TV camera as shown in FIG.
Is converted by an X-ray beam source 101 that emits X-rays to the subject P, an image intensifier 103 that converts the X-rays that have passed through the subject P into an optical image, and an image intensifier 103 Camera 105 that captures the obtained optical image and obtains an image signal (pixel value) for each pixel
An A / D converter 107 for converting an image signal obtained by the TV camera 105 into a digital signal; a defective pixel correction processing device 109 for correcting an image signal of a defective pixel among the obtained image signals; A D / A converter 111 that converts the image signal corrected by the correction processing device 109 and the image signal of the normal pixel into an analog signal, and a monitor 113 that displays the image signal converted into an analog signal by the D / A converter 111. And an image processing device 115 that performs image processing such as threshold processing using an image signal whose defective pixel value has been corrected by the defective pixel correction processing device 109.

Here, the CC used for the TV camera 105
Since the characteristics of the defective pixel generated in D are various, the defective pixel correction processing device 109 performs correction by a correction processing method according to the characteristic. A representative example of the defective pixel correction processing method will be described below.

[0005] White defective pixels (white point defects, white line defects, white point defect group, etc.) that appear to appear slightly whitish than the surrounding pixels on the image are defects caused by dark current generated in the CCD. This occurs because the image signal is slightly larger than the surrounding pixels by the level of the dark current generated in the defective pixel. This dark current is characterized by being constant without depending on the amount of incident light. For this reason, an attempt has been made to measure in advance a dark current at a predetermined incident light amount, specifically when imaging is performed in a state where light is not incident on a camera, and to subtract the value in real time from the captured X-ray image signal. And good results have been obtained.

When the dark current is extremely large and the pixel is saturated with white (when the charge of the pixel is saturated),
Since the information of the pixel is completely lost, the effect of the defective pixel correction by the above-described subtraction processing cannot be expected, and a pixel replacement method or an interpolation processing method using a peripheral pixel of the defective pixel is performed.

[0007] Further, black defective pixels (black point defects, a group of black point defects) which look extremely black compared to the surrounding pixels are CC
D: This is a defect that occurs when dust or the like adheres to the chip in the manufacturing process and the incident light is blocked. Since this defective pixel is a pixel that does not contain information on the pixel as in the case of the saturated white defective pixel, a pixel replacement method or an interpolation processing method using a peripheral pixel of the defective pixel has been attempted. .

As a pixel replacement method, an adjacent pixel replacement method for replacing a normal pixel adjacent to a defective pixel with a defective pixel and a median filter processing method for replacing the pixel with an intermediate value of a neighboring image signal have been attempted.

[0009] As interpolation processing methods for interpolating defective pixels from neighboring pixels, bi-linear interpolation and higher-order interpolation processing (spline interpolation, Lagrange interpolation, SINC function)
And so on.

[0010]

However, the above-described correction processing method is effective for an area having a smooth pixel distribution (light contrast), but is effective for an area where an image signal greatly changes (contrast is low). (Extremely strong area), the correction trace may be clearly recognized. For example, in an upper gastrointestinal tract examination, correction marks are clearly visible at the boundary between the image of barium and the background image placed in the subject and at the stomach folds and the like. The reason why the correction cannot be performed well can be explained as follows.

In the adjacent pixel replacement method, as shown in FIG. 19A, when the contrast sharply changes between the first pixel row and the second pixel row, a defective pixel is detected in the second pixel row. When this is interpolated with the pixels in the bright region (pixels in the first column), the image signal of the defective pixel is expected to be 10, but as shown in FIG. 1
00, the correction mark can be clearly confirmed.

In the median process, the region of interest (R
When there are a plurality of defective pixels in OI), the probability that the selected median value is a defective pixel increases, and a phenomenon that the defective pixel is replaced with the defective pixel may occur. For example, in the example shown in FIG. 20, when the pixel of the image signal 90 is a defective pixel, if median processing is performed on the image signal of that row, an image signal of an intermediate value among the image signals 100, 90, and 50, that is, The image signal having the second largest value is the median value. In this case, the median value becomes the image signal value 90 of the defective pixel, and the image signal of the defective pixel becomes the image signal of the pixel as it is, which is the same as when no processing is performed.

In the bilinear interpolation method, for defective pixels in an area corresponding to a subject extending in the horizontal or vertical direction, defective pixels are interpolated by pixels containing no information on the subject. Improper interpolation. For example, FIG.
In the example shown in FIG. 21A, when the subject is an elongated object extending in the vertical direction, and when the pixel of the image signal value 5 in the area corresponding to the subject is a defective pixel, FIG.
Since the interpolation is performed using the image signal of the pixel indicated by the white circle in FIG. 21B, the image signal of the defective pixel is interpolated to a value 100 irrelevant to the information of the subject as shown in FIG.

Regarding other methods, the effect is not remarkably exhibited as compared with the above-mentioned interpolation processing method, although the algorithm is complicated and the scale of the processing circuit is increased. SIN in this
In the high-dimensional interpolation processing method using the C function, the interpolation processing is performed in consideration of pixels distant from the defective pixel. Therefore, when there is a completely different image signal at a position distant from the defective pixel, the information is added to the interpolation result. , Resulting in improper interpolation.
For example, in the example shown in FIG. 22, in the case where the pixel indicated by the mark x is a defective pixel, interpolation is performed using the image signal of the pixel indicated by the white circle. However, the pixel region used for interpolation is large, and there is a possibility that two or more subjects are included in the region. Therefore, the image signal of the defective pixel is interpolated with the image signal of a subject different from the subject indicated by the cross mark. there's a possibility that.

SUMMARY OF THE INVENTION The present invention has been made in consideration of the above problems, and has as its object to provide a defective pixel correction processing device for an image sensor which can appropriately correct an image signal of a defective pixel.

[0016]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention relates to a defective pixel correction processing apparatus for an image sensor which corrects an image signal of a defective pixel of an image sensor having a plurality of light receiving means arranged as pixels. And a defective pixel detecting means for detecting a defective pixel, a defective pixel classifying means for classifying a defective pixel based on a detection result of the defective pixel detecting means, and a classification result of the defective pixel classifying means, The gist of the present invention is to include a defective pixel correction processing unit that corrects a defective pixel by interpolating an image signal of the defective pixel using an image signal of a pixel adjacent to the defective pixel except for the defective pixel.

According to the first aspect of the present invention, the defective pixel is classified by the defective pixel classifying means based on the detection result by the defective pixel detecting means for detecting the defective pixel. In accordance with the result, the image signal of the defective pixel is interpolated using the image signal of the pixel adjacent to the defective pixel except for the defective pixel. Thereby, the image signal of the defective pixel can be appropriately corrected.

The defect classifying means creates a normal / defective label image in which the defective pixel is at a high level and the normal pixel is at a low level. Classify by pattern, perform labeling for each of the predetermined number of pixel groups corresponding to this type,
The defective pixel correction processing unit interpolates an image signal of a defective pixel using an image signal of a pixel adjacent to the defective pixel except for the defective pixel for each of the predetermined number of pixel groups in accordance with the labeling by the defect classification unit. It is desirable.

Further, the defect classifying means creates a normal / defective label image in which the defective pixel is at a high level and the normal pixel is at a low level, and based on the normal / defect label image, a defect for each of a predetermined number of pixel groups. Classification by pixel pattern is performed, and labeling is performed for each of the predetermined number of pixel groups in accordance with the type, and the defective pixel correction processing unit calculates a correction coefficient corresponding to the labeling by the defect classification unit to the predetermined number of pixels. A correction coefficient generation unit that is generated for each pixel group, and is associated with labeling by the defect classification unit, and for each of the predetermined number of pixel groups, an image signal of a pixel adjacent to the defective pixel except for the defective pixel; It is desirable to interpolate the image signal of the defective pixel using the correction coefficient generated by the generator.

[0020]

Embodiments of the present invention will be described below with reference to the drawings. The defect pixel correction processing apparatus of the image sensor of the present embodiment is provided in the X-ray diagnostic apparatus 100 as shown in FIG. Image to TV camera 105
The image signal of the defective pixel is interpolated among the image signals obtained by the imaging.

In particular, in this embodiment, the TV camera 105
Detects defective pixels of the CCD used in the process, creates a normal / defective label image that is divided into defective pixels and normal pixels based on the detection results, and further forms defective pixel patterns based on the normal / defective label images. Create a labeling image classified according to. Then, based on the labeling image, the defective pixel is excluded, and the image signal of the defective pixel is interpolated using the image signal (pixel value) of the pixel adjacent to the defective pixel.

FIG. 1 is a block diagram showing an embodiment of an apparatus for correcting a defective pixel of an image sensor according to the present invention. As shown in FIG. 1, a defective pixel correction processing device 10 for an image sensor according to the present embodiment includes a switch 11, a defective pixel detection unit 13 as a defective pixel detection unit and a defective pixel classification unit, a correction data ROM 15, It has a coefficient generation unit 17, a buffer memory 19, and a defective pixel correction processing unit 21 as defective pixel correction processing means.

The switch 11 sends the supply destination of the image signal converted into a digital signal by the A / D converter 107 shown in FIG. 18 to the defective pixel detection unit 13 (terminal B) or the buffer memory 19 side (terminal A). Switch. This switching is performed by control means (not shown).

As shown in FIG. 2, the defective pixel detector 13 includes a frame addition processor 23, a low-pass filter processor (LPF processor) 25, a subtractor 27, and a normal / defective pixel separator 29. , A defective pixel classifier 31;
Defective pixels are detected and classified.

The frame addition processor 23 removes noise components from the image signal by adding image signals of an arbitrary number of frames supplied from the A / D converter 107 via the switch 11.

The LPF processor 25 includes a frame adder 23
Accordingly, a spatial high-frequency component in an image signal from which a noise component has been removed, that is, a steep change caused by a defective pixel is removed by low-pass filter processing to spatially leave only a low-frequency component.

The subtracter 27 calculates the difference between the image signal from which the noise component has been removed by the frame adder 23 and the image signal from which only the low-frequency component has been spatially reduced by the LPF processor 25 and extracts only defective pixels. I do.

The normal / defective pixel separator 29 is a subtractor 2
A normal / defective label image is created in which the defective pixel extracted by 7 is “1” and the other normal pixels are “0”. For example, if the subtractor 27 extracts a pixel indicated by the mark x shown in FIG. 3A as a defective pixel, the normal / defective pixel separator 29 sets the defective pixel indicated by the mark x to “1” and sets the other pixels to “1”. "0"
A normal / defective label image as shown in FIG.

The defective pixel classifier 31 further classifies defective pixels based on the normal / defective label image created by the normal / defective pixel separator 29, and performs further labeling according to the type. .

As a method of this labeling, a predetermined R
OI, for example, a pixel adjacent to the defective pixel is checked for each 3 × 3 pixel group as shown in FIG. 4, and if the number of “1” is 0 as shown in FIG. As shown in FIG. 4B, when the number of “1” is one, two consecutive point defects are detected.
If the number of “1” is two as shown in FIG. 4C, a line defect or a point defect group, and if the number of “1” is three or more as shown in FIG. They are classified and labeled according to these types.

As a specific method, the name of each pixel adjacent to the defective pixel of the ROI is set as shown in FIG.
I11, I12, I13, I21, I23, I31, I
32, 133, the sum SUM of adjacent pixels (= I11 +
I12 + I13 + I21 + I23 + I31 + I32 + 133)
Isolated point defect when SUM = 0, 2 when SUM = 1
A continuous point defect is classified as a line defect or a point defect group when SUM = 2, and a point defect group when SUM ≧ 3.

Further, in the case of a two-point defect (SUM = 2), as shown in FIGS.
type). According to this classification method, the total of adjacent pixels excluding a specific pixel is checked and classified as follows.

[0033]

[Number 1] SUM = I12 + I13 + I21 + I23 + I31 + I32 + 133 = 0 ... type1 SUM = I11 + I13 + I21 + I23 + I31 + I32 + 133 = 0 ... type2 SUM = I11 + I12 + I21 + I23 + I31 + I32 + 133 = 0 ... type3 SUM = I11 + I12 + I13 + I21 + I31 + I32 + 133 = 0 ... type4 SUM = I11 + I12 + I13 + I21 + I23 + I31 + I32 = 0 ... type5 SUM = I11 + I12 + I13 + I21 + I23 + I31 + 133 = 0 ... type6 SUM = I11 + I12 + I13 + I21 + I23 + I32 + 133 = 0... Type7 SUM = I11 + I12 + I13 + I23 + I31 + I32 + 133 = 0.
(J) is a line defect (a transfer defect that occurs when charges are transferred in the horizontal or vertical direction in the CCD), and the other defect patterns are more likely to occur than other defect patterns. It is wasteful to provide a circuit for correcting this. In other words, it can be considered that a CCD having a point defect group cannot be used as a CCD for the TV camera 105 because it is a low quality product.

In the method of classifying line defects, as in the case of a two-point defect, the sum of adjacent pixels excluding a specific pixel is checked and classified as follows.

[0035]

SUM = I11 + I21 + I31 + I13 + I23 + 133 = 0 .... type9 SUM = I11 + I12 + I13 + I31 + I32 + 133 = 0 .... type10 In the above-described classification method, a normal / defective label image is used to classify pixels based on the sum of labels of pixels adjacent to defective pixels. Are checked in the order of I11, I12, I13,... And the number of "1" is counted.

Then, labeling is performed on the defective pixels classified as described above, and a labeling image of the defective pixels is created by combining them. For example, FIG.
As shown in FIG. 7A, the number of isolated point defects is 2, the type of two continuous point defects is 3 as shown in FIG. 7B, and the type 2 of two continuous point defects is 4 as shown in FIG. As shown in FIG. 7D, type 3 of the two consecutive point defects is 5, and FIG.
As shown in FIG. 7, the type 4 of the two continuous point defects is 6, and FIG.
As shown in FIG. 7 (f), the type 5 of the two continuous point defects is 7, and as shown in FIG.
As shown in FIG. 7H, the type 7 of the two consecutive point defects is 9, and as shown in FIG.
Is 10, and as shown in FIG.
Reference numeral 9 denotes 11, and as shown in FIG.
e10 is 12, and the point defect group is 13 as shown in FIG.
And

The correction data ROM 15 stores the labeling image created by the defective pixel classifier 31. Further, the correction data ROM 15 stores normal X-ray images or X-rays.
When the defective pixel is interpolated in the case of fluoroscopy, the stored labeling image is output to the correction coefficient generator 17.

The correction coefficient generator 17 is provided with a correction data ROM
The weight at the time of interpolating the defective pixel by the defective pixel correction processing unit 21 is generated as a correction coefficient in accordance with the labeling image output from the reference numeral 15. Here, as the simplest method, the same weight is assigned to all pixels used for interpolation, that is, defective pixels are interpolated by the average value of adjacent normal pixels.

For example, in the case of an isolated point defect as shown in FIG.
In the case of two consecutive point defects as shown in FIGS. 8B to 8I, the weight is 0.142 for all adjacent normal pixels.
857, and in the case of a line defect as shown in FIGS. 8 (j) and 8 (k), the weight is 0.16 for all adjacent normal pixels.
6667.

Further, the weight is inversely proportional to the distance (inverse) because the former is smaller than the distance to the pixel located at the four corners, and the distance to the next pixel is smaller and closer to the characteristic of the defective pixel. Ratio). In this case, for example, as shown in FIG.
The upper and lower, left and right adjacent pixels are set to 0.103553, and the diagonally adjacent pixels are set to weight 0.146445.
As shown in (d), (f), and (h), t of two consecutive point defects
In the case of types 1, 3, 5, and 7, adjacent pixels in the upper, lower, left, and right directions have a weight of 0.163363, and adjacent pixels in the oblique direction have a weight of 0.1.
15515 and FIGS. 9 (c), (e), (g), (i)
As shown in FIG. 9, in the case of types 2, 4, 6, and 8 of two consecutive point defects, the weight of the adjacent pixels in the upper, lower, left, and right directions is 0.171573, and the weight of the adjacent pixels in the oblique direction is 0.12132.
As shown in (j) and (k), in the case of a line defect, the upper, lower, left, and right adjacent pixels have a weight of 0.207107, and the diagonally adjacent pixels have a weight of 0.14647. 8 and 9, the pixels in the thick frames represent defective pixels, and the pixels marked with * are the pixels to be interpolated. The interpolation coefficient of the defective pixel not to be interpolated is set to 0.

The buffer memory 19 is used for the purpose of adjusting timing in the vertical direction, has a capacity capable of storing two or more lines of image signals, and has a structure capable of simultaneously outputting a plurality of lines of image signals. For example, when simultaneously outputting three lines, as shown in FIG. 10, four line memories LM1 to LM
Multiplexer M for reading M4 and 3 lines simultaneously
P.

Here, an operation example in the case where three lines are simultaneously output will be described with reference to the timing chart of FIG.
Line image signals of line numbers 1, 2, and 3 are written in the order of line memories LM1, LM2, and LM3 (timing charts W1, W2, and W3). Next, at the same time when the line image signal of the line number 4 is written to the line memory LM4 (timing chart W4), the line memory LM
1, LM2, and LM3 simultaneously output line image signals of line numbers 1, 2, and 3 (timing chart R
1, R2, R3). Next, at the same time when the line image signal of the line number 5 is written to the line memory LM1 (timing chart W5), the line memories LM2 and LM
3 and LM4 simultaneously output line image signals of line numbers 2, 3 and 4 (timing charts R2 and R3).
R4). Thereafter, the same operation is performed.

The defective pixel correction processing unit 21 multiplies the line image signal output from the buffer memory 19 by the correction coefficient generated by the correction coefficient generation unit 17 as shown in FIG. 33-12, ..., 33-nm,
Multipliers 33-11, 33-12,..., 33-nm;
3-11, 33-12, ..., 33-nm and multipliers 33-11,
33-12,..., 33-nm are added by a predetermined number of pixels (the number of normal pixels in the ROI),
And a horizontal / vertical addition processing unit 35 that outputs the image signal of the defective pixel.

For example, when performing the interpolation process of the defective pixel using the pixels in the 3 × 3 ROI, the defective pixel correction processing unit 2
1 includes nine multipliers 33-11 and 33 as shown in FIG.
-12,33-13,33-21,33-22,33-23,33-
31, 33-32 and 33-33 and three latches 37-1 and 37-2 for adjusting the horizontal and vertical timings.
37-3 and adders 39-1, 41-1, 3 of two lines (a total of six) for adding three data for each line
9-2, 41-2, 39-3, 41-3, an adder 43 for adding the added data of the first and second lines, and an adder 43 for adding the added data of the first and second lines. Latch 4 for latching addition data of the third line during addition
5 and an adder 47 that further adds the added data on the third line to the data added by the adder 43.
The latches 37-1, 37-2, 37-3 and the adders 37-1,
39-1, 37-2, 39-2, 37-3, 39-3 and adder 4
3, a latch 45, and an adder 47 constitute a horizontal / vertical addition processing unit 35.

Here, the defective pixel correction processing unit 2 in the case of performing the defective pixel interpolation processing using the pixels in the 3 × 3 ROI
An operation example 1 will be described with reference to a timing chart shown in FIG. At this time, as shown in FIG.
In the following description, the image signal of the first line (line) is P11, the image signal of the second line is P12,. Further, the defective pixel correction processing unit 21 operates in response to a clock transmitted from a control unit (not shown).

First, at the clock C1, the first image signals P11, P21 and P31 of the first ROI are converted to the multipliers 33 of each line.
-11, 33-21, and 33-31. Next, at the clock C2, the second image signals P12, P22,
P32 is a multiplier 33-12, 33-22, 33-32 for each line.
Is input to At this time, the multipliers 33-11, 33-21, 3
The first image signal of the first ROI is input to 3-31.

Next, in the lock C3, the third image signal P13, P23, P33 of the first ROI is output to the multiplier 33-1 of each line.
3, 33-23 and 33-33. At this time, the first image signal of the first ROI is input to the multipliers 33-11, 33-21, 33-31, and the multipliers 33-12, 33-22,
33-32 receives the second image signal of the first ROI.

Next, at the clock C4, the first correction coefficient is input from the correction coefficient generation unit 17 for each line, and the first ROI of the next ROI is supplied to the multipliers 33-11, 33-21 and 33-31.
The second image signals P14, P24, P34 are input. At this time, the first R is supplied to the multipliers 33-12, 33-22, and 33-32.
The second image signal of the OI is input, and the multiplier 33-13,
The third image signal of the first ROI is input to 33-23 and 33-33.

Next, at the clock C5, the second correction coefficient is input from the correction coefficient generation unit 17 for each line, and is also supplied to the multipliers 33-12, 33-22, and 33-32 for the next ROI.
The second image signals P15, P25, P35 are input. At this time, the next RO is input to the multipliers 33-11, 33-21, and 33-31.
The first image signal of I is input, and the multipliers 33-13, 3
The third image signal of the first ROI is input to 3-23 and 33-33.

Next, at the clock C6, the third correction coefficient is input from the correction coefficient generation unit 17 for each line, and is supplied to the multipliers 33-13, 33-23, and 33-33 for the next ROI.
The second image signals P16, P26, P36 are input. At this time, the next RO is input to the multipliers 33-11, 33-21, and 33-31.
The first image signal of I is input, and the multipliers 33-12, 33
3-22 and 33-32 receive the second image signal of the next ROI.

Then, at the clock C7, the result D1 of the first horizontal / vertical addition process of the ROI is output. Thereafter, the same operation is performed.

The defective pixel correction processing device 10 for the image pickup device
Multiplier 33 because the circuit scale of
-11, 33-12, 33-13, 33-21, 33-22, 33
-23, 33-31, 33-32, and 33-33 cannot be used, as shown in FIG.
3-12, 33-13, 33-21, 33-22, 33-23, 3
A ROM for inputting an image signal and a correction coefficient instead of 3-31, 33-32, and 33-33 and outputting a multiplication result corresponding to these values, that is, a lookup table (LUT) 49
-1, 49-2,..., 49-n (n = number of lines) may be provided.

Next, the operation of the defective pixel correction processing device 10 for an image sensor according to the present embodiment will be described. First, TV camera 1
In order to detect a defective pixel of the CCD used in 05, the switch 11 is switched to the terminal B side, and then an image signal is obtained by taking an image with the TV camera 105 in the absence of the subject. The image signal obtained by the TV camera 105 is converted into a digital signal by the A / D converter 107,
The signal is supplied to the defective pixel detection unit 13 via the switch 11.

When the image signal converted into the digital signal is supplied, in the defective pixel detecting section 13, an image signal of an arbitrary number of frames is added by a frame addition processor 23 to remove noise components. Is spatially reduced to only low frequency components. And a frame adder 2
The difference between the image signal from which the noise component has been removed by 3 and the image signal spatially reduced to only the low-frequency component by the LPF processor 25 is obtained by the subtractor 27, and the defective pixel is extracted.

Then, a normal / defective label image in which the extracted defective pixel is “1” and other normal pixels are “0” is created by the defective pixel separator 29, and this normal / defective label image is further processed. Based on the ROI (for example, 3 × 3)
), And labeling corresponding to the type is performed by the defective pixel detector 31. The labeling image obtained by this labeling is temporarily stored in the correction data ROM 15. Thereafter, the defective pixel correction processing unit 21 performs a correction process of the defective pixel based on the labeling image stored in the correction data ROM 15. Note that this defective pixel detection operation is first performed when the defective pixel correction processing device 10 for the image sensor is manufactured.
This is performed when the correction data ROM 15 is rewritten for the purpose of interpolating a newly generated defective pixel due to deterioration of the CD due to aging.

In this state, the switch 11 is switched to the terminal A by control means (not shown) to perform normal X-ray photography or X-ray fluoroscopy. When normal X-ray photography or X-ray fluoroscopy is performed, the labeling image stored in the correction data ROM 15 is supplied to the correction coefficient generation unit 17, and the correction coefficient corresponding to the labeling image is corrected by the correction coefficient generation unit 1.
7 to the defective pixel correction processing unit 21.

The defective pixel correction processor 21 multiplies the correction coefficient output from the correction coefficient generator 17 by the corresponding image signal, and adds the result to the number of pixels in the ROI to obtain a defective pixel value. . This operation is performed for all ROIs having defective pixels corresponding to the labeling image, and the image signals of the defective pixels are interpolated.

The interpolated image signal is supplied to the D / A converter 1
The signal is converted into an analog signal by 11 and displayed on the monitor 113 together with the image signal of the normal pixel. When performing image processing such as threshold processing, the interpolated image signal is supplied to the image processing device 115.

As described above, the defective pixel correction processing device 10 of the image sensor according to the present embodiment detects the defective pixel of the CCD,
Based on this detection result, create a normal / defective label image divided into defective pixels and normal pixels, further create a labeling image in which defective pixels are classified according to the pattern based on the normal / defective label images, and Since the defective pixel is excluded based on the labeling image and the image signal of the defective pixel is interpolated using the image signal of the pixel adjacent to the defective pixel, the image signal of the defective pixel can be appropriately corrected. .

Further, since the defective pixel correction processing device 10 of the image pickup device according to the present embodiment interpolates only the normal pixels adjacent to the defective pixel, it is possible to reduce the influence of other defective pixels and the influence of distant pixels. Can be made, and the interpolation trace becomes inconspicuous. Furthermore, the defective pixel correction processing device 10 of the image sensor according to the present embodiment creates a normal / defective label image from the detected defective pixel, and classifies and labels the defective pixel based on the label image. Labeling can be performed effectively.

In the image sensor defective pixel correction processing apparatus 10 of the present embodiment, the frame addition processor 23 and the LPF
The defective pixel detection unit 1 includes a processor 25, a subtracter 27, a normal / defective pixel separator 29, and a defective pixel classifier 31.
3 is configured, but when the circuit size is limited, software or hardware having the same function as the defective pixel detection unit 13 may be used.
In this case, the configuration is such that the defective pixel detection unit 13 is removed from the defective pixel correction processing device 10 of the image sensor shown in FIG. 1 as shown in FIG. 17, and the labeling image is stored in the correction data ROM 15 directly from the software or hardware. To do.

In the present embodiment, the defective pixel correction processing apparatus 10 for an image sensor has been described by way of example in which a defective pixel of a CCD used in a TV camera 105 is interpolated. However, the present invention is not limited to this. Instead, the present invention can be applied to a case where a defective pixel of another image sensor, for example, a MOS image sensor is interpolated.

Furthermore, in the defective pixel correction processing apparatus 10 for the image sensor of the present embodiment, interpolation processing and the like are performed using the ROI of a 3 × 3 pixel group, but the present invention is not limited to this. Alternatively, ROIs of other pixel groups may be used.

Further, the defective pixel correction processing device 10 of the image sensor according to the present embodiment has been described as an example in which the defective pixel of the CCD of the TV camera 105 mounted on the X-ray diagnostic device 100 is interpolated. The present invention is not limited to this, and can be applied to an image sensor used in any device, such as when interpolating a defective pixel of an image sensor used in an endoscope device.

[0065]

As described above, according to the present invention, defective pixels are classified based on the result of detection of defective pixels, and corresponding to the classification result, except for defective pixels, pixels adjacent to defective pixels are removed. Since the image signal of the defective pixel is interpolated using the image signal of (1), the image signal of the defective pixel can be appropriately corrected.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a defective pixel correction processing device for an image sensor according to the present invention.

FIG. 2 is a block diagram illustrating a defective pixel detection unit of the defective pixel correction processing device of the image sensor illustrated in FIG. 1;

FIG. 3 is a diagram showing a defective pixel extracted by a normal / defective pixel separator and a normal / defective label image.

FIG. 4 is a diagram showing a classification example of defective pixels within a predetermined ROI (3 × 3 pixels).

FIG. 5 shows an ROI (3 ×) used for describing labeling.
FIG. 3 is a diagram showing names of pixels in (3 pixels).

FIG. 6 is a diagram showing a classification example in which defective pixels within a predetermined ROI (3 × 3 pixels) are divided into 10 types.

FIG. 7 is a diagram illustrating an example of a case where labeling is performed for each type of defective pixel illustrated in FIG. 6;

8 is a diagram illustrating correction coefficients (simple average) for each type of defective pixel illustrated in FIG. 6;

9 is a diagram showing correction coefficients (reverse ratios of distances) for each type of defective pixel shown in FIG. 6;

FIG. 10 is a block diagram illustrating a buffer memory of the defective pixel correction processing device of the image sensor illustrated in FIG. 1;

FIG. 11 is a timing chart showing an operation timing when three lines of the buffer memory are simultaneously output.

FIG. 12 is a block diagram illustrating a defective pixel correction processing unit of the defective pixel correction processing device of the image sensor illustrated in FIG. 1;

FIG. 13 is a block diagram showing a defective pixel correction processing unit when correcting a defective pixel using pixels in a 3 × 3 ROI.

FIG. 14 is a block diagram illustrating operation timings of a defective pixel correction processing unit when a defective pixel correction process is performed using pixels in a 3 × 3 ROI.

15 is a diagram showing names of pixels used for explaining the operation timing shown in FIG.

FIG. 16 is a diagram illustrating another configuration of the defective pixel correction processing unit.

FIG. 17 is a block diagram showing a defective pixel correction processing device for an image sensor according to the present invention when a defective pixel detection unit is replaced with hardware or software.

FIG. 18 is a block diagram showing a schematic configuration of an X-ray diagnostic apparatus equipped with a TV camera and a defective pixel correction processing device.

FIG. 19 is a diagram showing image signal values before and after replacement by an adjacent pixel replacement method.

FIG. 20 is a diagram showing image signal values before and after replacement by median processing.

FIG. 21 is a diagram illustrating an image signal value before interpolation by the bilinear interpolation method, a diagram illustrating pixels used in the bilinear interpolation method, and an image signal value after interpolation by the bilinear interpolation method. FIG.

FIG. 22 is a diagram illustrating pixels used for high-dimensional interpolation processing using a SINC function.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 10 Defective pixel correction processing device of image sensor 11 Switch 13 Defective pixel detection unit 15 Correction data ROM 17 Correction coefficient generation unit 19 Buffer memory 21 Defective pixel correction processing unit 23 Frame addition processor 25 LPF processor 27 Subtractor 29 Normal / defective Pixel Separator 31 Defective Pixel Classifier 33-11, 33-12, 33-13, 33-21, 33-22,
33-23, 33-31, 33-32, 33-33 33 Subtractor 35 Horizontal / Vertical Addition Processing Unit 37-1, 37-237-3, 45 Latch 39-1, 41-1, 39-2, 41 -2,39-3,41-3,4
3,47 Adder LM Line memory MP Multiplexer P Subject

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Koichiro Nabuchi 1385-1, Shimoishigami, Otawara City, Tochigi Prefecture Inside Toshiba Nasu Plant (72) Inventor Akira Tsukamoto, 1385-1, Shimoishigami, Otawara City, Tochigi Prefecture Inside the Toshiba Nasu Factory (72) Inventor Tohru Kato 1385-1 Shimoishigami, Otawara City, Tochigi Prefecture Inside the Toshiba Nasu Factory (72) Inventor Seiichiro Nagai 1385-1, 1 Shimoishigami Shimoishigami, Otawara City, Tochigi Prefecture Dical Engineering Co., Ltd.

Claims (3)

[Claims] 1. A defective pixel correction processing device for an image sensor for correcting an image signal of a defective pixel of an image sensor having a plurality of light receiving means arranged in a pixel, the defective pixel detecting means detecting a defective pixel. A defective pixel classifying unit for classifying a defective pixel based on the detection result of the defective pixel detecting unit; and an image signal of a pixel adjacent to the defective pixel except for the defective pixel, corresponding to the classification result of the defective pixel classifying unit. A defective pixel correction processing unit for correcting a defective pixel by interpolating an image signal of the defective pixel using the method. 2. The defect classifying means creates a normal / defective label image in which a defective pixel is at a high level and a normal pixel is at a low level, and generates a defect for each of a predetermined number of pixel groups based on the normal / defect label image. Classification by pixel pattern is performed, and labeling is performed for each of the predetermined number of pixel groups in accordance with the type, and the defective pixel correction processing unit is configured to perform the predetermined number of pixel groups in accordance with the labeling by the defect classification unit. 2. An image signal of a defective pixel is interpolated by using an image signal of a pixel adjacent to the defective pixel except for the defective pixel every time.
13. A defective pixel correction processing device for an imaging device according to claim 1. 3. The defect classifying means creates a normal / defective label image in which a defective pixel is at a high level and a normal pixel is at a low level, and a defect for each of a predetermined number of pixel groups is generated based on the normal / defect label image. Classification according to the pattern of pixels is performed, and labeling is performed for each of the predetermined number of pixel groups in accordance with the type, and the defective pixel correction processing unit calculates a correction coefficient corresponding to the labeling performed by the defect classification unit to the predetermined number of pixels. A correction coefficient generation unit that is generated for each pixel group, and corresponding to the labeling by the defect classification unit, an image signal of a pixel adjacent to the defective pixel except for the defective pixel for each of the predetermined number of pixel groups, and the correction coefficient 2. The defective pixel correction processing device for an image sensor according to claim 1, wherein the image signal of the defective pixel is interpolated using the correction coefficient generated by the generating unit.