JP2672285B2 - Defect detection method for solid-state image sensor - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting defects in a solid-state image pickup device, for example, when a solid-state image pickup device used in a video camera is manufactured, the quality of the product is judged. It is suitable for application in some cases. 2. Description of the Related Art A solid-state image sensor of this type has a large number of pixels formed in a plane lattice, and has a uniform characteristic over the entire light-receiving surface of the image sensor during manufacture. It is practically difficult to obtain, and the photoelectric conversion output obtained from the pixels arranged in sequence contains various noise components. That is, when detection data RD composed of a photoelectric conversion output is detected for one horizontal section, FIG.
As shown in (A), first, the detection data RD has a shading component SH in which the average signal level gradually changes when the solid-state image sensor is horizontally scanned. The shading component SH is generated based on unevenness of aperture, conversion, unevenness of etching, unevenness of dyeing of a filter, and the like of each pixel, and has a relatively low frequency. Secondly, the detection data RD has a noise component K0 which is generated on the basis of unevenness between pixels. This noise component K0 has a relatively high frequency and a relatively large fluctuation range. The above is a phenomenon that can be seen in a solid-state imaging device that does not include a defective pixel. However, if there is a defective pixel, the detection data RD is that if there is a defect, a white defect WD that rises in a pulse shape at a position corresponding to the defect. Alternatively, a black defect BD occurs. In order to detect the defective pixels by extracting the defects WD and BD based on the defective pixels from the detection data RD including various fluctuation components as described above. In the past, a method of extracting the detection data RD using a spatial filter has been used. However, according to this method, the size of the defect to be detected,
It is necessary to set the size and weighting function of the spatial filter to adapt to this based on the type or frequency characteristics, but it is often difficult to set the optimum value in practice.
For example, it is extremely difficult in practice to design a spatial filter having an optimal frequency characteristic when detecting a defect from the detection data RD based on the frequency characteristic, and a simple filter such as a 3 × 3 or 5 × 5 filter is used. This is practically difficult to achieve. Further, according to the conventional defect detection method, when a defect is discriminated by comparing the output obtained from the spatial filter supplied with the detection data RD with a reference value in the comparator, the defect is detected. When the signal level of the detection data RD changes when the set state or the output level changes, the true data obtained from the defective pixel cannot be estimated, which makes it difficult to automatically determine the defective pixel. . For example, a set obtained by applying Laplacian-1-1-1-18-1-1-1 to a defect changes as shown in FIG. Further, according to the conventional defect detection method, when the defect detection processing is performed using the spatial filter, it is not possible to obtain the determination result for the pixels in the area of the outer peripheral edge of the solid-state image pickup device. There is. In this connection, this conventional defect detection method generally takes a wide range of spatial filtering and performs filtering a plurality of times. In this way, it is unavoidable to generate a blank area at the periphery. For example, when a 3 × 3 spatial filter is used for the defect detection data RD, as shown in FIG. 4, the calculation cannot be performed for one bit of pixels in the outer peripheral area. The present invention has been made in consideration of the above points, and obtains a defect detection signal by using a defect removal filter capable of realizing excellent smoothing of a minute space when obtaining a shaded extraction output from an image signal. At the same time, it is intended to effectively solve the problem that the defect detection processing cannot be performed for the pixels in the blank area formed in the peripheral portion of the solid-state image sensor by digitally operating the defect removal filter. is there. In order to solve such a problem, according to the present invention, detection data S1 is sequentially obtained from a plurality of adjacent pixels M1-M5 of the solid-state image pickup device 1, and the detection data S1 is obtained as a median. By performing a filtering process using the filter 9, the detection data having an intermediate value among the detection data of the plurality of pixels M1 to M5 is converted into positions x (i-2) j to x (i + 2) of the plurality of pixels. Among the pixels constituting the image pickup surface of the solid-state image pickup device 1, the regions 21 to 24 of the second pixel portion at the outer periphery of the pixels constituting the image pickup surface of the solid-state image pickup device 1 are sequentially selected as the data of the pixel at the substantially central position xij of j. Data D which is obtained as the first shading data corresponding to the area of the first pixel portion to be excluded and is located at the outer peripheral edge of the first shading data
1 to D4 are used to interpolate as second shading data corresponding to the area of the second pixel portion in the outer peripheral edge of the imaging surface, and the first and second data are obtained from the data of each pixel of the detection data S1. The subtraction data S6 is obtained by subtracting the corresponding data of the shading data, and the subtraction data S6 is obtained.
The defect detection signals WD and BD included in the detection data S1 are extracted by comparing 6 with the reference value. The detection data having the intermediate value among the detection data S1 of the plurality of pixels M1 to M5 is sequentially selected as the data of the pixel at the substantially central position among the plurality of pixels M1 to M5. By using the median filter 9, it is possible to achieve much better smoothing of the minute space, and as a result, an accurate defect detection signal can be obtained.
At the same time as this processing, by interpolating the data of the pixel group in the peripheral portion of the shaded extraction output as the data of the pixel group in the blank area, the shaded component included in the extracted data is extracted and output. Therefore, when subtracting the solid-state image pickup device 1, it is possible to surely remove the shading component from the pixels in the peripheral portion of the solid-state image pickup device 1. By doing so, it is possible to reliably detect the defect generated in the outer peripheral portion of the solid-state image pickup device 1. be able to. An embodiment of the present invention will be described in detail below with reference to the drawings. In FIG. 1, reference numeral 1 denotes a solid-state image pickup device, which is formed by arranging a large number of pixels such as CCDs (Charge Coupled Devices) in a plane lattice pattern in the H direction and the V direction. Photoelectric conversion. The photoelectric conversion output of each pixel is sequentially scanned by the horizontal clock signal SH and the vertical clock signal SV generated by the clock driver 3 in units of 1H, and the time-series detection data R is output.
As D, it is input to the analog-digital conversion circuit 5 through the input circuit 4, converted into a digital signal, and then input to the picture memory 7 of the shading extraction circuit 6. Each time the picture memory 7 receives the detection data RD arriving every 1H, the picture memory 7 executes a synchronous addition operation for each pixel by using the accumulator 8, thereby reducing a noise component included in the detection data RD. After performing the processing, the detection data is stored and held.
Thus, as shown in FIG. 2B, the detection data RD (FIG.
The input data S1 obtained by suppressing the high frequency noise component K0 based on the unevenness for each pixel included in (A) and converting it into the small noise component K1 is sent from the picture memory 7. The solid-state image pickup device 1 is used as the input data S1.
Signals WD and BD and 1H
A shading component SH that fluctuates slowly during the interval
And are sent as they are. Where the shielding component S
Since H has a property of changing with a similar tendency with respect to the entire line signal for 1H that sequentially arrives, H is not suppressed by the synchronous addition for each frame and is output as it is. Also, the defect signals WD and BD are generated as similar signals for a plurality of adjacent line signals, and therefore remain without being suppressed. This is because, in the case where a CCD is used as the solid-state imaging device 1, the abnormality that occurs in the pixel due to its nature is not continuous in the H direction in most cases but only in the V direction. Input data S1 having such signal contents
Is input to the defect removing filter 9. The defect removing filter 9 is composed of a 1 × 5 median filter as shown in FIG. 5, and the intermediate value of the data of 5 pixels successively following one line signal of the input data S1 is used as the intermediate position data. Send out. That is, in FIG. 5, the i-th line included in the j-th line output of the input data S1.
The data "40" of the th pixel (whose address is xij) is input to the central bit position, that is, the third bit position M3, and the addresses x (i-
1) The data “12” of the pixel of j, x (i-2) j,
Store "10" in the second and first bit positions M2, M1,
Moreover, the data "9" and "11" at the addresses x (i + 1) j and x (i + 2) j are input to the fourth and fifth bit positions M4 and M5. In this state, the filter 9 outputs the data representing the intermediate value of the data input to the first to fifth bit positions M1 to M5 (in the case of this embodiment, the fifth bit data "11"). Is output as output S2. In such an operation, the content "40" of the data input to the third bit position M3 shows a large value representing a white defect, so that the other first, second, fourth and fifth bits are used. Contents of position M1, M2, M4, M5 "1
“0”, “12”, “9”, and “11” have signal levels indicating that the corresponding pixels are normal. In such a case, the data at the center position xij is set to the intermediate value data "11" (address x (i + 2) j by the median filter operation).
Is transmitted as a filter output S2. In this manner, the defect removing filter 9 is used.
Means that even if a defect is included in the input data S1, a filter output S2 from which the defect has been removed can be transmitted. When the 1 × 5 median filter is used as the defect removing filter 9 as described above, if there are three or more defects in five consecutive bits of line input, the defect removing operation cannot be performed. In such a case, it is determined that the solid-state image sensor under inspection is defective. This filter output S2 is the buffer memory 1
After the noise component is further suppressed by performing a simultaneous addition operation using the accumulator 8 as shown in FIG.
A shading extraction output S3 as shown in FIG. Here, the shading extraction output S3 is compared with the input data S1 (FIG. 2B) and the defect data WD and BD are compared.
Is removed, and the noise component K2 based on the unevenness of each pixel is further reduced. The shading extraction output S3 is provided to the low-pass filter 11. The low-pass filter 11 smoothes the noise component K2 contained in the shaded extraction output S3. For example, a 3 × 3 low-pass filter as shown in FIG. 6 or 7 can be applied. Thus, at the output end of the low-pass filter 11, as shown in FIG. 2 (D), a smoothed output S4 consisting of substantially only the fading component SH is obtained, which is supplied to the peripheral edge interpolation circuit 12. Since the peripheral portion interpolating circuit 12 cannot avoid a blank area which cannot be calculated by calculation in the peripheral portion of the solid-state image pickup device 1 during the processing in the defect removing filter 9 and the low-pass filter 11, this blank space is inevitable. It is intended to interpolate the data in the area so that the subsequent processing can be performed. By the way, if there is a pixel group in which it is not actually determined whether or not the fixed image sensor 1 can be used for the outer peripheral edge portion, it means that the peripheral edge portion cannot perform the photoelectric conversion function, so the fixed image sensor is effectively used. In order to do so, it is better that there is no such undecidable area. Actually, if the defect removing filter 9 is a 1 × 5 median filter and the low pass filter 11 is a 3 × 3 filter as in the embodiment of FIG. 1, the smoothed output S4 is shown in FIG. As shown, 3-bit blank areas 21 and 22 are formed in the outer peripheral edge in the H direction, and a 1-bit blank area 23 is formed in the outer peripheral edge in the V direction.
And 24 occur. The blank areas 21 to 24 are interpolated by the peripheral interpolating circuit 12 in the order shown in FIG. That is, first, the blank area 21 on the left edge is interpolated using the data D1 of the pixel group of 3 bits inside the blank area 21 as shown in FIG. 9 (A). Thus, as shown in FIG. 9 (B), two data areas having the same data D1 are contiguously formed on the left peripheral portion. Next, the peripheral portion interpolation circuit 12 interpolates the data D2 of the pixel group of 1 bit inside the blank area 23 of the upper edge portion to obtain two data having the same data D2 as shown in FIG. 9C. Interpolation is performed to connect the data areas. Subsequently, as shown in FIG. 9C, the peripheral portion interpolation circuit 12 performs an interpolation operation on the blank area 22 on the right side edge by using the data D3 of 3 bits inside the blank area 22, and as shown in FIG. 9D. Thus, two areas having the same data D3 are formed so as to be connected to each other. After that, the peripheral edge interpolation circuit 12 inserts one bit of data D4 inside the blank area 24 at the lower edge.
Is interpolated to form two areas having the same data D4 on the lower edge as shown in FIG. 9 (E). In this way, the peripheral portion interpolation circuit 12 generates an interpolation output S5 having corresponding data over the entire pixels of the solid-state imaging device 1, and subtracts this from the input data S1 sent from the picture memory 7 in the subtraction circuit 13. Since the input data S1 is the same data as the noise component K0 of the detection data RD (FIG. 2A) obtained in the solid-state image pickup device 1 as described above with reference to FIG. The outputs of the pixels of the entire solid-state imaging device 1 are output including those of the peripheral portion. Therefore, the subtraction output S6 of the subtraction circuit 13 is the input data S
1 (FIG. 2 (B)), the shading component SH is subtracted by the interpolation output S5 (FIG. 2 (E)) to obtain the white defect WD and the black defect B as shown in FIG. 2 (F).
Only D remains to be output. Since such subtraction is executed over the entire surface of the solid-state image pickup device 1 including the outer peripheral edge portion, if the white defect WD or the black defect BD occurs in the peripheral pixel, it can be reliably detected. . Incidentally, the signal component corresponding to the outer peripheral portion of the interpolation output S5 is interpolated with the same data as the data of the pixel group inside thereof, but actually obtains a curve similar to the curve at the outer edge of the shading component SH. Therefore, it is possible to prevent the effect of the shading waveform from remaining on the result of subtraction from the input data S1. The subtraction output S6 of the subtraction circuit 13 is input to the defect determination circuit 14. Upon receipt of the subtraction output S6, the defect judgment circuit 14 has, for example, four comparison levels CO as shown in FIG.
M1 to COM4 are set, and it is determined whether the defect WD or BD exceeds these comparison levels COM1 to COM4. Based on the result of the determination, firstly, the number of defects is integrated to perform so-called score calculation, thereby evaluating the score of the solid-state image sensor 1 currently inspected. Secondly, the defect judgment circuit 14 detects the address where the defect WD or BD occurs and performs so-called shape recognition, score calculation, and post-processing. Third, the defect determination circuit 14 creates a defect map indicating the distribution and size of the defect based on the shape recognition and the point calculation. Thus, the defect determination circuit 14 sends out a defect detection signal ADD containing the result of the defect determination. In the above configuration, when the solid-state image pickup device 1 obtains the detection data RD as shown in FIG. 10A, the filter output S2 is shown by the operation of the median filter in the defect removing filter 9. As shown in FIG. 10 (B), only the shading component SH is formed, which is smoothed by the low-pass filter 11 and the rear peripheral portion interpolation circuit 1
In step 2, interpolation calculation is performed. When the subtraction circuit 13 subtracts from the input data S1 using the interpolation output S5,
The subtraction output S6 can obtain a signal similar to that obtained when the defects WD and BD are separated and extracted from the shading component SH as shown in FIG. Therefore, the defect determination circuit 14 can easily determine the addresses of the defects WD and BD. In the embodiment shown in FIG. 5, the defect removing filter 9 has a 1 × 5 median filter for black and white images, but the defect removing filter 9 for color images is used.
What is necessary is just to apply the thing of the structure of FIG. That is, the solid-state imaging device 1 has a configuration in which pixels corresponding to the three primary color signals R, G, and B are sequentially arranged as shown in FIG. 11A, and each of the three primary color signal pixels has a unique color filter. Since it is mounted, it is necessary to filter the defect removing filter 9 for each color. In such a case, the primary color signals R, G, and B are respectively converted by using a mask 21 formed by forming a mask portion SK for two bits in the middle as shown in FIG. 11B. The separated outputs may be applied to a median filter. FIG. 12 shows still another embodiment of the defect removing filter 9, in which two adjacent bits of the input data S1 extracted from the picture memory 7 are compared and their maximum or minimum values are compared. By selecting the value, the filter output which does not include the black defect or the white defect is obtained. In the case of this embodiment, the defect removing filter 9 is used to obtain the filter output S2 of the pixel xij before and after 5 times.
Using the data of two adjacent bits, the maximum value is selected from the data of two adjacent bits in the first step, and the minimum value of the adjacent data is selected in the second step for the selected output. In the step, the minimum value is selected from the selected adjacent data, and in the fourth step, the maximum value is selected from the two data. In this way, it is possible to obtain a filter output obtained by removing data due to a defect based on a defective pixel as data at a pixel xij position.
Therefore, the calculation of the median filter of FIG. 5 takes too much time, or can be used as a simple configuration instead of a device having a problem that it cannot be applied in hardware. Instead of the defect removing filter 9 described above, a space filter or a two-dimensional filter may be used. Further, the low-pass filter 11 having the configuration shown in FIG.
When a color image is processed as, the configuration of FIG. 13 can be applied. That is, the shading extraction output S3 received from the buffer memory 10 is stored in the memory area M11 as shown in FIG. 13, for example, for the data of the R signal, and the memory area M11 is shifted downward by one bit to fetch the data R2. Then, the memory area M11 is shifted upward by 1 bit to take in the lower data R3, and then the memory area M11 is shifted right by 1 bit to take in the data R4. The low-pass filter 11 has the following (1)
Formula [Equation 1] Is used to obtain the averaged output MW, and this is output as the smoothed output S4. By using the low-pass filter 11 that performs signal processing in this way, it is possible to realize a low-pass filter that can be applied to image signals of any color pattern and that can simultaneously process all pixels. As described above, according to the present invention, the detection data having the intermediate value among the detection data of the plurality of pixels is sequentially used as the data of the pixel at the substantially central position of the plurality of pixels. By using the median filter having the function of selecting, it is possible to achieve much better smoothing of the minute space, and as a result, an accurate defect detection signal can be obtained. According to the present invention, at the same time as this processing, in order to obtain the shaded extraction output in the shaded extraction circuit, when the detection data is arithmetically processed by the median filter, the pixels in the peripheral portion of the solid-state image sensor are It becomes impossible to obtain the shaded extraction output of, but by interpolating the data of the pixel group in the peripheral portion of this shaded extraction output as the data of the pixel group in the blank area, the shaded data included in the extracted data When the components are subtracted by the shaded extraction output, it is possible to reliably remove the shaded components from the pixels at the peripheral edge of the solid-state image sensor, thus ensuring the defects generated at the outer peripheral edge of the solid-state image sensor. Can be detected.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a defect detection method for a solid-state image sensor according to the present invention. FIG. 2 is a signal waveform diagram showing signals of respective parts in FIG. FIG. 3 is a diagram for explaining a conventionally used spatial filter. FIG. 4 is a schematic diagram showing a blank area that occurs when a space filter is used. 5 is a schematic diagram showing a defect removing filter 9 of FIG. 1. FIG. FIG. 6 is a diagram for explaining a low-pass filter 11 of FIG. FIG. 7 is a diagram for explaining the low-pass filter 11 of FIG. FIG. 8 is a diagram for explaining the low-pass filter 11 in the figure. 9 is a schematic diagram for explaining an interpolation operation of a peripheral edge interpolation circuit 12 of FIG. FIG. 10 is a signal waveform diagram of each part showing an experimental result by the defect detection apparatus having the configuration of FIG. 11 is a schematic diagram showing another embodiment of the defect removing filter 9 of FIG. 1. FIG. 12 is a schematic diagram showing another embodiment of the defect removing filter 9 of FIG. FIG. 13 is a schematic diagram showing another embodiment of the low-pass filter 11 shown in FIG. [Explanation of reference numerals] 1 ... Solid-state image sensor, 2 ... Light source, 3 ... Clock driver, 6 ... Shading extraction circuit, 7 ... Picture memory, 8 ... Akymulator, 9 ... Defect removal filter, 10 ... Buffer memory, 11 low-pass filter, 12 peripheral interpolation circuit, 13 subtraction circuit, 1
4 ... Defect determination circuit.

   ────────────────────────────────────────────────── ─── Continuation of front page    (72) Inventor Akio Kanamaru               6-7-35 Kita Shinagawa, Shinagawa-ku, Tokyo               ー Inc.                (56) References JP-A-52-100829 (JP, A)                 JP-A-55-104173 (JP, A)

Claims (1)

(57) [Claims] Obtaining detection data sequentially from a plurality of adjacent pixels of the solid-state image sensor, and filtering the detection data using a median filter to obtain detection data having an intermediate value among the detection data of the plurality of pixels, A region of the second pixel portion at the outer peripheral edge of the pixels forming the image pickup surface of the solid-state image pickup device is sequentially selected as the data of the pixel at the substantially central position of the positions of the plurality of pixels. Is obtained as the first shading data corresponding to the area of the first pixel portion excluding the above, and the data at the outer circumferential edge of the imaging surface is used by using the data at the outer circumferential edge of the first shading data. interpolating a second Shiedein grayed data corresponding to the region of the second pixel portion, the first from the data of each pixel of the detection data And subtracting corresponding data of the second shading data to obtain subtraction data, and comparing the subtraction data with a reference value to extract a defect detection signal included in the detection data. Defect detection method for solid-state imaging device.