JPS61288667A - solid-state imaging device - Google Patents

Abstract

PURPOSE:To obtain the suitable correcting value concerning the picture element of the maximum appearance frequency gain value by selecting the correspondence that the product of the central gain value and the gain correcting coefficient value of the distribution comes to be mutually the same level, using the correcting value of the picture element with the maximum signal versus noise ratio in the picture element and executing the displaying. CONSTITUTION:The value of the gain correcting coefficient is for example set beforehand as shown in Figs. B and C rows. When the calculation, which multiplies the gain correcting coefficient value to the gain value, is executed as shown by a B row, the suitable correction is executed in the scope of gain value 5-10 bits by the correction, and the considerable scope of the central part of distribution is suitably improved. Then, still, the correction of the signal of four bits is suppressed. Next, the input of either of the gain value or the gain correcting coefficient value to a multiplier is shifted to one digit and the correspondence shown in the C row is selected. In the correction, the suitable correction is executed by the scope of gain values 4-9 bits, and the picture element of the central part of the gain distribution is sufficiently included.

Description

[Detailed Description of the Invention] [Summary] The present invention, in a solid-state imaging device, easily corrects the central portion of the distribution when correcting the gain of the image signal, and if necessary, replaces data of pixels corresponding to the edges of the distribution. By doing this, image display is improved.

[Industrial application field]

The present invention relates to solid-state imaging devices, and particularly relates to improvements in image signal correction methods thereof.

Imaging devices divide optical images into a large number of pixels and process the data, but in order to obtain surface-quality image signals, corrections are made more effectively than before for variations in characteristics between pixels. This is requested.

[Conventional technology]

In a solid-state imaging device, a photoelectric conversion element such as a photovoltaic type, a photoconductive type, or a Mis type is usually arranged in each pixel region of the light receiving surface, and the electrical signals generated by these photoelectric conversion elements are transferred by charge coupling. The signal is multiplexed and detected in time series using a charge transfer device (CTD) such as a charge transfer device (CCD), and the signal is corrected after non-pulling and digital conversion, and then displayed on a display device.

Conventionally, the compensation has been considered to be divided into defective pixel correction in which data of another pixel is substituted for a defective pixel having no sensitivity at all, and correction of variations in sensitivity between pixels.

The former defective pixel correction is performed when, for example, pixel X is a defective pixel with no sensitivity among pixels arranged two-dimensionally as shown in Figure 2(a), and a normal defective pixel among the eight pixels adjacent to this pixel Pixel data is also displayed at pixel coordinates X.

This defective pixel correction is performed with a circuit configuration as shown in the block diagram of FIG. Each pixel data recorded in is sent to the display device 23 via the driver 22. In contrast, in the defective pixel correction U3 work, the address of the pixel to replace it is converted or the address of the necessary pixel is sent to the display device 23 via the driver 22. For example, the address conversion memory 25 in which .

The latter sensitivity variation correction is performed before the 1rI regular defective pixel correction, and is generally performed using the following calculation formula.

(p - p+,)/(po-r'+,) where P is the output when looking at the object to be imaged, PL is the output when looking at the standard Turkesian I with a uniform small amount of incident light, Pl+ is the output when looking at a reference Kukerat with a uniform large amount of incident light.

This sensitivity variation correction is divided into a numerator offset correction and a denominator gain correction in the O;1 notation.

Offset correction is correction for variations in the DC component unrelated to the intensity of incident light, and gain correction is correction for variations in the ratio of the amount of change in output to the amount of change in incident light. This is done as shown in the figure. That is, the offset correction value of each pixel is stored in the offset memory 28 and the gain correction coefficient is stored in the gain memory 29 in advance, the offset correction value is subtracted by the adder 26 from the digital image signal, and the gain is calculated by the multiplier 27. Multiply by correction factor.

As is known from the above equation, the offset correction value is set based on the output P, - of each pixel when looking at a reference target with a uniform small amount of incident light, and the gain correction coefficient is set based on the reference target with a small amount of incident light and large amount. It is set based on a value obtained by multiplying the reciprocal of the output difference PHPt of each pixel when looking at the target by a constant.

In the conventional gain correction method, as shown in row D of FIG. 4, a pixel with the largest gain is used as a reference, and a coefficient is used to adjust the level of pixels with a smaller gain to this pixel.

That is, for example, when the gain of each pixel is represented by a maximum of 8 bits and varies up to a minimum of 1 bits, an 8-bit gain correction coefficient is required in order for the number of effective digits of the 8-bit gain compensation iE result to be 8 bits. , the correction result is a maximum of 16 hints. However, the digits lower than the effective minimum digit shown in the middle diagram of the figure are within the error range. In order to match the level of the correction result of a gain with a small number of digits, a gain correction coefficient of 9 to 15 points is required as shown in the figure.

In images obtained by such correction, large flickers appear in low-gain pixels with a small number of effective digits, which is extremely annoying to the eyes.

In the correction method according to Japanese Patent Application No. 57-18198/l, which was previously provided by the present inventor, when setting the gain correction coefficient, as shown in line E of FIG. ) are ignored and processed all at once, and the level of the correction result of pixels with a small number of effective digits is lowered to suppress the flickering described above.

Note that the correction method according to the invention has the effect of compressing the maximum number of digits of the gain correction coefficient, and therefore the calculation hint length.

[Problems to be Solved by the Invention] Among the conventionally known signal correction methods for solid-state imaging devices, gain correction is based on the gain with the highest pitch I-number as described above.
1L, however, in the distribution of the sensitivity of the photoelectric conversion element, the distribution frequency near the maximum value is generally small, and even if this is faithfully displayed, the effect is small.

Further, in the correction method of the invention, depending on the distribution pattern of the sensitivity of the photoelectric conversion element, etc., pixels with a gain having a high distribution frequency may be ignored, and a good image cannot be obtained.

In order to escape from the current state of affairs as described above and further improve the images of solid-state imaging devices, there is a need for improvements in image signal correction methods.

[Failure to solve the problem]

The above problem is solved by shifting at least one of the gain value of each pixel and the preset gain correction coefficient value as necessary to obtain the gain value and the gain correction coefficient value near the center of the distribution of the gain value. Select a correspondence between the gain value and the gain correction coefficient value such that the product of the two values is at the same level, and upon imaging, perform a gain correction calculation using the gain correction coefficient value according to the correspondence, and obtain the same level. For a pixel with a gain value that yields a product of This problem is solved by the solid-state imaging device according to the present invention, which displays an image using the correction value of the pixel with the maximum signal-to-noise ratio in the pixel where the product of .

[For production]

In the solid-state imaging device according to the present invention, combinations of gain correction coefficient values are set in advance. It goes without saying that this value can be expressed as the reciprocal of the output difference P, I Pt of each pixel when looking at the reference target with large and small incident light amounts, using the above formula, but it is usually the minimum value. The number of digits is smaller than the maximum number of digits of the gain, that is, for a gain value with a maximum number of digits or a number of digits close to the maximum, the number of effective digits after correction is reduced. Regarding the value, the number of digits of the correction coefficient value is suppressed.

At least one of the gain correction coefficient value and the gain value of each pixel, that is, the output difference PH−PL, is shifted as necessary and inputted to the gain correction multiplier, and the majority frequency near the center of the distribution of gain values is A correspondence between the values such that the product of the value and the positive coefficient value of the gain cine 11 is at the same level is selected.

In this way, the preset and selected gain correction coefficient value and the above-mentioned correspondence are combined and correspond to the conventional gain correction coefficient, and during imaging, the gain correction calculation is performed using this gain correction coefficient value according to this correspondence. conduct.

Through this gain correction calculation, appropriate correction values can be obtained at least for pixels having a large number of gain values near the center of the distribution.

However, for a pixel whose gain value is larger than the gain value for which a product of the same level can be obtained, the operation length of the multiplier is overflowed and an appropriate correction value cannot be obtained. In addition, the correction value is replaced with the correction value of the pixel with the maximum signal-to-noise ratio in the pixel for which a product of the same level can be obtained.

Note that defective pixel correction in which data of another pixel is substituted for a defective pixel having no or almost no sensitivity is performed in the same manner as in the conventional example.

According to the present invention, by performing the above-mentioned correction after offset correction, a good image without peculiar pixels can be displayed.

〔Example〕

Hereinafter, the present invention will be explained in detail by way of examples with reference to FIG. However, line A in the figure shows the conventional method according to the invention for comparison, line B shows the intermediate stage of the correction operation according to the invention, and line 0 shows the completion stage of the correction operation according to the invention.

The solid-state imaging device of this embodiment is equipped with a 12-bit gain pressure multiplier, and the gain correction coefficient is limited to a maximum of 12 bits as shown in the column of gain correction coefficients in FIG. Also, dogs, the basis of small incident light intensity [1! Assume that the distribution of the output difference PIIP+ of each pixel when looking at the cuget, as shown in the column of gain and gain distribution in the figure, has a maximum of 10 bits and a central range of 7 to 4 bits.

In this case, if the above-mentioned prior art is used to correct the maximum sensitivity standard in which the number of effective digits after correction for a pixel with a gain of 10 I. The pixels of bin) are corrected accurately, but 7 Hiso I
- For the following pixels, the correction coefficient is rounded to a constant value of 12 hits, and the level after gain correction is suppressed, so that the central range of the gain distribution, which accounts for the majority of pixels, is not faithfully displayed.

In contrast, in the gain correction of the present invention, the value of the gain correction coefficient is set in advance, for example, as shown in lines B and C in the figure. In this example, the maximum number of digits for the gain is 10 bits, and the minimum number of digits for the correction coefficient value is 7 bits, and it is expected that the number of effective digits will decrease after correction for a gain with a large number of digits. The number of digits is <12 bits as described above, and as in the invention, the number of digits is suppressed for gains with a small number of digits.

Assume that the calculation of multiplying the gain value by this gain correction coefficient value is performed as shown in line B. In this correction, the gain value is 5 to 10.
Appropriate correction has been made for the bit range, and a considerable range in the central part of the distribution is correct, and it is an improvement compared to the correction for row A, and the correction for the 4-bit signal is suppressed. .

In this embodiment, in this case, either the gain value or the gain correction coefficient value input to the multiplier is shifted by one digit l-, and the correspondence shown in the 0th row is selected. In this correction, the gain value is 4.
Appropriate correction is performed in the range of ~9 pi I-, and pixels in the central part of the gain distribution are sufficiently included.

During imaging, gain correction calculations are performed using the gain correction coefficient values in correspondence with each other in this manner.

However, in this example, the maximum gain of 10 pips 1-length overflows the operation length of the multiplier and an appropriate correction value cannot be obtained. Of the pixels that have undergone gain correction,
The correction value of the pixel with the highest signal-to-noise ratio is replaced and displayed.

Furthermore, if there are defective pixels with no or almost no sensitivity,
Defective pixel correction is performed to replace this pixel with data of another pixel in the same manner as in the prior art example. With the above correction, a good image without any unusual pixels is displayed.

Furthermore, if the gain correction coefficient value is shifted by 1 bit from the value in the above embodiment, it becomes 12 bits for a gain of 4 bits or less, 11 bits for a gain of 5 bits, etc.
If the gain is set to 6 bits for a gain of 0 bits, the central part of the gain distribution of this embodiment will be included in the range of appropriate correction even in the absence of the shift function according to the present invention. If the central part of the gain distribution shifts downward, the central part will not be sufficiently included in the appropriate correction range, and if it shifts upward, the number of effective value digits will be lost.

〔Effect of the invention〕

As explained above, according to the present invention, in a solid-state imaging device of a given scale, an appropriate correction value can be easily obtained for the pixel having the gain value that appears most frequently, and Effective processing is also performed on pixels, resulting in a good image in which information is sufficiently displayed and there are no unusual pixels.

[Brief explanation of drawings]

Fig. 1 is a diagram showing an example of the gain correction method according to the present invention, Fig. 2 (a) is a schematic diagram of defective pixel correction, Fig. 2 (bl is a block diagram of defective pixel correction, and Fig. 3 is sensitivity variation correction FIG. 4 is a block diagram showing an example of a conventional gain correction method.

Claims (1)

[Claims] At least one of the gain value of each pixel and a preset gain correction coefficient value is shifted as necessary to obtain a gain value near the center of the distribution of the gain value and the gain correction coefficient. Select a correspondence between the gain value and the gain correction coefficient value whose products with the numerical values are at the same level, perform gain correction calculation using the gain correction coefficient value according to the correspondence during imaging, For a pixel with a gain value from which a product of levels can be obtained, the calculation result is used as a correction value, and for a pixel whose gain value is larger than the gain value from which a product of the same level can be obtained, a pixel that is located near the pixel and has the same A solid-state imaging device characterized in that an image is displayed using the correction value of a pixel with a maximum signal-to-noise ratio among pixels for which a product of levels can be obtained.