JP3788997B2 - Image signal processing device - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to an image signal processing apparatus and an image signal processing method , and more particularly to an image signal processing apparatus and an image signal processing method that are preferably used for controlling non-linear processing for converting a dynamic range of a signal processing circuit.
[0002]
[Prior art]
In general, a conventional imaging apparatus has a configuration as shown in the block diagram of FIG. Incident light is photoelectrically converted by the CCD 20, and the gain of this output signal is adjusted by the AGC circuit 21 and converted into an 8-bit digital signal by the A / D converter 22. Then, the normal γ correction is performed by the γ correction circuit 23, and the high luminance signal is nonlinearly suppressed by the KNEE circuit 24 to ensure a dynamic range. The output compressed by the KNEE circuit 24 is supplied to a signal processing circuit 25 that performs image arithmetic processing with 8-bit resolution. The output signal of the signal processing circuit 25 is output as a color video signal, and is input to the AE circuit 26 to control the iris and shutter speed, and the output signal is supplied from the AE circuit 26 to the AGC circuit 21. .
[0003]
[Problems to be solved by the invention]
In such a conventional image pickup device, a non-linear process such as γ correction or KNEE correction is performed with an 8-bit digital signal, and signal processing is performed with the same 8-bit as the input signal based on the information. The range is small.
[0004]
In view of the above problems, an object of the present invention is to provide an image signal processing apparatus capable of performing arithmetic processing such as fine non-linear processing for each screen in accordance with a shooting screen.
[0005]
[Means for Solving the Problems]
An image signal processing apparatus according to the present invention includes an A / D converter that converts an input signal into an n-bit digital signal, and an image that stores the n-bit input signal converted by the A / D converter for each screen. Storage means, knee correction means for performing knee correction processing on an n-bit input signal of one screen stored in the image storage means, converting the signal into an m (m <n) bit signal, and the knee correction means Control means for controlling the characteristics of knee correction processing, and n-bit input signals for one screen stored in the image storage means based on n-bit input signals for one screen converted by the A / D conversion means. Evaluation means for pre-evaluating an input signal, wherein the control means determines an inclination of knee correction based on an evaluation result of the evaluation means , and further inputs n bits of one screen stored in the storage means Included in signal A knee correction slope reference point is determined based on the maximum level signal value, and the knee correction means is determined by the control means for an n-bit input signal of one screen stored in the image storage means. A knee correction process based on the characteristics specified by the inclination and the reference point is performed, converted into an m (m <n) bit signal of one screen, and output .
[0006]
[Action]
The present invention evaluates an image of an input signal, and performs processing on the input signal so as to maximize the variance of the signal based on the evaluation result, thereby performing detailed processing for each screen according to the screen. It is possible. In addition, by providing the image storage means, optimal calculation processing for each screen is executed without time delay.
[0007]
【Example】
Hereinafter, an image signal processing apparatus of the present invention will be described in detail with reference to the drawings by taking an imaging apparatus as an example.
[0008]
FIG. 1 shows an overall block diagram of an imaging apparatus according to the present invention. Incident light from a subject passes through a lens 1 and an iris 2 for adjusting the amount of light, and enters a photoelectric conversion unit including a CCD (solid-state imaging device) 3. Then, it is converted into a color video signal by the following processing.
[0009]
In the CCD 3, the photocharge generated in the light receiving unit is transferred to the transfer unit and taken out as an output signal. The gain of this signal is adjusted by the AGC circuit 4 and converted into a 12-bit digital signal by the A / D converter 5. This output signal is stored in an image memory (field memory) 6. Then, the normal γ correction is performed by the γ correction circuit 7, and the high luminance signal is nonlinearly suppressed by the KNEE circuit 8 to ensure the dynamic range.
[0010]
On the other hand, the digital signal after A / D conversion is input to the evaluation circuit 10 as a 12-bit output. In the evaluation circuit 10, evaluation based on image information over the entire screen is performed with reference to the information in the histogram memory 11, and an output signal based on the evaluation result is sent to the control circuit 12 that controls the KNEE circuit 8 and the AGC circuit 4. Supply. The control circuit 12 appropriately performs KNEE correction of the luminance signal as will be described later, and fine processing can be performed for each screen according to the imaging screen. Here, in the present embodiment, a 12-bit digital signal is subjected to nonlinear processing such as γ correction or KNEE correction to obtain an 8-bit signal, so that the dynamic range of the high luminance signal can be further increased. The output compressed by the KNEE circuit 8 is supplied as 8-bit data so as to match the signal processing circuit 9 that performs image arithmetic processing with 8-bit resolution. The output signal of the signal processing circuit 9 is output as a color video signal, and is also input to the AE circuit 13 to generate an aperture control signal for controlling the iris and a shutter speed control signal for controlling the CCD drive clock. 14.
[0011]
Next, the operation of the KNEE circuit will be described.
[0012]
FIG. 2 is a graph showing a nonlinear characteristic curve according to an embodiment of the present invention. The vertical axis A in FIG. 2 takes the incident light quantity, the horizontal axis B in the figure takes a signal converted to 12 bits, and the signal corresponding to the incident light quantity is quantized by A / D conversion and output in 12 bits. The conversion process is shown. Further, the 12-bit input signal taken along the horizontal axis B in FIG. 2 is compressed to 8 bits by γ conversion and KNEE correction, and the vertical signal C is taken as the vertical axis C. Indicates.
[0013]
When 12-bit data (horizontal axis B) is non-linearly converted by digital processing, the output signal level is determined by the γ curve until the slope line defining the KNEE characteristic intersects the γ curve from the 0 level. Γ characteristic (for example, y = x ^0.45 ), and KNEEE compression is performed when exceeding this. When the input signal exceeds a predetermined level, the output data is set to the white clip level (a constant value defined by the number of output bits, that is, the maximum output value = 255 level).
[0014]
Therefore, output saturation can be prevented until the level of the input signal reaches a predetermined level.
[0015]
KNEE compression is determined by the slope of the KNEE characteristic. In general, both ends of the oblique lines (start point and end point shown in FIG. 2) are set, and the slope of the KNEE characteristic is determined. As an example of changing the KNEE characteristic in accordance with the histogram, for a histogram in which the appearance frequency of luminance data is high in the brighter side, the inclination angle of the KNEE characteristic is increased so that it is inclined away from the white clip level, and a high-level input signal To ensure resolution. On the other hand, for a histogram in which the appearance frequency of luminance data is high in the darker side, the inclination angle of the KNEE characteristic is reduced so that it approaches the white clip level and the resolution of the low-level input signal can be increased. Note that the histogram in this embodiment counts and evaluates the appearance frequency distribution of the number of samples in a grouping in which brightness is divided into 100 level units. Thereby, the capacity of the histogram memory can be reduced.
[0016]
Hereinafter, a method for controlling the evaluation value and the KNEE characteristic will be described with specific examples.
[0017]
In the first method, as shown in FIG. 3, the end point is the intersection of the highest level of the input signal (4095 on the B axis) and the highest level of the output signal (255 on the C axis). The slope of the KNEE characteristic is set according to the histogram of the input signal, and the start point that is the intersection of the γ curve and the output signal at 200 levels (100%) is obtained. As described above, the slope of the KNEE characteristic is increased, for example, for the histogram with the higher appearance frequency of the luminance data in the brighter side, and for the histogram with the higher appearance frequency of the luminance data in the darker side, the inclination angle of the KNEE characteristic is increased. Reduce the tilt angle.
[0018]
In the second method, as shown in FIG. 4, the intersection of the highest level in one screen of the input signal and the highest level that can be output (255 on the C axis) is the end point, and KNEEE is set according to the histogram of the input signal. A starting point which is an intersection with the γ curve is obtained from the slope of the characteristic. The slope of the KNEE characteristic is the same as in the first method. For example, for a histogram in which the appearance frequency of luminance data is high on the bright side, the inclination angle of the KNEE characteristic is increased and on the dark side, , KNEE characteristic inclination angle is reduced.
[0019]
In the third method, as shown in FIG. 5, the origin is the intersection of the extended line of the KNEE characteristic slope and the vertical axis (C), and the position of the origin of this slope is varied on the vertical axis (C). A starting point which is an intersection with the γ curve is obtained by setting a KNEE inclination according to the histogram of the input signal and translating on the C axis with a constant inclination. For example, for a histogram in which the appearance frequency of luminance data is high on the brighter side, the origin is set to the low level side of the output signal on the vertical axis (C), thereby ensuring the resolution of the high luminance region and the high white clip point. Can be set to level. On the other hand, for a histogram in which the appearance frequency of luminance data is high in the darker side, the origin is set to the high level side of the output signal on the vertical axis (C), thereby ensuring the resolution of the low luminance region.
[0020]
In the fourth method, as shown in FIG. 6, the origin of the slope of the KNEE characteristic is fixed to 100% (200 levels) of the output signal on the vertical axis (C), and KNEE is set according to the histogram of the input signal. A starting point which is an intersection with the γ curve is obtained from the slope of the characteristic. In contrast to the first method, the slope of the KNEE characteristic is opposite to the first method. For example, for a histogram in which the appearance frequency of luminance data is high on the brighter side, the inclination angle of the KNEE characteristic is reduced and on the dark side , Increase the inclination angle of the KNEE characteristic.
[0021]
As described above, optimal image processing is performed on the output of the image memory based on the image data over the entire screen of the shooting screen.
[0022]
In this way, by using the image memory, it is possible to correct the time delay required for generating the evaluation and control signals, so that the optimal image processing calculation for each screen can be executed without any temporal “deviation”. It has become possible.
[0023]
As another embodiment, information before being stored in the image memory is added to the evaluation means, and past information (previous screen) is also used, and nonlinear processing can be performed according to the result.
[0024]
In the above-described embodiment, the evaluation means uses an evaluation using a luminance histogram of an image input signal, and the calculation means uses a non-linear compression calculation. However, the present invention is limited to such an embodiment. As evaluation means, evaluation is performed using a luminance peak value, luminance Min-Max, luminance average value, luminance central value, luminance dispersion value, etc. Non-linear conversion (Min-Max fixed), It is also possible to configure the image signal processing apparatus according to the present invention using a device that performs nonlinear expansion calculation, nonlinear shift, linear compression calculation, linear expansion calculation, addition shift, subtraction shift, and the like.
[0025]
Further, color information, for example, white balance may be controlled according to the result of the evaluation means. As an evaluation means for evaluating the color information, a means for evaluating a color histogram, a color peak value, a color Min-Max, a color average value, a color median value, a color dispersion value, and the like can be used.
[0026]
Furthermore, an input 8-bit digital signal can be converted into an output signal that has an output of 6 bits and applied to a display such as a liquid crystal display device. The parameter setting for determining the KNEE inclination is not limited to the present embodiment.
[0027]
【The invention's effect】
As described above, according to the image signal processing device of the present invention, the image of the input signal is evaluated, and the input signal is subjected to arithmetic processing so that the variance of the signal is maximized based on the evaluation result. Accordingly, detailed processing can be performed for each screen. In addition, by providing the image storage means, it is possible to execute optimal calculation processing for each screen without time delay.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an embodiment of an imaging apparatus according to the present invention.
FIG. 2 is a graph showing a nonlinear characteristic curve according to an embodiment of the present invention.
FIG. 3 is a graph showing a nonlinear characteristic curve according to another embodiment of the present invention.
FIG. 4 is a graph showing a nonlinear characteristic curve according to another embodiment of the present invention.
FIG. 5 is a graph showing a nonlinear characteristic curve according to another embodiment of the present invention.
FIG. 6 is a graph showing a nonlinear characteristic curve according to another embodiment of the present invention.
FIG. 7 is a block diagram illustrating an example of a conventional imaging apparatus.
[Explanation of symbols]
1 Lens 2 Iris 3 CCD
4 AGC circuit 5 A / D converter 6 Image memory 7 γ correction circuit 8 KNEE circuit 9 Signal processing circuit 10 Evaluation circuit 11 Histogram memory 12 Control circuit 13 AE circuit 14 Clock generation circuit

Claims (5)

A / D conversion means for converting an input signal into an n-bit digital signal;
Image storage means for storing an n-bit input signal converted by the A / D conversion means for each screen;
Knee correction means for performing knee correction processing on an n-bit input signal of one screen stored in the image storage means, converting it to an m (m <n) bit signal, and outputting
Control means for controlling the characteristics of knee correction processing of the knee correction means;
Evaluation means for evaluating in advance an n-bit input signal of one screen stored in the image storage means based on an n-bit input signal for one screen converted by the A / D conversion means;
The control means determines a slope of knee correction based on the evaluation result of the evaluation means , and further performs knee correction based on a maximum level signal value included in an n-bit input signal of one screen stored in the storage means. The knee correction means is characterized in that the knee correction means is characterized by the inclination and reference point determined by the control means for an n-bit input signal of one screen stored in the image storage means. An image signal processing apparatus characterized in that a knee correction process based on the above is performed, converted into an m (m <n) bit signal of one screen and output .   The image signal processing apparatus according to claim 1, wherein the evaluation unit performs evaluation using a histogram created based on information of an input signal for one screen. When the histogram of the evaluation means is a histogram with a higher appearance frequency of brightness data, the control means increases the slope of the knee correction, and the appearance frequency of the brightness data with a darker histogram of the evaluation means. The image signal processing apparatus according to claim 2, wherein when the histogram is a high histogram, the control unit reduces the slope of the knee correction . The control means determines an end point of a knee correction slope as the reference point based on a maximum level signal value included in an n-bit input signal of one screen stored in the storage means, and the knee correction means includes: The image signal processing apparatus according to claim 1, wherein knee correction processing is performed based on characteristics specified by an inclination and an end point determined by the control unit. An A / D conversion step of converting an input signal into an n-bit digital signal;
  An image storage step of storing the n-bit input signal converted by the A / D conversion step in an image storage unit for each screen;
  An evaluation step of evaluating an n-bit input signal for one screen stored in the image storage means based on an n-bit input signal for one screen converted by the A / D conversion means;
  The knee correction slope is determined based on the evaluation result of the evaluation step, and the knee correction slope reference point is based on the maximum level signal value contained in the n-bit input signal of one screen stored in the storage means. A determination step for determining
  The n-bit input signal of one screen stored in the image storage means is subjected to knee correction processing based on the characteristics determined by the inclination determined by the control means and the reference point, and m (m <m n) a knee correction step of converting to a bit signal and outputting the signal, and an image signal processing method.

Images