JP4090141B2 - Gradation correction method for imaging apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a gradation correction method for an image pickup apparatus used in a video camera or the like, and in particular, has improved gradation characteristics so that a natural-feeling image can be reproduced.
[0002]
[Prior art]
The gradation correction device of the imaging device corrects the video signal so that the luminance of the reproduced image corresponds to the luminance of the subject. Until now, in order to effectively use an image receiving apparatus having a small dynamic range, various ingenuities have been applied to this gradation correction.
[0003]
As shown in FIG. 4, a conventional image pickup apparatus including a gradation correction device includes an image pickup element 1 that alternately outputs two types of video signals having different exposure amounts within one field period, and before the video signal is amplified. A processing unit 2; an A / D converter 3 that converts an analog video signal into a digital signal; a time axis converter 4 that aligns and separates the time axes of two types of video signals with different exposure amounts; Level synthesis means 5 for synthesizing two kinds of video signals, histogram data detection means 6 for detecting a histogram of luminance of the synthesized image signal, a microcomputer 7 for calculating gradation characteristics based on this histogram, and a set Gradation correction means 8 for performing gradation correction of the composite image signal in accordance with the gradation characteristics, and the composite image signal after gradation correction output from the gradation correction means 8 is output by the main video signal processing circuit 9. It is physical, and output as a video signal.
[0004]
In this apparatus, the image pickup device 1 performs horizontal CCD transfer at twice the normal transfer rate, and two types of video signals, a video signal with a long exposure amount and a video signal with a short exposure amount, within one field period. Output alternately. The output video signal is subjected to pre-processing 2 such as CDS, AGC, etc., and then converted into a digital signal by the A / D converter 3, and the time axis converter 4 at a standard speed and at the same timing for a long time. The exposure image signal and the short-time exposure image signal are separated, and these signals are weighted and synthesized at a predetermined luminance level by the level synthesis means 5.
[0005]
The histogram data detection unit 6 detects a histogram of the composite image signal output from the level synthesis unit 5. In this detection, it is determined to which of the gradation levels divided in multiple stages the level of the composite image signal is a total of m × n pixels of horizontal m points and vertical n points in the screen. The histogram shows how many pixels having a certain gradation exist in the screen. The histogram data detection means 6 outputs the detected histogram data to the microcomputer 7.
[0006]
The microcomputer 7 uses the histogram data to calculate gradation characteristics according to the procedure described later, and sets the calculation result in the gradation correction means 8.
[0007]
The tone correction unit 8 performs tone correction on the synthesized image signal output from the level synthesis unit 5 so as to realize the tone characteristics set by the microcomputer 7 and outputs an image signal. This image signal is processed by the main video signal processing means 9 to become a video output.
[0008]
The microcomputer 7 uses the histogram data to obtain the gradation characteristics according to the procedure shown in FIG.
[0009]
The gradation characteristics are basically set so that the contrast is enhanced as the gradation level having a higher frequency in the histogram (that is, the gradation level having a higher luminance frequency). This characteristic is obtained by normalizing the histogram data of each gradation level and cumulatively integrating the normalized value of each gradation level. For this standardization,
Step 1: First, a total value Hall of histogram data H [n] (here, gradation level n is set to 16 levels from 0 to 15) for each gradation level is calculated.
Step 2: The histogram data H [n] for each gradation is divided by the total value Hall of the histogram data to calculate a normalized value Hk1 [n] for each gradation level.
[0010]
The normalized histogram data is cumulatively integrated from Hk1 [0] to Hk1 [n], and the cumulative integral value (ΣHk1 [n]) corresponding to the gradation Hk1 [n] of the input image signal is output. By doing so, it is possible to obtain gradation characteristics that enhance the contrast of gradations with high luminance frequency. However, if the tone correction is performed with the tone characteristics as it is, a slight tone difference in a tone with a high luminance frequency is emphasized more than necessary, and the S / N deteriorates, or a region with a low luminance frequency. The contrast may be lost, resulting in unnatural gradation reproduction.
[0011]
Therefore, in order to limit the slope of this gradation characteristic,
Step 3: When the normalized histogram data Hk1 [n] is smaller than the lower limit value C pmin of the slope, the lower limit value is set as new normalized histogram data Hk2 [n] (= C pmin ), Further, Hk1 [n] is, when the inclination is larger than the upper limit value C pma x is this new histogram data the upper limit value has been normalized Hk2 [n] (= C pmax ). When C pmin ≦ Hk1 [n] ≦ C pmax , use Hk1 [n] as it is for Hk2 [n]
Step 4: Gradation characteristics are obtained by cumulative integration of the histogram data Hk2 [n] after tilt restriction.
[0012]
further,
Step 5: Finally, normalization processing for reproducing the peak value KI of the input image is performed.
[0013]
This last normalization process normalizes the maximum value of the video to the upper limit value of the dynamic range in the signal processing circuit after the gradation correction process.
[0014]
The tone characteristics thus obtained enable tone correction in which the contrast is enhanced as the tone level of the histogram is higher. In addition, it is possible to prevent a situation in which the maximum value of the image is out of the dynamic range of the subsequent signal processing circuit due to the gradation correction.
[0015]
[Problems to be solved by the invention]
However, with the conventional tone correction method, the tone characteristics are standardized by the upper limit value of the dynamic range of the signal processing circuit after tone correction processing, regardless of what subject is captured. Therefore, even when a dark subject that does not cause saturation of the dynamic range is captured, correction that raises the gradation unnecessarily is performed, resulting in an unnatural image. Had.
[0016]
The present invention solves such a conventional problem, and effectively uses the dynamic range of the main signal processing circuit and the monitor after the gradation correction processing and also for a low-luminance subject image. An object of the present invention is to provide a gradation correction method that enables gradation correction to a simple image.
[0017]
[Means for Solving the Problems]
Therefore, in the gradation correction method of the present invention, the maximum gradation value of the histogram data is detected and the gradation characteristics corresponding to the maximum gradation value are calculated.
[0018]
Therefore, it is possible to perform optimum gradation correction in accordance with various subjects, and natural images can be reproduced even for low-luminance subjects while effectively utilizing the dynamic range of the subsequent signal processing.
[0019]
DETAILED DESCRIPTION OF THE INVENTION
According to the first aspect of the present invention, a sum of histogram data of each gradation of an input image is obtained, a normalized value obtained by dividing the histogram data of each gradation by the sum is obtained, and the normalized value is a gradation. When the slope is smaller than the lower limit of the slope that limits the slope of the characteristic, the normalized value is replaced with the lower limit of the slope, and when the normalized value is larger than the upper limit of the slope that limits the slope of the gradation characteristic, The standardized value is replaced with the upper limit value of the slope, and the standardized value is cumulatively integrated up to each gradation to obtain a cumulative integrated value for each gradation, and the gradation of the input image is determined according to the difference between the cumulative integrated values. In the gradation correction method for performing correction, when the maximum gradation value of the histogram data is detected and the maximum gradation value is within a standard level, the upper limit value of the slope and the lower limit value of the slope are set to be equal values. , The maximum gradation value is The slope upper limit value is greater than the slope lower limit value when it is greater than the quasi-level and within the dynamic range of the post-processing circuit, and the slope is also greater when the maximum gradation value is greater than the dynamic range of the post-processing circuit. Is set to be larger than the lower limit value of the slope. By using the gradation characteristic thus obtained, gradation correction according to the luminance of the subject can be performed.
[0020]
According to the second aspect of the present invention, the sum of histogram data of each gradation of the input image is obtained, a normalized value obtained by dividing the histogram data of each gradation by the sum is obtained, and the normalized value is an inclination of the gradation characteristic. When the normalized value is smaller than the lower limit value of the slope that limits the slope, the normalized value is replaced with the lower limit value of the slope, and when the normalized value is larger than the upper limit value of the slope that limits the slope of the gradation characteristic, the normalized value Is replaced with the upper limit value of the slope, the normalized value is cumulatively integrated up to each gradation, a cumulative integrated value is obtained for each gradation, and the gradation of the input image is corrected according to the difference between the cumulative integrated values of the gradations. In the gradation correction method, the maximum gradation value of the histogram data is detected, and the maximum gradation value of the histogram after gradation correction when the maximum gradation value is not less than a standard level and within the dynamic range of the post-processing circuit. Enter When the maximum gradation value is larger than the dynamic range of the post-processing circuit, the maximum gradation value of the histogram after gradation correction is set as the upper limit of the dynamic range of the post-processing circuit. In this way, the maximum gradation value of the histogram after gradation correction is changed. By using the gradation characteristic thus obtained, gradation correction according to the luminance of the subject can be performed. .
[0023]
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0024]
The gradation correction method of the present invention is implemented by the gradation correction apparatus of the imaging apparatus shown in FIG. The configuration of this apparatus is as described above.
[0025]
The image sensor 1 of this apparatus performs horizontal CCD transfer at twice the normal transfer rate, and two types of video signals, a video signal with a long exposure amount and a video signal with a short exposure amount, within one field period. Output alternately. The output video signal is subjected to pre-processing 2 such as CDS, AGC, etc., and then converted into a digital signal by the A / D converter 3, and the time axis converter 4 at a standard speed and at the same timing for a long time. The exposure image signal and the short-time exposure image signal are separated.
[0026]
The level synthesizing unit 5 synthesizes these signals with a predetermined luminance level. As shown in FIG. 5, the composition is performed by defining a certain composition level and composition width. For the long-exposure image signal Ylong, a signal K × Ylong multiplied by a gain K for starting compression from the synthesis level or higher is created, and for the short-exposure image signal Yshort, an offset value Yoffset is added. After that, a signal (1-K) × (Yshort + Yoffset) multiplied by a gain (1-K) for compression at a synthesis level or lower is created, and these are added to obtain a synthesized signal Ymix.
[0027]
The histogram detection means 6 detects the histogram data of the image of this synthesized signal. In this detection, the gradation level is divided into 16 gradations (a sufficient characteristic can be obtained with this number of divisions), and the level of the composite signal Ymix at the total m × n pixels of horizontal m points and vertical n points on the screen. Which gradation level belongs is determined, and histogram data indicating how many pixels having a certain gradation exist in the screen is obtained.
[0028]
The microcomputer 7 uses the histogram data to calculate the gradation characteristics using the algorithm shown in FIG. The procedure of this algorithm is basically the same as that of the conventional one (FIG. 6), and “normalization of histogram data”, “slope limitation processing”, “cumulative integration using histogram data subjected to slope limitation processing” and “ The gradation characteristics are calculated in the flow of “normalization according to the maximum gradation value”. However, the difference from the prior art is that the maximum value of the histogram data is obtained, and the “inclination limiting process” and the “normalization by the maximum gradation value” process are different depending on this maximum value. In this embodiment, the maximum value of the histogram data is
(1) When it is within the standard level (2) When it is above the standard level and within the dynamic range of the post-processing system circuit (3) When it is above the dynamic range of the post-processing system circuit Accordingly, the “inclination limiting process” and the “normalization by maximum gradation value” process are different.
[0029]
The procedure of FIG. 3 will be described in detail.
[0030]
Step 11: First, a total value Hall of histogram data of all gradations is calculated. If the histogram data of the nth gradation is H [n], Hall is given by (Equation 1).
[0031]
Hall = ΣH [n] (n = 0 to 15) (Equation 1)
Step 12: At this time, H [n] is sequentially compared with zero from n = 0 to 15, and finally n where H [n] ≠ 0 is held as the maximum gradation value Nmax.
[0032]
Step 13: Next, according to (Equation 2), the histogram data of each gradation is divided by the total value Hall to calculate the first normalized value Hk1 [n].
[0033]
Hk1 [n] = H [n] / Hall (Equation 2)
FIG. 2 shows an example of a histogram when the maximum luminance level of the subject is in each of the cases (1) to (3) described above. In the upper part of FIG. 2, an example of histogram data subjected to the first normalization is shown as a bar graph. The horizontal axis represents the gradation level n (= 0 to 15), and the vertical axis represents the normalized value Hk1 [n]. The normalized histogram data is cumulatively integrated for each gradation from Hk1 [0] to Hk1 [15], thereby obtaining gradation characteristics indicated by a broken line in the upper part of FIG. That is, the cumulative integral value Hks1 [n] is expressed as (Equation 3) using Hk1 [n].
[0034]
Hks1 [n] = ΣHk1 [k] (k = 0 to n) (Equation 3)
At this time, the vertical axis of the upper diagram in FIG. 2 represents Hks1 [n].
[0035]
As described above, the gradation characteristics represented by the broken line can selectively improve the contrast with a high luminance frequency, but can improve the contrast. The difference is emphasized more than necessary, the S / N deteriorates, or the contrast at the gradation with a low luminance frequency becomes too small, and the gradation reproduction is unnatural. Therefore, in order to limit the gradient of the gradation characteristics,
Step 14: If Hk1 [n] is smaller than the lower limit value C pmin of the slope, Hk2 [n] = C pmin is set. If Hk1 [n] is larger than the upper limit value C pmax of the slope, Hk2 [n] = C pmax . At this time, the inclination limit values C pmin and C pmax are changed according to which of the above-mentioned (1) to (3) the maximum gradation value Nmax belongs to.
[0036]
In this embodiment, in the case of (1) is a limiting value C pmin = C pma x, also (2) (3) In the case of, different values for the limit value C pmin and C pmax Thus, the displacement width of the gradation characteristic is defined.
[0037]
Step 15: The cumulative integral value Hks2 [n] is calculated by (Equation 4) using the histogram data Hk2 subjected to this inclination limiting process.
[0038]
Hks2 [n] = ΣHk2 [k] (k = 0 to n) (Equation 4)
The histogram data Hk2 on which the slope restriction processing is performed with respect to the histogram characteristics in the upper part of FIG. 2 is a bar graph in the lower part of FIG. This is indicated by a broken line. When (1) “the maximum gradation value is equal to or less than the standard level”, the input value and the output value are the same, and gradation correction is not performed.
[0039]
Step 16: Finally, a second normalization process for reproducing the peak value of the input image is performed. In this second standardization process,
Hksk2 [n] = Hks2 [n] x Kl (Equation 5)
Thus, the normalized value Hksk2 [n] is obtained. At this time, the value of Kl is changed according to which of the above-mentioned (1) to (3) the maximum gradation value Nmax belongs to.
[0040]
In this embodiment, in the case of (1), Kl = 1, and in the case of (2), the maximum gradation value of the gradation correction output signal is the same as the maximum gradation value of the input signal. The value of Kl is set as follows. In the case of (3), the value of Kl is set so that the maximum gradation value of the gradation correction output signal becomes the upper limit of the dynamic range of the post-processing system circuit. When the second normalization is performed, the gradation characteristic Hksk indicated by the one-dot chain line in the lower part of FIG. 2 is obtained.
[0041]
FIG. 1 shows the gradation characteristics in the cases (1), (2) and (3) thus obtained. The lower part of FIG. 1 shows histogram data in each case when the Ymix signal is used as the gradation correction input, and the upper part of FIG. 1 shows the gradation characteristic calculated from the histogram characteristic chart of the lower part.
[0042]
The horizontal axis of this figure shows the luminance of the composite signal Ymix for gradation correction input in 16 gradations, with the lower 8 gradations being the long exposure signal Ylong and the upper 8 gradations being the short exposure signal Yshort.
[0043]
The standard level that divides between (1) and (2) is set to about half of the long-time exposure signal. In this example, the standard level is set to the third gradation in the input gradation value. Further, the level that is “within the dynamic range of the post-processing system circuit” that divides between (2) and (3) is the fifth gradation in the input gradation value. Further, it is preferable to set an optimum value for the gradient limit value for each gradation.
[0044]
With the above algorithm, gradation characteristics corresponding to the luminance of the subject can be obtained. By setting this calculation result in the gradation correction unit 8, the gradation correction unit 8 can perform optimal gradation correction on the synthesized signal in accordance with the luminance of the subject. That is, gradation correction is not performed on a subject that does not have a high luminance portion, and the gradation of the high luminance portion is enhanced as the level of the high luminance portion increases. Also, if the high luminance level exceeds the dynamic range of the subsequent signal processing, the output signal falls outside the dynamic range by standardizing the maximum value of the combined luminance signal to the maximum value determined by the dynamic range of the subsequent signal processing. Can be prevented, and the dynamic range of the subsequent signal processing can be effectively utilized.
[0045]
As a result, an optimum gradation correction effect can be obtained for any subject.
[0046]
【The invention's effect】
As is apparent from the above description, in the gradation correction method of the present invention, the maximum standardized value of the gradation correction calculation is changed according to the input image, so that the dynamics of the video signal processing circuit can be applied to any subject. By effectively utilizing the range and maintaining the gradation reproducibility of the image in the high luminance part, it is possible to obtain natural gradation reproduction in which the S / N deterioration is not noticeable in the low luminance image.
[Brief description of the drawings]
FIG. 1 is a gradation characteristic diagram obtained by a gradation correction method according to an embodiment of the present invention;
FIG. 2 is an explanatory diagram showing a procedure for obtaining a gradation characteristic diagram by the gradation correction method of the embodiment;
FIG. 3 is a flowchart showing the procedure of the gradation correction method of the present invention;
FIG. 4 is a configuration diagram of a solid-state imaging device including a gradation correction device;
FIG. 5 is a conceptual diagram showing a luminance signal synthesis method;
FIG. 6 is a flowchart showing a procedure of a conventional gradation correction method.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Image pick-up element 2 Pre-processing part 3 A / D converter 4 Time-axis converter 5 Synthesis | combination means 6 Histogram detection means 7 Microcomputer 8 Tone correction means 9 Main video signal processing means

Claims (2)

Find the sum of the histogram data of each gradation of the input image,
Obtain a normalized value obtained by dividing the histogram data of each gradation by the sum,
When the normalized value is smaller than the lower limit value of the slope that limits the gradient of the gradation characteristics, the normalized value is replaced with the lower limit value of the slope,
When the normalized value is larger than the upper limit value of the slope that limits the slope of the gradation characteristic, the normalized value is replaced with the upper limit value of the slope,
A cumulative integration value for each gradation obtained by accumulating the normalized value to each gradation is obtained,
In a gradation correction method for performing gradation correction of an input image according to a difference between gradations of the cumulative integral value,
Detecting the maximum gradation value of the histogram data;
When the maximum gradation value is within the standard level, the upper limit value of the slope and the lower limit value of the slope are equal values,
When the maximum gradation value is larger than the standard level and within the dynamic range of the post-processing circuit, the upper limit value of the slope is made larger than the lower limit value of the slope,
The gradation correction method, wherein the upper limit value of the slope is made larger than the lower limit value of the slope even when the maximum gradation value is larger than the dynamic range of the post-processing circuit. Find the sum of the histogram data of each gradation of the input image,
Obtain a normalized value obtained by dividing the histogram data of each gradation by the sum,
When the normalized value is smaller than the lower limit value of the slope that limits the gradient of the gradation characteristics, the normalized value is replaced with the lower limit value of the slope,
When the normalized value is larger than the upper limit value of the slope that limits the slope of the gradation characteristic, the normalized value is replaced with the upper limit value of the slope,
A cumulative integration value for each gradation obtained by accumulating the normalized value to each gradation is obtained,
In a gradation correction method for performing gradation correction of an input image according to a difference between gradations of the cumulative integral value,
Detecting the maximum gradation value of the histogram data;
When the maximum gradation value is not less than the standard level and within the dynamic range of the post-processing circuit, the maximum gradation value of the histogram after gradation correction is made equal to the maximum gradation value of the input signal, and the maximum gradation value is If it is larger than the dynamic range of the post-processing circuit, the maximum gradation value of the histogram after gradation correction is set so that the maximum gradation value of the histogram after gradation correction becomes the upper limit of the dynamic range of the post-processing circuit. A gradation correction method characterized by changing.

Images