JPS6122514B2 - - Google Patents

Description

[Detailed description of the invention]

For example, there is a CCD as a solid-state image sensor, which can be roughly divided into frame transfer type and interline transfer type.
In either type, the charge obtained by the imaging section is supplied in parallel to a horizontal register, which is read out serially at a horizontal scanning speed. By the way, when charges are read out from the horizontal register in series, that is, when charges are transferred to each bit of the horizontal register, untransferred charges are generated. For this reason, when a color-separated image of an object is projected onto a single CCD to obtain, for example, red, green, and blue point-sequential signals, color mixture occurs due to the untransferred charges. The present invention attempts to remove color mixture components from color signals that cause such color mixture. First, let's consider CCD color mixing. Now, for simplicity, we will use a CCD as shown in Figure 1.
1 is composed of an imaging section 2 with a horizontal scanning speed of 1, and a horizontal register 3, and in front of each bit (unit imaging section) of the imaging section 2, there is a filter R that transmits, for example, red light and green light, respectively. Suppose that G is arranged alternately. It is assumed that green light is supplied to such a CCD 1 with uniform reference, and then the charge obtained in the imaging section 2 by this green light is transferred to the horizontal register 3 (at this time, transfer efficiency is assumed to be 100%). Then, in this state, the level of charge in the register 3 is shown as in the column [OT] in FIG. Then, in the register 3, when the charge is read out for each read pulse (transfer pulse),
Assuming that the transfer efficiency η per one bit is 100%, the charge in the register 3 changes as shown in FIG. 2 for each read pulse. However, [T] indicates one cycle of the read pulse, that is, one bit is read from the register 3 every 1T. However, when the transfer efficiency η is not 100%, if the non-transfer efficiency is ε(=1−η), then
The change in charge in register 3 is as shown in Figure 3 (the charge in the last bit of register 3 is
(assumed to be read at η = 100% by the output gate). In other words, if the green signal and red signal are R and G, then in 2nT time, At time (2n+1)T, becomes. However, 2n is the number of bits in register 3,
Or it may be the number of transfer stages. Therefore, if the color mixing rate is Δ=R/G, it becomes as shown by the solid line in Figure 4, and when the non-transfer efficiency ε is, for example, 10 ^-4 or less, Δ=2nε/1-2nε 〓2nε It can be expressed as Moreover, as shown by the broken line in FIG.
Assuming that the axes of G and B (B is a blue signal) are shifted by 120°, and the phase shift of signal G due to color mixing is θ, then θ=tan ^-1 Rγsin60°/Gγ−Rγcos6
0° γ: Gamma correction value, as shown in FIG. Practically, θ≦5°. From the following, the following can be said (hereinafter, N=2n). Color mixture that occurs in a practical range of the transfer efficiency η and the number of transfer stages (number of bits) N, for example, ε≦10 ^-4 and N≦500, affects only the signal after one bit. The color mixing rate Δ is Δ=Nε. The present invention focuses on these points and removes color mixture. That is, a dot-sequential signal having color signals Si (i=1 to m, N≫m) dot-sequentially is transferred through the register 3 by N bits and mixed, and as a result, the signal Si becomes the signal
Suppose that it is taken out as Si. Then, since the signal Si' is a linear combination of the signals Si from above, the matrix connecting the signals Si and Si' is a color mixing matrix F=(fij)
Then, the following formula is obtained.

[Table] According to equation (i), the signal Si' is a linear combination of the signals Si, so the signal Si can be obtained from the inverse matrix F ^-1 of the matrix F and the signal Si'. , that is, It is. Therefore, when calculating the matrix F ^-1 ,
It becomes as follows (proof is omitted). F ^-1 = (gij) = (g′ij)/｜F｜

[Table] |F|=(1-Δ) ^m −(-Δ) ^m (i, j=1,2...m) And if Δ≪1,

[Table] becomes. That is, the signal Si is, for example, the signals R, B, G,
For a point sequential signal with this order, Next door or,

[Table] = 〓
becomes. FIG. 7 shows an example of the present invention in which color mixture is removed based on equation (iii). That is, the image of the subject 5 is projected onto the color separation filter 7 by the imaging lens 6 to form a color separated image of the subject 5, which is then supplied to the CCD 8.
For example, dot-sequential signals R', B', and G' are taken out from the CCD 8. In this case, the signals R′ to G′ are
It is in phase in any horizontal scanning period, has a mixed color component as described above, and is expressed by equation (ii). Then, the point sequential signals from the CCD 8 are supplied to sampling and hold circuits 11R to 11G, and signals R' to G' are respectively taken out and synchronized, and the synchronized signals R' to G' are
The signals are supplied to level correction circuits (amplifiers) 13R and 13G through amplifiers 12R to 12G, and are converted into signals (1+Δ)R' to (1+Δ)G' with a level of (1+Δ), and these level-corrected signals are sent to the subtraction circuit 14.
Supplied to R~14G. In addition, signals R'~ from amplifiers 12R~12G
G' is supplied to attenuators 15R to 15G to produce signals ΔR' to ΔG' having a level of Δ, and these signals are supplied to subtraction circuits 14B to 14R. Then, in the subtraction circuits 14R to 14G, subtraction shown by equation (iii) is performed, and therefore the subtraction circuits 14R to 1
4G provides signals R, B, and G without color mixture,
This is taken out to output terminals 16R to 16G. In this way, according to the present invention, even if charge remains untransferred in the horizontal register of the CCD 8 and color mixture occurs in its output signal, the color mixture can be removed. In the examples shown in FIGS. 8 and 9, the level correction circuits 13R to 13G are omitted. That is, in the example of FIG.
The signal R' from the attenuator 15R is supplied to the subtraction circuit 14R, and the signal ΔR' from the attenuator 15R is supplied to the subtraction circuit 14R to become the signal (1+Δ)R'.
G' is supplied to a subtraction circuit 14R and is made into a signal R. Further, the signals B' and G' are processed in the same manner to become the signals B and G. In the example shown in FIG. 9, the color mixing rate Δ is sufficiently small and is assumed to be 1+Δ=1. Furthermore, considering the horizontal register of CCD8, the number of signal (charge) transfer stages N is small on the left side of the screen, and large on the right side, and since Δ=Nε,
Strictly speaking, the gains of the level correction circuits 13R to 13G and the attenuators 15R to 15G may be changed depending on the screen position, that is, the horizontal scanning position, but in that case, the sawtooth wave signal is changed from the horizontal synchronizing pulse. With this signal, circuits 13R to 13G, 15
What is necessary is to control the gain of R to 15G. In addition, in the CCD8, the charge obtained in the imaging section is transferred in the vertical direction and supplied to the horizontal register, so if there is a problem with the remaining bits being transferred in the vertical direction, the same applies as described above. It is sufficient to perform signal processing.

[Brief explanation of the drawing]

1 to 5 and 6 are diagrams for explaining the present invention, and FIGS. 7 to 9 are system diagrams of an example of the present invention, respectively. 8 is a CCD, 11R to 11G are sampling and hold circuits, 13R to 13G are level correction circuits,
15R to 15G are attenuators.

Claims (1)

[Claims] 1 Multiple color signals S ₁ output from the solid-state image sensor
The plurality of color signals S _{1 -}
Synchronize Sm and this synchronized multiple color signals
Any color signal Sn from S ₁ to Sm (however, 1≦
n≦m), at least 1 of this color signal Sn
A color mixture removal circuit that removes color mixture components contained in the plurality of simultaneous color signals S ₁ to Sm by subtracting a color signal obtained before sampling by a level ratio that depends on a color mixture rate.