JPS6226635B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to color separation decoding of images, and more particularly to a color image information processing method for obtaining color information from color separation information. In color image reading, a plurality of solid-state imaging devices, such as CCD devices and BBD devices, in which 37.5 x 37.5 μm ² photoelectric conversion cells are densely integrated in a straight line or planar on a semiconductor chip, each have an image forming device. A lens projects the same point, line, or part of the image through an optical filter, and color separation image information for the same image position is obtained from each element. One of the problems is that each solid-state image sensor There are differences in reading position shift and reduction rate. Causes of reading position deviation include deviations in the mounting position of the imaging lens, deviations in the viewing direction, etc., and deviations in the mounting positions of each solid-state image sensor.The difference in reduction ratio is due to distance deviation and the light of the imaging lens. This is caused by a misalignment of the axis or a difference in focal length. In any case, if these deviations exist, moiré will appear differently or red ghosts will appear around black images when recording color images using the output image signals of each solid-state image sensor. or conversely, color shift occurs, such as black ghosts appearing around red images. An object of the present invention is to improve the above-mentioned color shift by processing read electrical signals. In order to achieve the above object, in the present invention,
Each color component image information is binarized,
Binary information indicating the presence or absence of each color component is divided into pixels corresponding to a small area on the screen, and the distribution pattern of the binarized information for a predetermined number of pixels consisting of a pixel of interest and a plurality of surrounding pixels around it is determined. The information held in the memory is given to a color encoder, and the color encoder determines that the binary information on both sides of the pixel of interest in the distribution pattern is different and that the binary information of the pixel of interest is different. When the information is different from both neighboring binary information, information for changing the binary information of a pixel of interest to one of the neighboring binary information is obtained, and the binary information is changed accordingly. As a result, the color component image information becomes binary information,
When there is a color shift of less than 2 units in the small area, that is, less than 2 pixels in the pixel unit, when focusing on the color shift pixel, the binary information on both sides of the pixel of interest in the distribution pattern is different. Moreover, since the binary information of the pixel of interest is different from the binary information on both sides, the color encoder obtains information for changing the color misaligned pixel to its neighboring binary information, and the two values of the color misaligned pixel are Since the value information is changed like that,
It becomes binary information without color shift. FIG. 1 shows the configuration of an electric circuit for implementing the present invention. In FIG. 1, 1 is a document, and its image (one line) is projected onto a first linear CCD element _4R via a red filter _2R and an imaging lens _3R , and a cyan filter _2C. and projected onto the second linear CCD element _4C via the lens _3C ,
CCD elements 4 _R and 4 _C output analog signals to amplifiers 5 _R and 5 _C by self-scanning, respectively. Analog signal levels are adjusted in amplifiers _5R and _5C . The output analog signals of amplifiers _5R and _5C are respectively input to sample and hold circuits _6R and _6C , where the signal level at a certain point during one pixel section is held for one pixel section,
The analog signal at a constant level during one pixel section passes through low-pass filters 7 _R and 7 _C to the binarization circuit 8.
Applied to _R and _8C . The binarization circuits 8 _R and 8 _C compare the input analog signal level with the reference voltages V _rR and V _rC , respectively, create binarized image signals R _io and C _io using these as thresholds, and send them to the discrimination circuit 9. Apply. The discrimination circuit 9 includes serial input-5-bit parallel output shift registers _SHR , _SHC , _{a PROM9RB} storing color data, D-type flip-flops F1 to F3, and an AND gate 9.
It consists of R and 9C, and the shift register SH _R
Binarized image signals R _io and C _io are applied to each of S C and S _C . FIG. 2 shows electrical signals from various parts of the electrical circuit shown in FIG. 1 for the image on the original document 1. In this embodiment, as shown in FIG. 2, the binarized reference voltage V _rC is set relatively high with respect to the image reading analog signal (output of 7 _C ) which has been subjected to the cyan filter 2 _C. As a result, the image information detection width (width of "1") of the binary image signal (output of 8 _R : R _io ) read with the red filter becomes the same as the binary image signal (width of "1") read with the cyan filter applied. 8 _C 's output: C _io )'s image information detection width is narrower by Δd at both edges. The correspondence between the original image color and R _io and C _io is shown in Table 1 below. R _io "1", C _io "0"
By considering it as white,

[Table] As long as the reading position difference or reading area deviation between solid-state image sensors 4 _R and 4 _C is within the range of Δd, R _pu
Color shift (ghost) is less likely to occur in red and black two-color recording using _t and B _put . The width of the "1" level output image signals R _put and B _put with respect to the image length is determined by the "1" level width of the binary image signal read through the cyan filter, so the output of V _rc and 7 _C By adjusting the relative level of , it is possible to accurately determine the output image signal width (ie, recording width) with respect to the image length. The "1" width of R _io has no bearing on this. Let me explain in a little more detail. The input signal (R
_io , C _io ) and set the output signals (R _put , B _put ) to positive logic, No. 2 in Table 1 is reading with red information, No. 4 is reading with black information, and No. 1 and 3
means that there is no red information or black information. In this signal processing, if we first focus on reading black information, there is a possibility that No. 2 (red) and No. 3 will appear around black (No. 4), but No. 3 will not be read as red. No problem. No. 2 (red) before or after black (No. 4)
The problem is color shift. However, as mentioned above, the "1" section of the image signal R _io read with the red filter is made wider than the "1" section of the image signal C _io read with the cyan filter. Even if R _io and C _io are originally misaligned due to positional misalignment in the optical system, as long as this misalignment is within the range where R _io is wider than C _io , it will be read as black because C _io is "1"
, red (R _io = 0, C _io =
1) will never occur. In other words, color shift of red around black does not occur. Next, focusing on red information reading, red (No. 2) is
Since only C _io is "1", the relative deviation between C _io and R _io has no bearing on the red reading. In other words, there is no color shift of black around red. If Δd is increased, the red recording area becomes wider, so Δd cannot be increased too much. If Δd is made small, color shift due to misalignment of the reading optical system etc. cannot be completely prevented. Therefore, in this embodiment, the present invention is further implemented to sufficiently prevent color misregistration. This will be explained next. In the discrimination circuit 9, shift registers SH _R and SH _C
The outputs of are respectively applied to the address input terminals of _PROM9RB . To explain the correlation between the address data of P ROM9 _RB and memory data, P ROM9 _RB contains 5 bits of R _io and 5 bits of C _io in time series.
Color data is determined for each distribution mode of "1". Some examples of that data are shown in the third section.
To explain with reference to Figures a to 3g, the 5-bit outputs R _io and 5 bits of shift registers _SHR and _SHC , respectively,
Focusing on the center bit of C _io as an image signal for converting color image information data, the color instructions shown in the third column of Figures 3a to 3g should be made from the data of R _io and C _io . However, when the image signal data of interest shows a color (red or black) that stands out by one pixel, it is set to the same color as the color of the subsequent pixel (referring to the temporal relationship of the reading scan), and When the color of the pixel rises by one pixel, the color data of the same color as the previous pixel is stored in an address determined by R _io 5 bits and C _io 5 bits. For example, the third
As shown in figure a, the boundary between two black pixels and two white pixels,
Pixels that can be determined to be black from R _io and C _io are stored as black, that is, R _put "0" and B _put "1", and as shown in Figure 3b, the switching boundaries of several pixels of the same color are also R
_io , C _io, memorize the color that can be determined from the 3rd c
As shown in Figures 3d and 3f, only one pixel of red next to black continuous, black of only one pixel next to red continuous, and black of only one pixel next to white continuous are all subsequent colors. The same color is used. As shown in FIGS. 3e and 3g, the white and red one-pixel arcs remain as they are. As already explained, in this embodiment, the "1" width of R _io is set wider than the "1" width of C _io due to the settings of V _{rR and} V _rC . It's easy to become.
Therefore, in FIGS. 3e to 3g, R _io
If it is "1" (Figure 3f), ignore it and R _io
In the case of "0" (Figures 3e and 3g), emphasis is placed on it. The cases shown in FIGS. 3c and 3d can mainly be considered as ghost colors, but in order to prevent this, the color data for the arcuate colors is changed to the adjacent color. Therefore, color shift (ghost) is less likely to occur. Each bit R _pu of read data of P ROM9 _RB
t and B _put are applied to the D inputs of flip-flops F1 and F2, respectively. F1 and F2
are both set when the D input is "1" and the clock pulse φ _D arrives, and its Q output becomes "1", and when the D input is "0" and the clock pulse φ _D arrives, it is reset and its output becomes "0". It is what it is. Flip-flop F3 also operates in the same way as F1 and F2, but a color image information output command is sent to its D input.
DV (output: "1") is applied and the clock input terminal
Clock φ _B to be applied to shift registers S _R and S _C is applied to CK. Note that φ _B is slightly ahead of φ _D in phase. flip flop F
3 is in the set state while the color image information output command DV is "1" and gives a gate-on signal "1" to the AND gates 9R and 9C. As a result, from the AND gate 9R, it is "1" only when the color image information is red, and "0" otherwise, and from the AND gate 9C, it is "1" only when the color image information is black, and it is "1" only when the color image information is black. In this case, color data of "0" is output. As described above, according to the present invention, when there is a color shift of less than 2 pixels in the binary information of color component image information (Figures 3c and 3d), when focusing on the color shift pixel, 2 pixels are detected. Value information distribution pattern (Figure 3c,
3d), the binary information on both sides of the pixel of interest is different, and the binary information of the pixel of interest is different from the two on both sides of the pixel of interest.
Since it is different from any of the value information, the semiconductor memory (9 _RB ), which is a color encoder, obtains information to change the color misaligned pixel to its neighboring binary information,
Based on this, the binary information of the color-shifted pixel is changed. In other words, color shift is corrected. Therefore, binary information without color shift can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of a circuit configuration for implementing the present invention, and FIG. 2 is a waveform diagram showing electrical signals of each part thereof. Figures 3a to 3g are P
FIG. 7 is a plan view showing color image signal distribution for explaining memory data of ROM9 _RB . 1: Original, 2 _R : Red filter, 2 _C : Cyan filter, 4 _R , 4 _C : Solid-state image sensor, 5 _R , 5 _C : Amplifier, 6 _R , 6 _C : Sample and hold circuit, 7 _R , 7
_C : Low-pass filter, 8 _R , 8 _C : Binarization circuit,
9: Discrimination circuit, _SHR , SH _C : Shift register (memory), 9 _RB : PROM (color encoder).

Claims (1)

[Scope of Claims] 1. Binary information indicating the presence or absence of each color component, obtained by binarizing each color component image information, is divided into correspondences with pixels of a small area on the screen, and the pixel of interest is centered on the pixel of interest. Information indicating the distribution pattern of the binary information of a predetermined number of pixels consisting of a plurality of surrounding pixels is held in a memory, the information held in the memory is given to a color encoder, and the color encoder calculates the distribution pattern of the target pixel of the distribution pattern. When the binary information on both sides are different and the binary information of the pixel of interest is different from both of the binary information on both sides, the binary information of the pixel of interest is changed to one of the binary information on both sides. A color image information processing method that obtains information to be changed and changes binary information accordingly. 2. In the binarization process, the generation period of at least one piece of color component image information with a color component for the same image length is made longer by a predetermined period of 2Δd than that of other pieces of color component image information. The color image information processing method according to item 1.