JPS6129275A - Method of correcting half tone picture - Google Patents

Abstract

PURPOSE:To minimize round-off errors and to correct a picture signal precisely by correcting the picture signal with the aid of correction data collected accurately in the previous stage of an A/D converter. CONSTITUTION:A picture signal for including gradation is digitally converted by an A/D converter, and digital picture data is obtained. To attain this purpose, said picture signal is magnified larger than a normal picture signal by a programmable analog amplifier 70 in a previous stage of an A/D converter 71, when correction data on external light conditions at the time of inputting a picture and that on the light-receiving element dispersion, etc., are collected. Then the magnified signal is inputted to the A/D converter 71, and the correction data is accurately collected. Since the normal picture signal is corrected by said correction data, round-off errors can be minimzed compared with the digitally converted one handled by the same analog gains as those of the normal picture signal.

Description

DETAILED DESCRIPTION OF THE INVENTION FIELD OF INDUSTRIAL APPLICATION 1 The present invention relates to a method for processing image data including gradations, and for correcting conditions such as H at the time of input thereof.

Structure of the Work Example and Its Problems Conventionally, microcomputer-applied equipment such as office automation equipment has a small capacity of recording element or medium, so many processes have been limited to those that handle encoded data. However, recently,
With the development of large-capacity recording media such as large-capacity storage elements and optical disks, it has become possible to handle so-called non-encoded data (for example, hand-drawn drawings, arbitrary drawings, etc.). Therefore, document shapes can be displayed in high resolution (similar to facsimile).
We also provide a document file system that stores thousands to 11 documents on one optical disk and retrieves them as needed. It became possible.

Another request is to record diagrams and photographs with gradation, and there is also a request for color images. By the way, in the case of black and white binary information such as documents and figures, it is sufficient to determine whether it is white or black near the intermediate value of the read analog signal, and it is relatively easy to read errors due to uneven light source, uneven light receiving element, etc. The goal was achieved with some corrections.

However, in the case of an image with gradation, the unevenness immediately becomes gradation unevenness υ, which does not represent the true gradation.

Figure 1 is a diagram showing a conventional example, in which (1) is a line sensor, (2) is a buffer amplifier that receives the output of line sensor (1), and (3) is a digital output of the buffer amplifier (2). A/D converter that converts into a value, (4) is an A/D
It shows the output terminal group of the converter (3), (5) represents the original to be read, (6) and (7) represent the movement direction, (
8) is a lamp that illuminates the original (5), and (9) is a lamp that illuminates the original (5).
) onto the line sensor (1) t;
(10 is a window shown as a representative of outside light, and is a source of outside light given to the original (5). (B) is an arrow indicating outside light.

In operation, when the original (5) is illuminated with a lamp (8) and an image is formed by the line sensor (1) through the lens (9), the original (5) is illuminated from the line sensor (1). An outside signal corresponding to the gradation of the line shown in t is obtained. By receiving the output of the line sensor (1) at the buffer amplifier (2) and inputting it to the A/D converter (3), encoded gradation information is obtained at the output terminal group (4). In this state, by physically moving the line sensor (1) in the direction of the arrow (7), the original (fi) j:
6) When the reading line moves in the direction (6), the original document (5) t is scanned and read. In this state,
The number of dots on the line sensor (1) represents the total number of dots on one line, and the amount of movement of the line sensor (1) in units of lines represents the reading density in the vertical direction of the original (5). For example, set the width of manuscript (5) to 216 wm.
When the number of dots in the line sensor (υ is 1728), a resolution of approximately 8 Pel/kaku can be obtained. At this time, the lens (9) covers the width of 216 m of the original 1iiiIi (6) of the line sensor (1). A magnification is required to image a length corresponding to 1728 dots.

Next, the development to be corrected will be explained. FIG. 2 shows the phenomenon to be corrected as seen in the output signal of the line sensor (1). The line sensor (1) is, so to speak, an IC and is manufactured through a complicated semiconductor process, but some bit defects occur. This dot defect causes an extremely different output at a particular bit, as shown in Figure 2, when reading a completely white manuscript. The horizontal axis in FIG. 2 represents the elements, and the vertical axis represents the output level. FIG. 2 α') is an enlarged view of the output signal (4), showing that the level differs for each bit. This bit defect is the first
When reading with a groove formation like this, vertical stripes appear on the read data, as shown in Figure 2 (4).The primary problem is dark buttocks. This is due to variations in flow. This is because sensors such as CCDs are made of semiconductors, so there is a current that flows even when no light hits them. This is also a cause of vertical stripes due to bit-by-bit variation. Figure 2 (6) shows the output signal of the line sensor (1) when the lens (9) is covered. (B') is an enlarged view of the output signal 03). In addition, uneven illumination of the document (5) is another distortion that must be corrected. For example, if a lamp (8) is used to illuminate the area near the center, circular ring-shaped illumination irregularities like contour lines on a map will occur starting from the area near the lamp (8). Although it is not noticeable to the human eye, the output signal is proportional to the amount of light, so it clearly appears in the data, and the data has a circumferentially darkened periphery. Particularly in this case, it is not sufficient to perform line-by-line correction alone; it is also necessary to collect data for the vertical direction of the document. This situation is shown in FIG. 2(C). This corresponds to uneven illumination on the document surface, but there is also data Iζ for external light other than illumination lamps.
This causes distortion. This is shown in FIG. 2 (2). Now, in order to collect accurate data while dealing with various factors that cause distortion, the following corrections have been conventionally performed.

To correct the above-mentioned variation in ψ, place a white original, turn on the light, shift the focal point of the lens to make the light hitting the line sensor (1) uniform, and then adjust the output at that time. is A/D converted and stored bit by bit.

Then, the average value from the obtained data string for 1 line and 2 minutes is taken as 1.0, and the reciprocal of the data for each bit is taken to obtain a correction value for each bit. Next, the lights are turned off and the lens (9) is covered to prevent light from hitting the line sensor (1), and the output at that time is A/D converted to collect data for one line. This is taken as the dark current correction value. In order to measure illumination unevenness and external light unevenness, place a white document with no unevenness, focus, scan the document, collect autumnal equinox data, and use the reciprocal of this as a correction value. After performing the above operations, place the original document to be taken, collect all data, and perform correction processing. In other words, the data read from the original is subtracted by the dark current correction value, multiplied by the bit defect data correction value, and further multiplied by the illumination unevenness and external light unevenness correction values. data can be obtained. This situation is shown in M8 meat. If the original is white, the scanned data will contain all of the unevenness and variation mentioned above. Show this to the prisoner.

The bit defect correction value obtained in advance is shown in (6). Bit defect correction is performed by multiplying these values. This is shown in FIG. 8(C). Next, set the illumination unevenness correction value (2)
Shown below. Illumination unevenness is removed by further multiplication by this illumination unevenness correction value. This is shown in @). Similarly (F
By multiplying by the outside light unevenness correction value shown in ), all unevenness is corrected, and white data as shown in expansion) is obtained. As shown in t below, the current state of the art is that all corrections are performed by performing arithmetic processing on A/D converted data values.

However, although the data 1 after the A/D conversion value is subjected to a correction calculation process, the problem here is rounding errors and This is a problem of numerical expression level range.

That is, if the number of representation bits of A/D converted encoded data is infinite, the correction will be performed completely correctly. However, since processors and other devices only process data using a limited number of bits, calculation errors accumulate. In addition, although the correction value itself has only a small value, it is expressed in a limited number of bits by AID conversion, so small values are rounded up and information is lost, making it difficult to perform accurate correction. The current situation is that this cannot be achieved.

OBJECTS OF THE INVENTION It is an object of the present invention to provide a method for correcting halftone images that can minimize rounding errors such as those described in E and can perform accurate correction.

Structure of the Invention In the halftone image correction method of the present invention, when an image signal including gradations is digitally converted by an A/D converter to obtain digital image data, correction data at the time of inputting the image include: In the preceding stage of the A/D converter, the image signal is enlarged more than the normal image signal and inputted to the A/D converter to collect correction data with high accuracy, and the normal image signal is corrected with this correction data. It is characterized by

DESCRIPTION OF EMBODIMENTS The halftone image correction method of the present invention will be described below based on a specific embodiment. FIG. 4 shows the configuration of an analog amplifier provided before the A/D converter (3). - is an input terminal to which the output of sensor (1) is added. The symbol indicates the gain setting register. This register is programmable and controlled in some way by the microprocessor. - indicates an operational amplifier. - is a resistance group that determines the amplification factor of the operational amplifier. Each switch is connected in series with a pair of resistors. This switch is controlled by the set value obtained in the gain qualification register (2), and when one of the switches is turned on, the corresponding resistance in the resistor group becomes the return resistor of the operational amplifier. - is a resistor interposed between the input terminal - and the inverting terminal of the operational amplifier. If the resistance value of this resistor is R, and the resistance values of the resistor group are R from t, '2R, 4R, 8R, and 16R, then the amplification factor of the operational amplifier is as well known. It is expressed as the ratio of the resistance value of and the resistance value of the input side, and is 1
．． 2.4゜8.16...---...When one of the switches is turned on, one of the amplification factors listed in L is selected. Therefore, the information set in the register ) You can specify the amplification factor by turning on one of the switches. (641 is a D/A converter, and - is an offset setting register that provides digital data to the D/A converter. This register has a digital value that provides an offset voltage to the non-inverting input terminal of the operational amplifier. - is the output terminal of the operational amplifier -.

Now, suppose that the input voltage at the input terminal is Vins, the voltage at the output terminal is Vout, and the output voltage of the D/A converter is V, then these relationships are expressed by the following equation.

Vout = −G Vin −1(1+G)V+
-■Here, G is R/R, 2R/R94R/R08R/
R916R/R...--Represented by 1.2.4,8
．． 16..., and one of them is set by the register.

Now, in order to collect correction data with high accuracy, the following steps should be taken. For example, to collect the bit defect data shown in FIG. 2, it is sufficient to enlarge the bit defect portion. Fig. 6(a) and Fig. 2) are shown again to show its details.In other words, considering that the changing part due to bit defect I has a voltage change range of +V and -■, the overall average value voltage is Va. shall be. Therefore, considering the input voltage as Va±vl and substituting it into the first equation, the output voltage is: vout = -G (Va fVl) + (1,000G
) V4-=±GV1- GVa+(1+G)V...
・・・・・・・・・・・・・・・ ■. Looking at this second equation, it is possible to set the second term and subsequent terms to zero by appropriately giving +. That is, V+=(G
/1+G) Set as Va: Koyoshi, vout = + GV, 1 ---
−− ■. This shows that it is possible to cancel the average level of the input voltage and magnify only the changing part. However, the step of measuring the average level value is necessary. For example, set ■+ to 0 and input the input signal to A/D converter (3) F with gain G=1.
Import data into fI**#9#Li';fo-hf', etc., and measure Van Vl. By measuring vl, the approximate value of G is determined and Va is determined. (G/ 1 + G) Va is determined, and each G
and V+ in the register) and then the operational amplifier - outputs ±Gv1, which is input to A/
By performing D conversion, υ±v1 can be expanded to the range limit of the A/D converter (3) and A/D conversion can be performed.

FIG. 6 shows a specific control example of FIG. 4, and (to) shows the programmable analog amplifier shown in FIG. 4. The register shown in FIG. 4 is omitted in FIG. 5, and a (71)' indicated by G and 0 indicates an input terminal corresponding to the input terminal in FIG. 4). (71) is an A/D converter connected to a pass line of a microprocessor or the like. (72)
indicates a pass line. The register of the programmable amplifier (7) is also connected to the pass line (72). The pass line (7) exchanges data to the processor. All calculations can be created by a program.Ordinary image signals are corrected using correction data obtained from data captured with high precision in this way.

As described in the detailed description of the invention, the halftone image correction method of the present invention involves converting an image signal including gradations into digital data using an A/D converter to obtain digital image data. When recording correction data for light conditions and correction data for light-receiving element variations, etc., the image signal is magnified compared to the normal image signal and input to the A/D converter at the stage before the A/D conversion. Since correction data is collected with high precision and this correction data is used to correct the normal image C' It is possible to suppress the rounding error to the minimum compared to a digital signal obtained with the same analog gain as the signal, and to apply accurate correction to the normal image signal. be.
−

[Brief explanation of the drawing]

Fig. 1 is a block diagram of a conventional device when inputting medium-tuned image expansion data, Fig. 2 is an explanatory diagram of the phenomenon to be corrected, Fig. 8 is an explanatory diagram of the correction method, and Fig. 4 is a concrete diagram of the present invention. Fig. 5 is a block diagram of the main part of a practical example, Fig. 5 is an explanatory diagram of the analog control system shown in Fig. 4, and Fig. 6 is an explanatory diagram of signals of correction data due to bit defects. (1) ... line sensor, (2) ... buffer amplifier, 13?1), ... A/Dv@ device, so ... gain setting register, - ... resistance group, Kasumi ... operational amplifier : (Foundation)...Resistance, -...D/A converter, -...
- Offset setting register (70)...Programmable analog amplifier.

Claims (1)

[Claims] 1. When obtaining digital image data by digitally converting an image signal including gradation using an A/D converter, correction data for external light conditions at the time of image input and correction data for light receiving element variations, etc. At the time of acquisition, the image signal is magnified compared to the normal image signal in the previous stage of the A/D converter and inputted to the A/D converter to collect corrected data with high precision, and with this corrected data, the normal image signal is A method for correcting halftone images. 2. Insert an analog amplifier whose gain and offset can be externally controlled before the A/D conversion, and control this with a processor such as a microcomputer to match the dynamic range of the input analog signal with the dynamic range of the A/D conversion. 2. A method for correcting a halftone image according to claim 1, wherein said method is controlled to eliminate waste in the dynamic range of A/D conversion.