JPS6210069B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for setting a reference value for an image read signal when an image read signal read in analog quantities is binarized using a reference value as a boundary.

Conventionally, when image information is optically read as analog information by a scanner, and then converted into digital signals and binarized, the binary value of white or black is determined by comparing it with a predetermined threshold level. At this time, in order to respond to fluctuations in the image quality of the original, there are methods that follow the background level of the original and change the thread level, and statistical methods based on the density distribution around the pixel of interest. A method of determining the threshold level using a method has been proposed.

However, the former method has the disadvantage that unless the contrast of the document, that is, the difference between black and white, is large, normal information cannot be obtained when binarized, and the latter method only works with a very standard document, i.e., text and margins. This is effective if it is composed of , but if there is a large amount of black background part such as a frame, the disadvantage is that the slightly thin part caused by printing unevenness in the black background part will be judged as white. It was hot.

The present invention has been made focusing on the above points,
When performing binarization processing on an image read signal that has been read in an analog quantity using a reference value as a boundary, the level of the read signal is compared with the set level of a predetermined stage set in advance, and the frequency of signal occurrence for each set level is determined. In this method, the frequency of occurrence is accumulated for each level in order from the minimum or maximum set level, and the standard value is set from the set level when the cumulative value exceeds the standard reference value. When concentrating on a level or an upper level, the reference value is shifted away from the concentrated level, so that the image reading signal reference value prevents misreading of the background level in solid white areas and lack of information in solid black areas. It attempts to provide a setting method.

Embodiments of the present invention will be described below with reference to the drawings. Figures 1 to 3 are image density vs. appearance frequency characteristic tables when reading a manuscript. FIG. 2 shows a frequency characteristic table for almost only the background part (white solid), and FIG. 3 shows a frequency characteristic table for almost only the black background part (black solid).

Here, from the image information of about several lines, each read signal level is compared with the setting level of a predetermined stage set in advance to find the signal occurrence frequency for each setting level, and each level is set in order from the minimum or maximum setting level. If the threshold level is determined from the set level when the cumulative value exceeds a very standard reference value by accumulating the frequency of occurrence of
Even if it is changed for each line, the thread level will dig into the background area in the solid white area where information is concentrated in the W area as shown in Figure 2, and the thread level will dig into the background area as shown in Figure 2.
In a solid black area where information is concentrated in part B as shown in the figure, there is a risk that information will be deleted.

Alternatively, a fixed threshold level may be determined based on the overall frequency information after reading the image information of the entire object, but in this case, in addition to increasing the memory capacity, It is no longer possible to follow local changes such as areas where the light is becoming lighter or darker than other areas.

In such a case, more accurate reading can be achieved by shifting the threshold level S1 set by the method described above to S2.

FIG. 4 is a block diagram of a signal processing circuit using the image reading signal reference value setting method of the present invention.

first. Image signals from a line scanning type imaging device (not shown) are serially input one line at a time in the main scanning direction through a terminal V. The input image signal is converted into, for example, 4-bit digital information by an analog-to-digital converter A/D, and then converted into 4-bit digital information.
Shift register that buffers data for a line
Input to SR6. On the other hand, in parallel with this, by inputting this digital information to the decoder DE, one of the outputs 0 to 15 that matches the value is selected and outputs "1". This decoder DE
The output is an 11-bit level counter RC that is cleared to zero by the clear signal shown in the figure before one line of image signal is input as shown in Figure 5.
Each time the clock pulse shown in FIG. 5 is input to the ENABLE terminal E of the clock, the counter whose ENABLE terminal E is set to 1 is incremented by 1.

When this operation is performed for one line of image information, the level counter will change according to the input level of the image signal.
The number of appearances for each level for one line is stored in RC.

The contents of this counter are transferred to shift register SR1 by a load signal as shown in FIG.

Shift register SR1, SR2, SR3, SR4,
The SR5 sequentially stores data on the number of image level occurrences for one line every cycle time and transfers it to the adder circuit. During this transfer, data is sequentially input to the adder circuit ADD1 for each level of the level counter, that is, in units of 11 bits.

In this case, if the number of pixels in one line is 2000, the number of pixels in 5 lines is 10000, and if the cumulative frequency is 75% as the standard threshold level, that is, the threshold level S1 in Figures 1 to 3, then the comparator circuit
In RE1, 7500 is compared with the output of adder circuit ADD1, and when the cumulative frequency exceeds 7500, "1" is output to line 01. Then, the level at that time is stored in a buffer of a control circuit CRL, which will be described later.

On the other hand, in the adder circuit ADD2, the number of occurrences of each signal level L ₀ to L ₁₅ stored in the shift registers SR1 to SR5 is calculated as L ₀ +L ₁ , L ₁ +L ₂ . . .
..., L ₁₄ +L ₁₅ for each two sets of levels, and in comparison circuit RE2, the frequency characteristic of 60% for white solid and the frequency characteristic of 40% for black solid are compared with the adder circuit output. If the value of the adder circuit is large, the output of line 02 or line 03 is set to "1".

In addition, the control circuit CRL has a hexadecimal counter that determines the register position when inputting the data of shift registers SR1 to SR5 to the adder circuit.
Monitor _L0 as 0, _L1 as 1... _L15 as 15, and when the value stored in the buffer when the previous line 01 was output is within 7, the output of line 02 is "1". If so, add 1 to the buffer, and if the buffer is 8 or more and the output of line 03 is "1", subtract 1 from the buffer. Compare this value with circuit RE3
is given as a reference value and compared with the output of SR6 to generate a binary signal. Here, it takes one cycle time (one main scanning time) to allocate the image signal level for one line to the level counter, and by operating the adder circuit and the comparison circuit, the comparison circuit RE
Since it takes one cycle time to give the reference value to the line SR6, the threshold level of the output of the shift register SR6 is determined using the information of the two lines before and after that line.

According to the method of the present invention described above, when the frequency of occurrence of image information is concentrated at the white level or black level, the threshold level is shifted in a direction away from the concentrated level, so that solid white areas and solid black areas are Able to process information accurately.

[Brief explanation of the drawing]

1 to 3 are image density vs. occurrence frequency characteristic tables when reading a document, FIG. 4 is a block diagram of a signal processing circuit using the image reading signal setting method of the present invention, and FIG. FIG. 3 is a processed signal waveform diagram. A/D...Analog-digital converter, SR1~
SR6...Shift register, DE...Decoder,
RC...Level counter, E...ENABLE terminal,
ADD1, ADD2...Addition circuit, RE1, RE2...
...Comparison circuit, CRL...Control circuit.

Claims (1)

[Claims] 1. When performing binarization processing on an image read signal read in an analog quantity using a reference value as a boundary, the level of the read signal is compared with the set level of a predetermined stage set in advance, and the signal generation for each set level is determined. In the method of calculating the frequency, accumulating the frequency of occurrence for each level in order from the minimum or maximum setting level, and setting the reference value from the set level when the cumulative value exceeds the standard reference value, the frequency of occurrence is A method for setting a reference value for an image reading signal, characterized in that when the reference value is concentrated at a lower level or an upper level, the reference value is increased or decreased so as to move away from the level at which the signal is concentrated.