JPS6378668A - Image reader - Google Patents

Abstract

PURPOSE:To precisely binarize a signal even if a base density dynamically fluctuates by deciding a binarization threshold every main scan line with the aid of the black and white peak values in terms of the density of a picture signal inputted to an original. CONSTITUTION:A data W-PK about the white peak values of previous eight lines including a current one is stored in a memory, and data lying eight line before is updated for the unit of the line. Data obtained from a black peak detection circuit 11 is stored as a black peak value B-PK in the memory. According to a white peak value W-AVD obtained from said data, a slice level is determined. The inputted black peak value is compared with the white peak value of the current line, and a lower limit is set again. In case a picture(including a character) on the original and the base density dynamically fluctuate, a threshold for binarization can cope with that appropriately.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an image reading device that binarizes an image information signal obtained by scanning an original image and photoelectrically converting it into each pixel. be.

(Prior Art) Conventionally, in this type of image reading device, an electrical image (information) signal obtained from a COD (charge-coupled device) image sensor or the like as a photoelectric conversion element is generally
Binarization is performed by applying a voltage to one input of the comparator and integrating the image signal using an integrating circuit including a capacitor as a threshold value and applying it to the other input of the comparator.

[Problem that the invention seeks to solve]

However, in the conventional binarization method as described above, when the image information signal of the image does not change for a long time, the value of the previous period approaches the level of the input image signal as much as possible, so the binary value is not changed. The drawback was that the conversion operation became unstable.

Furthermore, due to the time constant of the integrating circuit for integration, the ability to follow changes in the density of the background or character parts of a document, for example, is not good, and if the background density changes dynamically, the correct 2
There was a problem that the value conversion process was not performed.

The present invention has been made in view of the above-mentioned problems, and aims to provide an image reading device that eliminates the above-mentioned drawbacks and can always perform stable binarization processing.

[Means for solving problems]

Therefore, in the present invention, there is provided a means for holding the self-peak value and the black peak value on the density of the input image (fine report) signal of the original Y1 draft for each main scanning line, and a means for holding the black peak value and the white peak value for each main scanning line. The present invention attempts to achieve the above object by providing means for determining a binarization threshold value for each main scanning line using i.

[Effect]

With the configuration described above, in the present invention, the threshold always appropriately follows dynamic fluctuations in the original W, the original image/characters, and the background density, and the Since the threshold value 1 does not follow, accurate binarization processing becomes possible.

〔Example〕

The present invention will be explained below based on examples.

FIG. 1 is a block diagram of an embodiment of a binarization processing circuit configuration according to the present invention.

(Structure) (Fig. 1) 101 is an optical lens, 102 is a photoelectric conversion element C
OD line image sensor 106 is a CCD drive circuit. The image (information) signal imaged on the CCD line image sensor 102 is amplified by the amplifier 103,
5-bit analog/digital (A/D) converter 10
4 is input. The image signal converted into a 5-bit digital signal is latched into a D flip-flop 105 by a clock generated for each pixel obtained from a timing signal generation circuit 107, and then output to a digital comparator 108 and a peak value for each main scanning line. The signal is input to a white peak detection circuit 110 and a black peak detection circuit 111, which detect the signal.

The data obtained from each peak detection circuit 110 and 111 is inputted to a CPU (central processing unit), which uses the navigation data to determine the slice level for each main scanning line, and determines the slice level for each main scanning line. is output to. In the digital comparator 108, the image signal obtained from the D flip-flop 105,
By comparing the slice level obtained from this D flip-flop 109 and latched for each main scanning line, 2
Outputs a value signal THOUT.

(Operation) (Fig. 2.3) Next, the slice level determination method will be explained based on Fig. 2.3. FIG. 2 is a sequence flowchart of the slice level determination operation, and FIG. 3 is an example of density data of a document image resulting from the second same operation. The range of the manually generated density data is 0 to 31, with 0 representing pure white and 31 representing pure black. Further, the program shown in FIG. 2 is activated for each main scanning line to determine the slice level for the next line.

Each step will be explained in order below. In step sp1, the density data obtained from the white peak detection circuit 110 (FIG. 1) is read into the CPU, and in step SP2, the process proceeds to step SP3 only when the read data is smaller than 10, and in other cases If so, terminate the program without updating the slice level. this is,
This is to prevent the slice level from following when the original image is abnormal.

In step SP3, the white peak value data W-PK for eight previous lines including the current line is stored in the memory, and the data for eight lines before is updated for each line. Next, in step SP4, the data obtained from the black peak detection circuit 111 is stored in the memory as a black peak value B-PK. In step SP5, the process proceeds to step SP6 only when this black peak value B-PK is greater than 10, and in other cases, the program is terminated without updating the slice level. This is because, as in step SP2, the slice level is not tracked for both the abnormal vll and the abnormal images.

Step SP6 is the white peak value W-AV of the current line.
D is being calculated and determined by the following formula.

(W-A V D) n-1) ・”
(+) In the above formula, (W-AVD)n is the white peak value of the nth line, and is the average value from the n-7 line to the beginning of the n line, that is, the average value in the vicinity of the current line, so to speak, in a short section. It is obtained by averaging two values: the white peak value and the W-AVD of the previous line, that is, the average white peak value over a long period from the start of scanning.

This is done for the white peak value, that is, the density level of the background, so that the slice level will not be affected by materials such as blueprinted originals, where the background density is not constant and tends to fluctuate over a short period of time. , use the average value of 8 lines,
This is also to stabilize the slice level by taking into consideration the background average density over a long period.

Next, in steps SP8 to SP5P15, the above (
1) The slice level is determined according to the white peak value W-AVD obtained according to the formula. Step SP7~5P1
1 is a case where W-AVD>6, and step SP11
~5P14 is a case where 6≦W-AVD≦4, and step 5P15 is a case where W-AVD≦4. In this case, the background level is considered to be sufficiently white, so the lower limit value of the slice level is is not set, and the calculated value is used as the slice level.

As described above, there are three types of processing depending on the background level, but since the only difference in processing content is the set values, only steps SP8 to SP5PIO will be described. That is, in step SP8, the slice level S of the next line is determined by finding the intermediate value between the white peak value of the current line determined by the method described above and the inputted black peak value.

In step SP9, the input black peak value and
Compare the difference with the white peak value of the current line, and the difference is 14
If it is larger, the process proceeds to step 5P16; otherwise, the slice level S is reset to the lower limit value 18 (step 5pto), and the process proceeds to step 5P16.

The reason for setting the above-mentioned lower limit here is to avoid the tendency for black streaks to occur in non-text areas when the background level increases (approaches black). In addition, the difference 14 between the white peak value and the black peak value and the lower limit value 18 were determined by practical calculations as an example in this embodiment, and of course they will vary depending on each application example. be. In step 5P16, the previously determined slice level S is written to the D flip-flop 109, and the program is ended.

Next, an example of the density data level of this embodiment will be specifically explained based on FIG. 3. M in the upper part of FIG. 3 is an original document, which has images of each of the characters A to E on the background. The figure below is a timing chart in the sub-scanning direction of density data for M in the figure above when actually scanning.B-PK, W-PK
, S represent the black peak value, white peak value and slice level, respectively.

The original iM has different background densities in the parts marked ■, ■, and ■, which correspond to the white peak value W-PK shown in the figure below.

Further, in the character (image) portion, the density is different between A, B and C, D, and E, which corresponds to the black peak value B-PK shown in the figure below. Furthermore, the shaded area in the figure below indicates that it will actually be black.

Each of the above parts will be explained below.

In the 0 part of the original document M, the white peak value W-PK of the background is 5, so step S in FIG.
It corresponds to P11 to 14, and the non-character part has a lower limit of 1.
4, and the character part has an intermediate value of 18. In the part (3), the white peak value W-PK is 8, so step S in Figure 2
Corresponds to P8 to P10, the non-text part has a lower limit value of 18, the text part B has an intermediate value of 18, and the character part C has a black peak value B-PK and a white peak value W-PK. Since the difference is within the predetermined value of 14, the lower limit value is 18. In this area, the background density is high, making it difficult to distinguish between text (image) and non-text areas.
However, according to this method, the lower limit value of the slice level is set according to the background level, and the white peak value W-PK and the black peak value B
By setting the difference from -PK according to the background level, determining non-text portions, and setting the lower limit value as the slice level, it is possible to accurately reproduce only the text portions.

Next, in the 0 part of the original document M, there are two white peak values W-PK, so step SPI in FIG.
5, and the intermediate value becomes the slice level. Also,
Since the black peak value B-PK of the non-text portion is 6, the slice level is not updated and remains constant at step SP in FIG.

(Other Embodiments) In the above embodiment, the background density level of the original document M is determined in only three ways due to the relationship between the processing speed of the CPU and the accumulation time of the CCD line image sensor 102 (FIG. 1). However, by increasing the processing speed of the CPU and adding judgment of the background density level, it becomes possible to set a more appropriate slice level S.

〔Effect of the invention〕

As described in detail above, according to the present invention, even if there are dynamic changes in the image (including characters) and background density of the original document, it is possible to appropriately follow the threshold value of the binarization process. In addition, since the image reading apparatus does not follow the density of the image, it is possible to provide an image reading apparatus that can accurately perform binarization processing.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the configuration of an embodiment of the binarization processing circuit according to the present invention, FIG. 2 is a slice level determination operation sequence flowchart according to FIG. 1, and FIG. 3 is density data of an original document image. This is an example.

Claims (4)

[Claims] (1) Scan the original image and convert the photoelectrically converted image signal into 2
The image reading device converts the image signal into a value by using a holding means for holding the white peak value and the black peak value of the image signal for each of the main scanning lines, and using the white peak value and the black peak value. threshold determining means for determining a threshold for value conversion for each main scanning line, and the threshold determining means determines the white peak value when the difference between the white peak value and the black peak value is smaller than a predetermined value. An image reading device that determines a threshold value based on. (2) The white peak value average value determining means determines the white peak value of the current line based on the white peak average value from the first line at the start of the main scanning and the white peak average value from the current line to several lines before. 2. The image reading device according to claim 1, wherein the image reading device determines a peak value. (3) The image reading device according to claim 2, wherein the threshold determining means for binarization sets the threshold to be an intermediate value between the white peak value and the black peak value. (4) The threshold determining means sets a threshold according to the white peak value, and when the difference between the black peak value and the white peak value is smaller than the set value, the set value is used as the threshold. An image reading device according to claim 1 or 2, characterized in that: