JPH05328126A - Method and device for image forming - Google Patents

Abstract

PURPOSE:To provide the method and device for image forming by optimizing the density of a copy picture by a neural network to output a vivid copy picture. CONSTITUTION:The density of an original detected by an AE sensor 1 is inputted to a DC controller 2 and is outputted as an optimum development DC control value to a development DC bias value control circuit 25 by the neural network consisting of software in a microcontroller part 22. Learning is performed to constitute the network which calculates the control value which outputs such development DC bias that a proper copy density is obtained, and thereby, the optimum development DC bias is obtained for various originals with a very high precision to obtain a vivid copy picture.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image forming apparatus and method for controlling the density of a copy image based on image data to form a copy image.

[0002]

2. Description of the Related Art Conventionally, for example, in an electrophotographic image forming apparatus, in order to maintain an optimum image density of a copied image,
The average density in an appropriate area of the original image is detected, and the value of the developing bias in the developing process and the light amount of the original illumination lamp are set to the optimum values based on the detected density. As a result, it is set so that a light copy image can be obtained for a document such as a newspaper that has a dark background and a low contrast.
It is set so that a normal copy image can be obtained for a black character having a certain degree. Further, if it is thinner than that, it is set so as to obtain a darker copy image.

[0003]

However, in the above conventional example, when optimizing the image density of the copied image,
Since the developing bias value and the light intensity of the lamp are calculated based on the average density in an appropriate area of the original image, the image contrast, pure background density, or character It was not possible to distinguish between the manuscript and the photo manuscript. Therefore, in the conventional method, the setting range of the set value that determines the copy density cannot be set sufficiently wide, and for example, even if the characters of a document thinly written with a pencil on white paper are automatically adjusted, the density is not sufficient. There was a problem that it could not be copied.

The present invention has been made to solve the above problems, and provides an image forming apparatus and method capable of outputting a clear copy image by optimizing the density of the copy image by a neural network. The purpose is to

[0005]

In order to achieve the above object, the image forming apparatus of the present invention has the following constitution.

In an image forming apparatus for forming a copy image by controlling the density of a copy image based on the image data, a detecting means for detecting the image density based on the image data, and an image density detected by the detecting means. And a control means for controlling the density of the copied image according to
The feature is that the density of the copied image is optimally controlled.

Further preferably, the detecting means detects the image density within a predetermined area.

In order to achieve the above object, another aspect of the invention is to control the density of a copied image based on image data,
In an image forming method for forming a copy image, a detection step of detecting an image density based on image data, and a control step of controlling the density of the copy image according to the image density detected in the detection step, The feature is that the density of the copied image is optimally controlled by the neural network.

[0009]

With the above construction, the image density can be detected based on the image data, the density of the copied image can be controlled according to the image density, and the density of the copied image can be optimally controlled by the neural network.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described in detail below with reference to the drawings.

In the following, an electrophotographic copying apparatus will be described as an example, but the image forming process is a known technique, and a detailed description thereof will be omitted. Also, in this example,
As a method of controlling the copy density, a so-called “method of changing the developing DC bias value” is applied, but it goes without saying that a method of controlling the irradiation intensity of the original document can be applied in the same manner.

<First Embodiment> FIG. 1 is a schematic diagram of a copy density optimization control system to which this embodiment is applied. 1 in the figure
Is an AE sensor that receives reflected light when an illumination (not shown) is applied to the document and detects the density of the document by its intensity, and includes an optical receiving element 11 and an amplifier 12. 2 is a DC controller for controlling the operation of the entire apparatus,
It has a reception buffer section 21 for receiving a signal from the E-sensor 1 and adjusting it to an appropriate voltage level, a microcontroller section 22, and a developing DC bias value control circuit 25 which is a means for controlling copy density. The microcontroller unit 22 includes an A / D conversion unit 23 that converts an analog signal from the reception buffer unit 21 into a digital signal, and an optimum development DC obtained by a neural network described later.
D for outputting the control value to the developing DC bias value control circuit 25
A / A converter 24.

FIG. 2 shows the DC controller 2 in FIG.
2 is a schematic diagram of a neural network configured by software in the microcontroller unit 22 of FIG. In FIG. 2, a plurality of (80 in this example) document density sampling data values stored in the RAM in the microcontroller unit 22 are input to the input layer. Further, information for obtaining the optimum value of the developing DC bias control value is set in the output layer. There are various possible output data formats from this output layer. The most direct method is to use the code value of the control data to the developing DC bias value control circuit 25. Also, a method of assigning various types of originals to each unit of the output layer, first inferring the types of originals, and using this result to determine the optimum value of the developing DC bias by an algorithm for another step, etc. Be done.

The operation of the present apparatus having the above configuration will be described below.

First, in order to detect the density of the document, the AE
The sensor 1 scans a part of the original, and the density of an appropriate width (in the long axis direction of the photoconductor drum) is set at a fixed interval and a fixed height.
Output to the DC controller 2. Then, the detected density data is received by the reception buffer unit 21 in the DC controller 2.
It is fetched as a digital value in the microcontroller unit 22 via. Specifically, the reflected light condensed by the optical lens (not shown) by 10 cm in the long axis direction of the photoconductor drum is collected 80 times every 1 mm during continuous movement of the document or the optical system mechanism. .. The state of the detection area at this time is as shown in FIG. The original density of an area of width 10 cm and height 1 mm is fetched in units of 1 byte (256 gradations) by one fetch, and a total of 80 bytes of sampling data is expanded on the RAM.

Next, using these data as input data,
The neural network calculates the optimum control value. This neural network is preliminarily trained to output optimum developing DC bias control value information for many test originals. Here, this learning procedure will be described.

4 to 8 show the results of sampling various test originals. The solid line shows the density, and the broken line shows the density conversion rate. First, in FIG. 4, the background is white and the characters are dark and the number of characters is small. In FIG. 5, the background is white and the characters are thin and the number of characters is large. In FIG. 6, the background is dark and the characters are normal and the number of characters is large. FIG. 7 is a picture with a light density, and FIG. 8 is a picture with a dark density.

For these, learning is performed so as to construct a network for calculating a control value for outputting the developing DC bias so that an appropriate copy density can be obtained.
An optimum developing DC bias can be obtained with very high accuracy for various originals, and the control range can be greatly expanded. For example, a general automatic density adjustment can sufficiently raise the developing DC bias to obtain a clear copy image even for an original document written with a thin pencil on white paper that cannot be completely corrected.

FIG. 9 is a diagram showing the control width of the developing DC bias control value. In the figure, what is indicated by a dotted line is the conventional control range. In the example of FIG. 9, a standard thin original is used as the test paper, and the development bias value at which this is output at an appropriate density is set as the upper limit. Also, a newspaper original is set as the standard dark original and this is the appropriate density. The developing bias value output at is set to the lower limit. On the other hand, according to the above-mentioned neural network, the control width can be expanded to the range shown by the solid line in the figure.

<Second Embodiment> A second embodiment of the present invention will be described below with reference to the drawings.

In the above-described first embodiment, the density distribution in a specific area of the image is continuously taken in, and the optimum control is executed by the neural network from that state. Although this method enables more detailed image discrimination, it has a relatively large memory capacity and a large amount of calculation, and is not suitable when a reduction in system scale and an increase in response speed are required preferentially. In view of this point, the second embodiment provides an apparatus capable of obtaining an optimum control amount at high speed.

FIG. 10 shows a neural network according to the second embodiment. As shown in the figure, in this embodiment, the information of the document image to be captured is reduced, and the density average value, maximum value, minimum value, density change rate average value, maximum value,
Only the minimum value of 6 data is used. In other words, while scanning with the AE sensor 1, if you perform real-time processing, the memory only needs to store 6 data, and the number of input units of the neural network is only 6. Therefore, the network scale becomes relatively small. There is an advantage that the processing time for calculating the optimum control amount is shortened.

<Third Embodiment> A third embodiment of the present invention will be described below with reference to the drawings.

FIG. 11 is a schematic diagram of the system in the third embodiment. As shown in the figure, this embodiment is a system in which all the processes from the detection of the original density to the output image density control value output are performed by hardware. That is, D
In the C controller 3, an integration circuit 31 for obtaining an average value, a maximum value, and a minimum value of the image density, a peak hold circuit 3
2. A minimum hold circuit 33 and a differentiating circuit 34, an integrating circuit 35, a peak hold circuit 36, and a minimum hold circuit 37 for obtaining the average value, the maximum value, and the minimum value of the image density change rate are used as the input section from the AE sensor 1. The output from them is input to the input layer of the neural network circuit 38. The decoder 39 decodes the output from the neural network circuit 38 and outputs the image density control signal. Further, in this example, an example in which the developing bias and the document illumination lamp are optimally controlled is shown.

Further, the DC controller 3 includes a microcomputer (MPU) 40, and the MPU 40 controls the entire circuit described above in accordance with a program stored in a ROM (not shown).

As described above, according to the third embodiment, since all processing is performed by hardware, the processing time can be greatly shortened. It is also possible to optimally control the developing bias and the original illumination lamp by associating them with each other.

<Fourth Embodiment> A fourth embodiment of the present invention will be described below with reference to the drawings.

The fourth embodiment will be described by taking as an example the case of applying it to a digital type electrophotographic printer which has 8-bit gradation data per pixel and outputs an image of 256 gradations. Generally, in the case of a digital type electrophotographic printer, a method of transferring a plurality of gradation data per pixel to an image density control device (for example, light irradiation time control or light intensity control) of the printer to change the image density is a method. Has been taken.

However, when functioning as a copying machine, if the reproducibility of the printer and the reading characteristic of the scanner are different, transferring the read data as it is to the printer side not only deteriorates the image reproducibility but also linearly. It is impossible to obtain good gradation.

Therefore, usually, the "original image data" read by the scanner and the "gradation control data" transferred to the printer.
The functional relationship of is optimized to maintain the consistency between the original image density and the output image density. This functional relationship is realized by a memory table usually called a γ table.

That is, in the fourth embodiment, a plurality of γ tables are prepared as shown in FIG. 12, and control is performed so that the γ table to be referred to is switched to the optimum one based on the same input data as described above. Is. FIG. 12 shows the input / output relationship of the γ table. In this example, four tables (A to D) are prepared.

FIG. 13 shows actual memory space allocation. In this example, since 8-bit (256) gradation is considered, the address of each table is set to the lower 8 bits 0.
If it is set to 0h to FFh, each table may be switched by switching the lower 9th and 10th bits of the address.

According to the embodiment described above, the image density of a part of the original image is input, and the spatial distribution and change rate of the density in the image within an appropriate area are used as the input data to the input layer.
Information that determines the control amount of the copy image density control is used as output data from the output layer, and the density of the copy image is optimized by a neural network that has an appropriate intermediate layer between them, thereby making it possible to obtain a wide variety of types. A wide range of originals can be handled, and a copy image with optimum copy density can be output.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. It goes without saying that the present invention can also be applied to the case where it is achieved by supplying a program to a system or an apparatus.

[0035]

As described above, according to the present invention,
By optimizing the density of the copied image using the neural network, a clear copied image can be output.

[Brief description of drawings]

FIG. 1 is a schematic diagram of a copy density optimization control system in a first embodiment.

FIG. 2 is a schematic diagram of a neural network according to the first embodiment.

FIG. 3 is a diagram for explaining a scanning range in the first embodiment.

FIG. 4 is a diagram showing a sampling result of a test document in the first embodiment.

FIG. 5 is a diagram showing a sampling result of a test document in the first embodiment.

FIG. 6 is a diagram showing a sampling result of a test document in the first embodiment.

FIG. 7 is a diagram showing a sampling result of a test document in the first embodiment.

FIG. 8 is a diagram showing a sampling result of a test document according to the first embodiment.

FIG. 9 is a diagram for explaining a control range in the first embodiment.

FIG. 10 is a schematic diagram of a neural network according to a second embodiment.

FIG. 11 is a schematic diagram of a copy density optimization control system in the third embodiment.

FIG. 12 is a diagram showing a gamma table in the fourth embodiment.

13 is a diagram showing a state where the table of FIG. 12 is assigned to a memory.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 AE sensor 2 DC controller 11 Light receiving element 12 Amplifier 21 Reception buffer 22 Micro controller section 23 A / D conversion section 24 D / A conversion section 25 Development DC bias value control circuit

Continuation of front page (51) Int.Cl. ⁵ Identification number Office reference number FI technical display location G06F 15/18 8945-5L G06G 7/60

Claims (3)

[Claims] 1. An image forming apparatus for forming a copy image by controlling the density of a copy image based on the image data, and detecting means for detecting the image density based on the image data, and an image detected by the detecting means. An image forming apparatus comprising: a control unit for controlling the density of a copied image in accordance with the density, and optimally controlling the density of the copied image by a neural network. 2. The image forming apparatus according to claim 1, wherein the detecting unit detects an image density within a predetermined area. 3. An image forming method for forming a copied image by controlling the density of the copied image based on the image data, the detecting step of detecting the image density based on the image data, and the image detected by the detecting step. An image forming method, comprising: controlling the density of the copied image according to the density, and controlling the density of the copied image optimally by a neural network.