JPS6288075A - Picture processor - Google Patents

Abstract

PURPOSE:To simply attain a high speed process by outputting a code regarding an observation direction assigned by a density string encoding part, and outputting the variable density structure information of a remarked picture element from a variable density structure information output part setting the code string as an input. CONSTITUTION:A multivalue density digital signal from a scanning and photoelectric transducing part 1, after it is stored temporarily at a multi-gradation picture storage part 2, is given to a density string output part 3. The density string output part 3 reads out the density signal of each picture element within a picture element area of 3X3 centering an optical remarked picture, and outputs picture element density strings regarding four observation directions. Density string encoding parts 4a-4d input the picture element density string classified at every observation direction obtained respectively from the density string output part 3, and output a density state after encoding. These four codes classified by every direction are given to a variable density structure information output part 5 and the density structure information of the remarked picture element is outputted.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an image processing device that detects pixels corresponding to a pre-designated shading structure from characters, two figures, photographs, and images containing a mixture thereof. -Yes.

[Conventional technology]

Conventionally, pixels from multi-tone digital images whose density is significantly higher or lower than surrounding pixels.

When detecting pixels at the edge of a figure whose density changes rapidly in a specific direction, pixels located at e-degree ridge lines, pixels located at valleys, etc., (i) a method using differential calculation;
(ii) A method based on template matching has been used.

(i) is a calculation equivalent to a first-order differential (Gradient) or a second-order differential (Laplacian). For example, in the example of FIG. 8, A-I indicates the density along with the position of each part, and the central pixel E is It is currently a pixel of interest. As the first derivative, (A+B+C-G-H-1)2+ (A+D+G-C
-F-I)2 or.

l A+B 〇-G-H-I 1 + l A+D+
GC code -11 (When considering direction, calculate each term separately) As a second derivative, E-(A+B+C-F+I+H+G+D) / 8
``t'' and detects a desired pixel based on the magnitude of the calculated value.

(ii) is a standard mask (temporary mask) of the structure you want to detect.
-1.) to detect a desired pixel by taking a correlation with the image. Examples of templates are shown in FIGS. 9(a) and 9(b).

9 [ff1(a) is for corner detection, FIG. 9(b) is for corner detection.
, is a template for the detection of di. Using this, for example, sequentially compare 3×3 pixels in 36×36 pixels,
The place where the sum is obtained is called a corner or an edge. This method can be easily applied to two images, but when applied to a multi-tone image, it is necessary to binarize it with an appropriate threshold or to calculate the average density of all pixels in the template. It is necessary to perform calculations to compare the density of each pixel and the density of each pixel.

[Problem that the invention seeks to solve]

Although these methods are widely used and effective in the field of image processing, all of them involve calculation of density differences between pixels and size comparisons. Therefore, in the case of images consisting of a large number of pixels, or in cases where extremely high-speed processing is required, it is necessary to use dedicated circuits or parallel calculations.
This necessitated the introduction of a new method, which caused problems such as identification of processing and complication of circuits.

An object of the present invention is to provide a fast, simple, and flexible image processing device that solves the above-mentioned problems.

[L stage for solving problems]

The image processing device according to the present invention has a specified view 311.
A density string encoding unit that receives a direction pixel density string as input and outputs a code for each direction, and inputs a code related to one direction obtained by this density string coding unit, or a code string that is a combination of codes related to a plurality of directions. and a grayscale structure information output unit that outputs grayscale structure information of the pixel of interest. In the present invention, the density sequence coding unit outputs codes related to one or more directions related to the designated observation direction. With these code strings as input, the shading structure information of the pixel of interest is output from the shading structure information output section.

〔Example〕

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.
As shown in Figure 2, the X3 pixel area is horizontal, vertical, right 45
°, left 45°c7) a, b.

This is an example in which C light structure information of a pixel of interest is output by observing it from four directions c and d. In the figure, 1 is a scanning/photoelectric conversion unit that scans an original image and converts it into a multi-level density digital signal at the middle level of the pixel, and 2 is a multilevel unit that stores the digital signal obtained from the scanning/photoelectric conversion unit 1. tone image storage section, 3
reads out the density signal of each pixel in a 3×3 pixel area centered on an arbitrary pixel of interest from the multi-gradation dual image storage unit 2)
, 2.4 Density string output units 4a to 4d each output a pixel density string related to the observation direction, each inputting a 1 degree sequence of pixels for each observation direction obtained from the density string output unit 3, and code the density change state. There is a CT density sequence coding section that converts the data into 1D image data and outputs it. Reference numeral 5 denotes a gray scale structure information output unit which receives the four directional codes outputted from the density sequence encoding units 4a to 4d and outputs the gray scale structure information of the pixel of interest.

For example, the density column output section 3 outputs nine 1's as shown in FIG.
Pixel shift registers 311 to 333 and two 1
It can be constructed more easily than the line shift registers 34 and 35. That is, for example, 3 out of 36x36 pixels
After one pixel is output from the shift registers 311 to 313 for the three pixels in row [1 of
After the 66 pixels are delayed by the 1 line 2/minute shift register 34, the output is made from the shift registers 321 to 323 by one pixel. Thereafter, a delay of one line is performed by the shift register 35 for 1917 minutes, and outputs are output from the shift registers 331 to 333 for one pixel.

Each of the density column encoding units 4a to 4d is composed of a register 41 and a ROM 42, as shown in FIG. 4, for example.

The register 41 has a 12-bit configuration and stores the density values of three pixels in the observation direction in four bits each. ROM42 is 3
Codes for all combinations of pixel density values are stored, and the codes are output using the density value string stored in the register 41 as an address.

FIG. 5 shows an example of a code expressed in 3 bits. In this figure, temperature changes are indicated by arrows, → indicates no change, and / and \ indicate increases or decreases in concentration changes. These changes are expressed in 8 ways, and the codes are ' o o o ''~''11
1'' is assigned. For convenience of explanation, code numbers 0 to 7 are assigned to these codes.

The gradation structure information output section 5 is composed of a register 51 and a ROM 52, for example, as shown in FIG. register 5
1 stores four codes for each observation direction. When adopting the 3-bit encoding method shown in Figure 5,
Register 51 has a 12-bit configuration. ROM52 is
The shading structure information for all combinations of four codes for each observation direction is stored in advance, and the shading structure information is output using the code string stored in the register 51 as an address.

FIG. 7 shows an explanatory diagram of the observation direction and the ROM 52 for detecting pixels located on ridge lines and pixels located on valleys.
An example of the content is shown below. In FIG. 7, a to d shown on the left side of the ROM 52 are the observation directions shown in FIG. 2, and the number F indicates one of the code numbers in FIG. 5 (in this example, 0, 1 Only ゜2 is used). and “'abc
d'', that is, the address is composed of octal numbers x 4, and the ROM 52 stores grayscale structure information, for example, the 7th
In the figure, '01' is stored as a ridge line pixel, '10' as a valley pixel, and so on. In the example in Figure 7, '
All except the 8 locations indicated by 'a b c d' °
It is “OO”.

By using the configuration shown in this example, a 3×3 pixel area can be divided into horizontal, vertical, 45° right, 45° left, a, b, c,
The process of observing from the four directions of d and detecting pixels located on ridge lines, pixels located in valleys, etc. can be performed without performing concentration difference calculations, size comparisons, etc., resulting in faster processing. can be realized. Furthermore, by changing the contents of the ROM 42 and ROM 52, it is possible to detect any shading structure that can be determined in the E-4 observation direction, realizing not only high speed but also a flexible configuration.

[Effects of the Invention] As explained above, the present invention includes a density sequence coding unit that receives a pixel density sequence in a specified observation direction and outputs a direction-specific code, and a density sequence coding unit that outputs a code for each direction. A gradation structure information output unit is provided which receives a code string related to one direction or a code string that is a combination of codes related to a plurality of directions as input and outputs gradation structure information of the pixel of interest, so that it is possible to calculate the density difference between pixels and to perform size comparisons. There is an advantage that the shading structure information of the pixel of interest can be directly obtained from the pixel density array in one direction or in multiple directions, and that high-speed processing can be easily realized. In addition, it is easy to change the encoding method of the density string and the method of converting the code string to shading structure information.
It has the advantage of being able to be processed flexibly depending on the application.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.
FIG. 2 is an explanatory diagram of the observation direction, FIG. 3 is a block diagram showing an example of the configuration of the concentration column output section in the embodiment of the present invention, and FIG. 4 is a block diagram showing an example of the configuration of the concentration column encoding section. Figure 5 is a diagram showing an example of the code, Figure 6 is a block diagram showing an example of the configuration of the gradation structure information output section, and Figure 7 is a diagram showing pixels located on the ridge line and pixels located in the valley. FIG. 8 is a diagram of a 3×3 pixel area for explaining the conventional differential operation; FIG. 9(a), (
b) is a diagram showing an example of a conventional template. In the figure, 1 is a scanning/photoelectric conversion unit, 2 is a multi-gradation image storage unit,
3 is a density string output section, 4a to 4d are density string coding sections, 5
is a shading structure information output unit. Figure 1 Figure 2 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9

Claims (1)

[Claims] A scanning/photoelectric conversion unit that scans an original image and converts it into a multi-level density digital signal for each pixel, a multi-gradation image storage unit that temporarily stores the digital signal obtained by the conversion means, and this multi-gradation image storage unit. In an image processing device that reads out the density signals of an arbitrary pixel of interest and each pixel in its surrounding pixel area to determine the density structure of the pixel of interest, a direction-specific code is generated by inputting a pixel density string in a specified observation direction. A density sequence encoding unit to output, and a gradation unit that outputs shading structure information of the pixel of interest by inputting a code related to one direction obtained by this density sequence coding unit, or a code sequence obtained by combining codes related to a plurality of directions. An image processing device comprising: a structure information output section.