JPH0225223B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Technical Field of the Invention The present invention relates to an image data processing device, and more particularly to a logical filter suitable for window operations on multivalued image data.

(b) Background of the technology In image data processing using a computer system, the image to be processed is usually divided into pixels arranged in a matrix (grid shape), and image information (for example, density) is transmitted for each pixel. By representing it as multivalued data and performing various arithmetic processing on it, you can calculate, for example, the normal line at each part of the outline of the image, calculate the distance between multiple separated images, calculate the size of the image, and calculate the size of the image. The noise is removed, etc.

(c) Prior art and problems The basic calculations for the image data processing include level conversion using a fixed threshold value for each pixel of the image given as the multi-value data, and conversion of the image given as the multi-value data. Examples of pixel-by-pixel calculations or window calculations between two images include maximum value, minimum value, or average value calculation within a window.

However, these processes involve calculating the normal line, calculating the distance between multiple separated images, calculating the image size, removing noise in the image, etc., and calculating the relationship between pixels in a certain window within the same image. The drawback is that it cannot be used for the desired purpose.

In addition, there is a spatial filter that finds the relationship between pixels in a certain window in the same image, but spatial filters require processing that requires a large amount of calculation such as multiplication or addition, and therefore takes a long time to process. There are drawbacks.

(d) Purpose of the Invention The purpose of the present invention is to calculate the normal line, calculate the distance between multiple separated images, calculate the size of the image, remove noise in the image, etc. An object of the present invention is to provide a logical filter for processing multivalued image data, which can easily and quickly perform processing for determining relationships between pixels.

(e) Structure of the Invention The logical filter for processing multivalued image data according to the present invention has a binarization circuit, a conversion table, and a switching circuit, and the binarization circuit has pixels arranged in a matrix. Input image data consisting of multivalued data for each, binarize it using two given thresholds,
When the image data is scanned by a 3 x 3 window, binary data for each 9 pixels is sequentially output, and the conversion table receives the binary data as input and converts it to a given preset value. The switching circuit outputs multi-value data and control values corresponding to the binary data, and the switching circuit converts the multi-value data output from the conversion table and the multi-value data of a predetermined pixel of the image data into
It is configured to switch according to the control value and output as multivalued data of the center pixel of the window.

(f) Embodiments of the Invention Figure 1 shows an example of an image data processing system using a logical filter for processing multivalued image data. An image memory stores each density for each pixel as multi-value data, 2 is a control module for controlling the logical filter 3 for processing multi-value image data, 4 is a control bus, 5 is a read bus, and 6 is a write bus. be.

FIG. 2 shows a configuration diagram of an embodiment of a logical filter for processing multivalued image data according to the present invention, 3-1,
3-2, 3-3, 3-4, 3-5, 3-6, 3-
7 and 3-8 scan image data given as multivalued data for each pixel arranged in a matrix using a 3 x 3 window, and apply image data corresponding to each element of the window to two threshold values. Binarize using 2 to obtain 3×3 binary data.
A value conversion circuit 3 is configured, and 3-1 compares the multi-value data f for each pixel with a first threshold value T1, and converts the binary data to "1" or "0" according to f≧T1 or f<T1. A comparator 3-2 that outputs "1" or "0" of the binary data compares the multi-value data f for each pixel with the second threshold T2 and outputs "1" or "0" of the binary data according to f≦T2 or f>T2. Output comparator, 3-3 is AND circuit, 3-
4 is a shift register that stores the output of the AND circuit 3-3 for each three pixels, and 3-5 is an AND circuit 3-3.
A line buffer 3-6 stores the output of the line buffer 3-5 for each three pixels.A shift register 3-7 delays the output of the line buffer 3-5 by one scanning line. Line buffer to give, 3-8 is line buffer 3-
7 is a shift register that stores the output of 7 for each three pixels; 7 is a multiplexer that selects either table address A to be given to conversion table 8 or conversion table 9 described later and the output of binarization circuit 3; 8
is a conversion table that outputs a binary signal “1” or “0” according to a predetermined rule according to the 3×3 binary data obtained by the binarization circuit 3;
9 is a conversion table for converting the 3×3 binary data obtained by the binarization circuit 3 into 8-bit multi-value data according to predetermined rules; 10 is a conversion table for converting the data by the output of the conversion table 8; A multiplexer 11 selects either the output of the table 9 or the output of a delay buffer 11 to be described later, and a delay buffer 11 provides a delay of one scanning line and two pixels to the image data f.

Note that the table address A is input to the conversion table 8 and the conversion table 9 via the multiplexer 7.
Specify the corresponding binary signal or multi-value data C
Memorize in advance.

In such a configuration, for example, when the first threshold T1 and the second threshold T2 are set to 2 and 5, respectively, and multivalued image data as shown in FIG. Binary data as shown in FIG. 3b is obtained, and the binary data in a 3×3 window is output in the order of scanning. However, in FIG. 3b, H represents binary data "1".

When the 3×3 binary data obtained in this way is given as a 9-bit table address, the conversion table 8 or conversion table 9 is
Element A8 at the center of the 3x3 window (see Figure 4)
The multiplexer 10 outputs a binary signal or multi-value data represented by 8 bits to each pixel corresponding to the pixel, and the multiplexer 10 converts the conversion table 9 depending on whether the output of the conversion table 8 is "1" or "0".
or the output of the delay buffer 11.

Next, an example of image processing according to the embodiment will be described.

FIG. 5 is an example of determining the direction of the normal line at each pixel forming the outline of an image. ), the direction of the normal line at each pixel forming the contour is obtained as b. However, the direction of the normal is defined as c, and according to this definition, the conversion table 9 has a correspondence like d (the left side is 3x3 binary data specifying the table address of the conversion table 9, On the right side is a code that indicates the direction, and of the 3 x 3 binary data, the × mark is “1” or “0”, and the blank part is “0”.
(hereinafter the same) is stored in advance, and "1" is stored in the conversion table 8 corresponding to all inputs.

Figure 6 is an example of calculating the distance between multiple images, and by performing n times of processing, distances from each contour of three separated images A, B, and C given by binary data as shown in a are obtained. The distance in the direction is determined as b.

Conversion table 9 contains four types of 3×3 like c.
For any of the binary data (the H mark represents "1" and the L mark represents "0" - the same applies hereafter), it is "2" in the first processing and "3" in the second processing. ,…
..., in the n-th processing, "n+1" is stored, and the conversion table 8 contains four types of 3 such as c.
"1" is stored for all of the x3 binary data, and "0" is stored for the other binary data.

Then, in the first processing, T1=T2=1, in the second processing, T1=1・T2=2,
In the third processing, by setting T1 = 1 and T2 = 3, distances can be obtained through three processings as illustrated in d, e, and f.

Figure 7 is an example of determining the size of an image, where it is determined by the distance inward from each contour of four partial images A, B, C, and D in an image given by binary data as shown in a. This is to find the size of each image, and the conversion table 9 contains four types of 3×
For any of the binary data of 3, the first processing is "2", the second processing is "3", etc.
..., in the nth processing, "n+1" is stored,
On the other hand, the conversion table 8 contains four types of 3 such as c.
"1" is stored for all of the x3 binary data, and "0" is stored for the other binary data.

For example, for figure d of c, the first
By setting T1=T2=1 for both the second processing, the second processing, and the third processing, the distances are obtained as illustrated in d, e, and f through the three processings, and the final The maximum value "4" in the center of the output data represents its size.

Figure 8 is an example of removing noise in an image.
Among the four partial images A, B, C, and D in an image given as binary data like a, patterns like A are considered to be noise, and b is obtained by removing them. It is something to do.

In this example, noise removal is performed in three processes, and in each process T1 = T2 = 1, and in the first process, the conversion table 8
corresponds to four types of 3 x 3 binary data such as c, and all store "1", and the conversion table 9 corresponds to four types of 3 x 3 binary data such as c. In the second processing, "0" is stored in the conversion table 8 corresponding to five types of 3x3 binary data such as d, and , conversion table 9 has d
"0" is stored for all 3x3 binary data except for the 5 types, and in the third processing, all 3x3 binary data are stored in the conversion table 8.
"1" is stored in the conversion table 9 corresponding to the binary data of 3 x 3 such as e, and "1" is stored in the conversion table 9 corresponding to the binary data of 3 x 3 such as e.
“0” is stored in all cases corresponding to 3×3 binary data other than ×3 binary data.

As a result, 3
By performing this process twice, desired noise can be removed.

As described above, in the present invention, the multivalued image data to be processed is converted into binary data by classifying it using two threshold values that define the required value range, and the binary image data is converted into 3×3 By indexing the image data using binary data of each 9 pixels when scanning with a window, the multi-value data and control value of the conversion value are determined, and the multi-value data and control value of the conversion value are switched according to the control value and output from the conversion table. Processing is performed to output the value of multivalued data or input data as the value of the center pixel of the window. This process is executed for all pixels, and if necessary, this series of processes is repeated as many times as necessary, and the threshold value and the contents of the conversion table are updated at each repetition, if necessary. In this way, by setting appropriate threshold values and conversion table contents, for one screen of multivalued image data, calculation of normal line, calculation of distance between multiple separated images, calculation of image size, calculation of image size, etc. Various processes such as removing noise inside the image can be performed.

(g) Effects of the Invention As explained above, according to the present invention, it is possible to calculate the normal line at each part of the outline of the three images, calculate the distance between multiple separated images, calculate the size of the image, and noise in the image. It is possible to easily and quickly perform processing such as removal of pixels to determine the relationship between pixels in a certain area within the same image.

[Brief explanation of drawings]

FIG. 1 is an example of an image data processing system using a logical filter for processing multivalued image data, FIG. 2 is a block diagram of an embodiment of the logical filter for processing multivalued image data according to the present invention, and FIG. FIG. 4 is an explanatory diagram of a binarization circuit, FIG. 4 is an explanatory diagram of a conversion table, and FIGS. 5, 6, 7, and 8 are explanatory diagrams of various calculation processes according to the embodiment. In the figure, 1 is an image memory, 3 is a logical filter for processing multivalued image data, and 8 and 9 are conversion tables.

Claims (1)

[Claims] 1. A binarization circuit, a conversion table, and a switching circuit. , binarize using two given thresholds, and sequentially output binary data for each 9 pixels when the image data is scanned by a 3×3 window, and the conversion table is The switching circuit receives binary data as input and outputs preset given multi-value data and control values corresponding to the binary data, and the switching circuit outputs the multi-value data and control values output from the conversion table. A logical filter for processing multivalued image data, characterized in that multivalued data of a predetermined pixel of the image data is switched in accordance with the control value and outputted as multivalued data of a center pixel of the window.