JPH05233802A - High-speed processing system for reducing median noise - Google Patents

Abstract

PURPOSE:To execute a processing at a high speed for reducing noise, which is added by a sensor or the like, contained on the image of image data expressed by the multi-valued level of image data recorded in the storage device of a computer while using a median filter. CONSTITUTION:A median extraction table 301 is provided for calculating a median value at high speed, and a processed image 202 reduced noise is prepared from a source image 201 by respective processing parts composed of an initial table preparation part 101, table update part 102 and median value extraction part 103. Thus, processing time for reducing noise in the image data composed of massive picture elements by using the median filter can be considerably shortened.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a median noise reduction processing system for multilevel digital images, and more particularly to a median noise reduction processing system using median values.

[0002]

2. Description of the Related Art Conventionally, a median noise reduction processing system of this kind collects a pixel and a level of an image surrounding the pixel and rearranges the levels of the pixels in ascending or descending order to obtain a median value. A processing method is performed in which the (median value) is obtained and the value of this median value is used as the value after noise reduction of a certain pixel as described above for the pixels of the entire image.

[0003]

In the above-mentioned conventional high-speed median noise reduction processing method, since the median value is calculated by rearranging a fixed number of pixels for all the pixels, the median noise reduction is reduced. It has the disadvantage of long processing time.

An object of the present invention is to have a median extraction table 301 in order to calculate a median value at high speed, and to use an initial table creation unit 101, a table updating unit 102, and a median value extraction unit 103 as processing units. A sensor having a function of creating a processed image 202 of which noise is reduced from the image 201, and a sensor included in the image of the image data represented by the multi-valued level of the image data recorded in the storage device of the computer. By performing the process of reducing the noise added by using the median filter at high speed, the above drawbacks are solved, and the noise reduction processing time by the median noise reduction process of image data composed of a large number of pixels is significantly increased. It is to provide a high-speed median noise reduction processing method that can be shortened.

[0005]

In the high-speed median noise reduction processing system of the present invention, data of image data read from an image input device such as a scanner and recorded in a storage device of a computer is 0 to n (n is 2 ^m -1, m is a natural number) In the median (median) noise reduction processing method of image data represented by multi-valued levels, noise reduction on image data (hereinafter referred to as original image) is reduced. An initial table creation unit that creates a median extraction table for a certain number of pixels from an image for which the median noise reduction process is performed, and a median value extraction unit that obtains a median value from the median extraction table created by the initial table creation unit. , Image data subjected to noise reduction processing using the median value obtained by the median value extraction unit (hereinafter referred to as a processed image). ) Creates a and a table updating unit for updating the median extraction table.

[0006]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram of a high-speed median noise reduction processing system according to an embodiment of the present invention, FIG. 2 is a block diagram of a median extraction table according to this embodiment, and FIG. 2A is a median extraction table according to this embodiment. Diagram of the summary table of Figure 2, Figure 2
(B) is a block diagram of a detailed table of the median extraction table of the present embodiment, FIG. 3 is a block diagram for explaining the processing of the initial table creating unit of the present embodiment, and FIG. 4 is a high-speed median noise reduction of the present embodiment. It is a flow chart of processing of a processing system.

In FIG. 1, the high-speed median noise reduction processing method of this embodiment comprises an initial table creating section 101, a median value extracting section 102, a table updating section 103,
It is composed of an original image 201, a processed image 202, and a median extraction table 301.

Here, the initial table creating section 101 picks up the pixel data required to newly create a median extraction table from the original image 201 and creates a median extraction table 301. The median value extraction unit 102 extracts a median value from the median extraction table 301 and stores the value in the processed image 202. The table updating unit 103 picks up the pixel data adjacent to the processed pixel from the original image 201 to obtain the median value for the adjacent processed pixel, and updates the median extraction table 301. Or later,
Median value extraction unit 102 for all images of the original image
And the processing of the table update unit 103 is performed, and the processed image 2
Create 02. When the processing is performed on all the pixels of the original image 201, this processing ends.

Next, details of the processing of the high-speed median noise reduction processing method of this embodiment will be described.

First, the details of the median extraction table 301 will be described with reference to FIG.

In FIG. 2, the median extraction table 3
01 is a summary table 302 and a detailed table 303
Composed of and.

Here, as the size of the table, the size value obtained by the square root of the data range of the pixel data is used, the summary table 302 is a 1 × size value table, and the detail table 303 is the size value. × It has a size value. Therefore, if the effective data range of the image data is 0 to 63 (2 ⁶ -1), the summary table 302 is a vertical 1 × horizontal 8 table, and the detailed table is a vertical 8 × horizontal 8 table.

In the following description, the size value is set to 8.

Next, details of the initial table creating unit 101 will be described with reference to FIG.

In FIG. 3, the initial table creating unit 101 of FIG. 1 is processed only once at the beginning.

The detail table creation processing unit 104 clears the contents of the detail table 303 (processing route 402),
The pixel level of 9 points of 3 pixels × 3 pixels 202 in the upper left corner of the original image 201 is read (processing route 401), and the quotient and the remainder divided by the size value (here, 8) are plotted on the vertical and horizontal axes of the detailed table 303. Then, the relevant part in the table is counted up (processing route 402). The summary table creation processing unit 105 clears the contents of the summary table 302, reads the contents of the detailed table 303 (processing route 403),
The sum of the number of level values corresponding to the level of the summary table is calculated, and the sum is stored in the summary table (processing route 404).

Next, the processing of the high speed median noise reduction processing method of this embodiment will be described with reference to FIG.

First, the median value extraction unit 102 will be described.

In FIG. 4, the median value extraction unit 102
Reads the summary table 302 of the median extraction table 301 already created by the initial table creation unit 101 (step 501) and the table update unit 103 in order from the top, and adds up the number of each level of the summary table 302. Go After the addition, if the result is not 5 or more, the number of the next level is added. If the added result is 5 or more, the summary table 30 that has become 5 or more
The detailed table 303 corresponding to the level value of 2 is read in order from the lowest level, the fifth data from the low level is searched, and the vertical level value of the detailed table 303 is found from the position of the table detected as the fifth data. The sum of the horizontal level values is calculated as a median value and set in the processed image (step 502).

Next, the table updating unit 103 will be described.

The table updating unit 103 uses the median extraction table 301 whose median value has already been calculated by the median value extraction unit 102 to calculate the median value of the adjacent pixel whose previous median value has been calculated. To update. The pixel level data necessary for creating the median extraction table 301 is 9 pixels, and 6 pixels out of them are the pixel data used for the previous calculation of the median value.
The 3 pixels that are unnecessary from 1 are deleted from the detail table 303, and the new 3 pixels necessary for the next median value calculation are stored in the detail table 303 (step 50).
4).

The storage method is similar to that of the initial table creating unit 101. After updating the detail table 303, the summary table 302 is created and updated in the same manner as the initial table creating unit 101.

The step 503 described above with reference to FIG. 4 is a step of determining the end of processing. Step 50
If the process of step 502 is performed on all the pixels of the image data in 3 (step 503, Yes), the process ends. If the process of step 502 has not been executed for all pixels of the image data (step 5
03, No), after carrying out the table updating unit 103 (Step 504), the median extracting unit (Step 502)
Return processing to again.

[0025]

As described above, according to the high-speed median noise reduction processing method of the present invention, the median value is calculated by using the median extraction table without using the rearrangement process (reit) to obtain the median value. High-speed processing for reducing noise added by a sensor or the like included in an image of image data represented by a multi-valued level of the image data recorded in the storage device using a median filter Thus, there is an effect that the noise reduction processing time by the median filter of the image data composed of a large number of pixels can be significantly shortened.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a high-speed median noise reduction processing system according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a median extraction table of the present embodiment. FIG. 2A is a configuration diagram of a summary table of the median extraction table of this embodiment. FIG. 2B is a configuration diagram of a detailed table of the median extraction table of this embodiment.

FIG. 3 is a block diagram for explaining processing of an initial table creating unit according to the present embodiment.

FIG. 4 is a flowchart of processing of a high-speed median noise reduction processing method according to this embodiment.

[Explanation of symbols]

 101 initial table creation unit 102 median value extraction unit 103 table update unit 104 detailed table creation processing unit 105 summary table creation processing unit 201 original image 202 processed image 301 median extraction table 302 summary table 303 detailed table 401 to 404 processing route

Claims (1)

[Claims] 1. Data of image data read from an image input device such as a scanner and recorded in a storage device of a computer is expressed by a multi-valued level of 0 to n (n is 2 ^m -1, m is a natural number). In the median (median) noise reduction processing method for image data that has been created, create a median extraction table that targets a certain number of pixels from the image for which you want to perform median noise reduction processing to reduce noise scattered on the image data. An initial table creating unit, a median value extracting unit that obtains a median value from the median extracting table created by the initial table creating unit, and an image that has been subjected to noise reduction processing using the median value obtained by the median value extracting unit. A high speed characterized by having a table updating unit for creating data and updating the median extraction table. Median noise reduction processing method.