JPH04205276A - Noise removing device and noise removing method for image processing - Google Patents

Abstract

PURPOSE:To allow the noise removal in real time with hardware constitution by subjecting the picture element information around reference picture elements to sequential processing by using an integer adding and computing means and decoding ROM. CONSTITUTION:The picture element information outputs the picture element information at the circumferential 6 points exclusive of the right and left of the reference point 35 and the reference point 35 to respective output terminals 28 to 33, 35 via shift registers 14, 15 of the number of the stages equal to the number of horizontal picture elements and two stages of shift registers 19 to 21. The decoding ROM 36 executes the adding and computing operation to output 1 when the added value is above a specified value and 0 when the value is below the specified value by adding the binarized values of the respective output terminals of the shift registers 19 to 21. Then, whether the respective picture elements are the signal picture elements or noise picture elements can be discriminated by utilizing the difference that the noise picture elements distribute discretely while the signal picture elements distribute two-dimensionally, linearly and continuously. The noise is removed in real time with the hardware constitution in this way.

Description

Detailed Description of the Invention [Field of Industrial Application] The present invention uses a television camera to sequentially scan an object irradiated with linear signal light in a direction perpendicular to the signal light. The present invention relates to an apparatus and method for removing randomly distributed noise in real time from a binarized two-dimensional television image when measuring a shape.

[Conventional technology]

In the past, noise was removed using software, such as by integrating multiple images, such as through averaging processing.

[Problem that the invention seeks to solve]

With conventional methods, it is difficult to remove noise in real time. The present invention aims to solve the above-mentioned drawbacks and to make it possible to remove noise in real time.

[Means to solve the problem]

The present invention makes use of the difference that signal pixels are continuously distributed linearly in two dimensions, while noise pixels are distributed discretely. This paper proposes a noise removal device with a node configuration, which performs sequential processing using .

〔Example〕

FIG. 1 is a diagram illustrating an object shape measurement method to which the present invention is applied. As shown in FIG. It measures the shape of an object by scanning in a direction perpendicular to the signal light.The linear signal light 2 is sequentially scanned, each received image signal is stored in a memory circuit, and these signals are output to the TV screen. It is measured by

It is preferable that the memory circuit used here has a memory configuration that matches the number of scanning pixels of the television screen. Now, to explain the case where one pixel is represented by an 8-bit luminance gradation and a memory circuit is used that stores 512 pixels x 512 pixels in accordance with the scanning of the television screen, each pixel 4 of the memory circuit has the following: If it corresponds to figure a, information about the signal pixel 5 as shown in figure b will be stored, but normally, in addition to signal pixels, noise pixels caused by clock noise of the video camera are also input and stored.

FIG. 2 is a diagram illustrating the noise removal device of the present invention.
An analog image signal input from an input terminal 6 is converted into a digital signal by an AD converter 7 and stored in a storage circuit 8. Next, noise pixels are removed from the signal binarized by the binarizer 9 by the signal discriminator 10, which is a feature of the present invention, and only the image signal is stored in the storage circuit 11, and the signal is output from the output terminal 12. It is structured in such a way that

Here, as described above, the memory circuit 8 is configured to store 512 pixels x 512 pixels according to the number of pixels of the television, and the binarizer 9 extracts the maximum value of change in the scanning direction. Use something that binarizes.

FIG. 3 is a diagram explaining the signal discrimination circuit 10 used in the present invention, in which two shift registers 14 and 15 are connected in series from an input terminal 13 to which binarized pixel information is input, and a shift register 14 and 15 are connected in series. Two-stage shift registers 19, 20 and 21 are connected to the input side 16 and output side 17 of the register 14 and the output side 18 of the shift register 15, respectively, and the shift register 19.21 has an input side 22.23 and an output side 24.2.
Output terminals 28, 31, 30, 33 and 29.32 are provided from the connection point 26.27 between the first and second stages of the shift register 20, respectively, and the connection point between the first and second stages of the shift register 20. 34 to A terminal 35 are provided, and the shift register 1
9.20.21 respective output terminals 28.29.30
．． 31, 32, 33 and 35 are manually input to a decode ROM 36, and an output terminal 37 is provided from the decode ROM 36.

Here, it is preferable to use a shift register 14, 15 with 512 stages, which is the number of stages corresponding to the number of horizontal pixels of the television screen.

To explain the operation of the above-mentioned signal discrimination circuit, pixel information is inputted sequentially from the input terminal 13 and is outputted through the shift registers 14 and 15 whose stages are equal to the number of horizontal pixels. When the pixel information stored at addresses 1, j) is output, the pixel information stored at addresses (1-1□J+1) and (i-1, j-1) are output to terminals 16 and 18, respectively. Output. From this, if it is assumed that the pixel information to be output to the output terminal 35 is stored at the reference point No., jo) as shown in FIG.
．． 30.31.32.33 respectively (io 1.
jo+1), (io, Jo+1), (io+1.ja
+1), (1o-1,j.-1), (io, jo
1) Output the pixel information stored in (io+1.j.-1), that is, the surrounding area 6 excluding the left and right sides of the reference point.
This will output the pixel information of the point.

Decode ROM36 is shift register 19.20.21
The binarized values of the respective output terminals are added, and when the added value is above a certain value, 1 is output, and when it is less than that, 0 is output.

Next, a method by which the signal discrimination circuit 10 removes noise from pixel information will be explained.

The present invention utilizes the difference that signal pixels are distributed continuously in a two-dimensional line, whereas noise pixels are distributed discretely, and performs an integer addition operation on pixel information around a reference pixel for each pixel. This is to determine whether a pixel is a signal pixel or a noise pixel. That is, when the binary information of the output terminal 35 shown in FIG. 3 is 0, it is not a signal pixel, but when it is 1, it is determined whether it is a signal pixel or a noise pixel by the following procedure. Since signal pixels are continuously distributed, if the reference point 35 shown in FIG. 4 is pixel information, the surrounding points 28.2
Any one of 9, 30, 31, 32, and 33 is always 1. On the other hand, in the case of a noise pixel, the above guarantee does not apply, and if at least all of the surrounding points are 0, the pixel becomes a noise pixel. The optimum threshold value, which is a value for determining whether a pixel is a signal pixel or a noise pixel in the decoding ROM 36, may be determined experimentally.

Here, the pixel information of the left and right points of the reference point 35 is not added because the shape is measured by irradiating a linear signal light and scanning in a direction perpendicular to the signal light, so even if the reference point is 1, In this case, the left and right points are 0.

As is clear from the above explanation, the AD used in the present invention
The converter 7, the memory circuit 8, the binarizer 9, and the memory circuit 11 can all be used with conventional functions, so they may be used.

FIG. 5 shows an example of signal processing using the signal discriminator shown in FIG. 3, in which figure a shows an image before processing and figure b shows an image after processing.

〔Effect of the invention〕

Since the present invention is configured as described above, it is possible to remove noise in real time using a hardware configuration.

[Brief explanation of the drawing]

FIG. 1 is a diagram explaining an object shape measurement method to which the present invention is applied, FIG. 2 is a diagram showing an embodiment of the present invention, FIG. 3 is a diagram showing the configuration of a signal discriminator used in the present invention, and FIG. FIG. 4 is a diagram showing the storage position of each pixel in the memory circuit, and FIG. 5 is a diagram showing the processing according to the present invention. 1: Object to be measured, 2: Signal light, 3: Video camera, 4:
pixel, 5: signal pixel, 6: input terminal, 7: AD converter,
8, 11: Memory circuit, 9: Binarizer, 10: Signal discriminator, 12: Output terminal, 13: Input terminal, ICl3: Shift register, 16, 17.18: Shift register terminal,
19, 20° 21: 2-stage shift register, 22, 23
, 24.25, 26, 27, 34: 2-stage shift register terminals, 28, 29, 30, 31, 32, 33. 35
: Output terminal, 36: f code)'ROM, 37: Output terminal

Claims (2)

[Claims] (1) A means for AD converting an analog image signal, a means for extracting the maximum value of change from the digitized signal and binarizing it, and determining whether the binarized signal is a signal pixel or a noise pixel. and a signal discriminating means for outputting the binarized signal of the reference point as well as the binarized signal of the surrounding points of the reference point, and a means for outputting the binarized signal of the surrounding points of the reference point; 1. A noise removal device for image processing, comprising: arithmetic processing means for comparing a value with a threshold value. (2) Using the binarized pixel signal, add the binarized signal of the reference point for determining whether it is a signal pixel or a noise pixel and the binarized signal of the surrounding points of the reference point, and the added value is the threshold value. A noise removal method characterized in that if the pixel is above a threshold value, it is determined to be a signal pixel, and if it is smaller than a threshold value, it is determined to be a noise pixel.