JPH01292583A - Endoscope picture processor - Google Patents

Abstract

PURPOSE:To reduce the influence of the noise component ant to improve the detecting sensitivity by converting the data on a certain noticeable picture element into the value calculated with use of the non-contiguous picture elements against to at least one of plural video signals. CONSTITUTION:An accumulating/multiplying device 16 multiplies the picture element data obtained around each noticeable picture element read out of a main memory by the coefficient corresponding to each noticeable picture element read out of a coefficient ROM 15 and at the same time performs the cumulative addition of said picture element data and coefficients to send the result of the addition to an abnormal value correcting circuit 17. The circuit 17 checks whether the calculated value is suited to the display or not and performs the correction. While the correct arithmetic operation is impossible owing to the omission of the picture element data in case the picture elements exist around a picture. Thus a peripheral part detecting circuit 18 detects those picture elements via a read address counter 13. Then the data for peripheral parts are read out of a ROM 19 and supplied to a data selector 20. The selector 20 selects the output received from the circuit 17 and delivers it to a sub- memory 21 as the picture data in case the correct arithmetic operation is carried out to the noticeable picture element.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] - The present invention relates to an endoscopic image processing device configured to be less susceptible to noise components.

[Conventional technology]

Conventionally, images from an endoscope were decomposed into multiple color signals to generate electricity.
Examples of endoscopic image processing devices that convert into signals include:
A configuration as shown in FIG. 9 has been proposed. That is, in FIG. 9, 101 is a COD for imaging a biological observation site, and is arranged at the tip of the endoscope insertion section. 102
is an amplifier 10 that amplifies the output signal of the CCD 101;
3 is a γ correction section, 104 is an A/D converter, and 105 is a changeover switch.

106, 107, and 108 are R, G, and B color image memories connected to the output side of the changeover switch 105;
09゜1, 0.111 is the image memo January 06. ．． 107
．． ．． This is a D/A converter connected to 10B. 11
5. , 116 and 117 are contour enhancement circuits, respectively D/
, A changeover switch 112 . ．． 113. , , 14. Reference numeral 118 denotes a control signal generator, which is connected to the selector switch 10.
5, R, G, 8 image memory 106°107,
108. D/A converter 109.110.111
, changeover switches 112, 113, and 114, a synchronizing signal generating circuit 119°, and a motor 121 that drives an RGB rotary filter 120. Reference numeral 122 denotes a light source lamp, and the light from the lamp 122 passes through the filter 120.
The light is irradiated onto the light guide 123 through the light guide 123, guided to the distal end of the endoscope insertion section, and illuminated in a field-sequential color manner.

In the endoscopic image processing device configured in this way, the illumination light from the lamp 122 is transmitted through the RGB rotating filter 120.
The light is then separated into the three primary colors of R, G', and B, and is sequentially incident on the light guide 123 to illuminate the living body. The observed image of one body is converted into an electrical signal by C'CD10L, inputted to the amplifier 102, amplified to a voltage level within a predetermined range, and then inputted to the γ correction unit 103 for γ correction. The γ-corrected signal is A/D converted by the A/D converter 104 and then enters the changeover switch 105, where it is sequentially switched by a control signal and sent to each R, G, and B image memory 106, 107.
．． Sequentially recorded Sarel on 108th. Each R', G, B image;
E + J 106: The image signals recorded in 1o7 and 1o8 are sequentially called out by the control signal from the control signal generation unit 118, and are sent to each D/A converter 1[)9.110
．． ．． D/A conversion is performed at step 111.

The image signal, which has become an analog signal, passes through edge enhancement circuits 115, 116, and 117 as necessary, and is then sent to the RGB image signal output terminal together with the synchronization signal from the synchronization signal generation circuit 119. The RGB image signals obtained in this manner are displayed on a TV monitor for endoscopic observation.

FIG. 10 is a prong diagram showing an example of the configuration of the edge enhancement circuit. The input signal (A) is the first delay line 20
The first and second delay lines 202 each delay the signal by one pixel. The output □ signal (C) from the second delay line 202 is delayed by '2 pixels □.
The input signal (A) is added to the adder 203'7 to obtain an output signal (D), and this output signal (D) is sent to the 1/2 inverter 2.
04, the signal is halved and then inverted to obtain an output signal (E). Next, this output signal (E) and the first
The output signal (B) of the delay line 201 and the adder 20
By adding 5, the contour ``strong'' precept (F) is obtained. Next, this contour enhancement component (F) is applied to a multiplier 20.
By amplifying it to a predetermined magnitude in step 6 and adding it to the output signal (B) of the '1st delay line in adder 207,
An output signal (G) with enhanced contours is obtained.

Furthermore, in the past, many methods have been proposed in which not only the above-mentioned contour emphasizing circuit performs the emphasis processing using hardware, but also the method using the software. Go to Figure 11~
(C) is an explanatory diagram regarding the Laplacian method as an example thereof. In other words, the 11th diagram defines the pixel of interest as A+=(1
, j is an integer greater than or equal to 1), and the pixel of interest A and its adjacent pixels are
The coefficients of the 3 Laplacian matrices are multiplied together. The total sum is then determined and used as the value of the pixel of interest. Through this processing, the contour component can be extracted.

Note that FIG. 11(C) shows an output image processed using the Laplacian matrix, and the output pixels of interest are designated as B,
, it is expressed as follows.

B+J=(0>・A f4-11 +j4I) +(-
1)・A i I,. 1゜+(0)・A, 1゜n fj
++1 + (i)・A f+ -1') j+(4
)・AtJ+(-1)・At+, )10(O)・A
(1-11<i-n ten(' 1 )・At(j-11
+(0) 'A"fi++l+J'-11 [Problem to be solved by the invention] By the way, in the conventional contour enhancement processing means, "
As described above, the contour component is extracted using the pixel of interest and its neighboring pixels. For this reason, there is a problem in that it is easily influenced by noise components and reacts more strongly to fine line components and isolated points than to contour components.

In addition, endoscopic images provide useful diagnostic information. Mucosal structures and blood vessel images have relatively gentle gradients, but
In order to extract such a structural pattern, the conventional contour enhancement processing method using adjacent pixels has a problem in that the detection sensitivity is insufficient.

The present invention was made in order to solve the above-mentioned problems in endoscopic image processing devices using conventional contour enhancement means. The object is to provide an image processing device with high detection sensitivity.

[Means and operations for solving the problems] In order to solve the above problems, the present invention provides an endoscopic image processing device that decomposes images from an endoscope into a plurality of video signals and converts them into electrical signals. In this method, image processing is performed on at least one of the plurality of video signals by converting data of a certain pixel of interest into a value calculated using at least information of pixels that are not adjacent to the pixel of interest.

By extracting contour components using information about pixels that are not adjacent to the pixel of interest in this way, it becomes less susceptible to noise components and allows the extraction of structural patterns with relatively gentle gradients in mucous membrane structures, blood vessel images, etc. becomes possible.

〔Example〕

Examples will be described below. FIG. 1 is a schematic diagram showing the overall configuration of an example of the configuration of an endoscope apparatus to which an endoscopic image processing apparatus according to the present invention is applied. In the figures, ■ is an endoscope whose main body 1a is connected to an observation device 2 including an image processing device, and whose insertion portion 1b is designed to be completely inserted into a living body 3. Observation device 2 includes a TV for observation.
A monitor 4 is connected, and an accessory device 5 such as a suction device is connected to the endoscope main body 1a. Illumination light is then supplied to the tip of the endoscope insertion section 1b, and the observation image obtained by illuminating the living body 3 is converted into an electrical signal by an imaging device such as a CCD placed at the tip of the insertion section 1b. The electrical signal is processed in the observation device 2, converted into a TV signal, and displayed on the TV monitor 4 for endoscopic observation.

FIG. 2 is a block diagram of the first embodiment of the endoscopic image processing apparatus according to the present invention. In the figure, numeral 11 is an A/D converter that converts an analog video input signal into a digital signal, and the video signal digitized by this A/I) converter 11 is recorded in the main memory 12. 13 is a read address counter which is a counter for reading out the pixel of interest on the main memory 12; 14 is a matrix counter; in order to perform arithmetic processing on the pixel of interest, as shown in FIG. It generates the address of a matrix of size (9×9) centered on . 15 is a coefficient ROM for storing each coefficient of the matrix of size (!JX9) shown in FIG.
9) is an accumulation multiplier that cumulatively multiplies the pixel data in the range of 9) and the coefficients of the coefficient ROM 15, respectively. Reference numeral 17 denotes an abnormal value correction circuit that corrects the value when the calculation result of the cumulative multiplier 16 is outside a predetermined dynamic range. 18 is a peripheral part detection circuit that detects the peripheral part of the video data stored in the main memory 12, and 19 is a peripheral part data ROM that generates the video data of the peripheral part. 20
is a data selector that selects the video data output from the abnormal value correction circuit 17 that has been subjected to arithmetic processing, and R for peripheral data.
The peripheral data output from the OM 19 is switched based on the detection signal from the peripheral detection circuit 18. 21 is a sub-memory for storing the video data selected by the data selector 20; 22 is the sub-memory 2;
D/ which converts the video data stored in 1 into an analog signal.
It is an A converter. - Next, the operation of the endoscopic image processing apparatus configured as above will be explained. First, a video input signal for image processing is converted from an analog signal to a digital signal by the A/D converter 11, and is recorded in the main memo January 2. When the readout address counter 13 specifies the address of the pixel of interest, the image data recorded in the main memory 12 is read out using the matrix counter 14, as shown in FIG. (9×9) By sequentially specifying the addresses of pixels on all sides, pixel data is read out from the main memory 12 and input to the cumulative multiplier 16, and at the same time, each pixel data is read out from the coefficient ROM 15 as shown in FIG. The coefficients for each position are read out and input to the cumulative multiplier 16.

The cumulative multiplier 16 multiplies (9 x 9) pixel data read from the main memo January 2 by the coefficient corresponding to each pixel data of interest read from the R9M15 for coefficients. Cumulative addition is performed and the resulting value is output to the abnormal value correction circuit 17. This abnormal value correction circuit 17 detects whether or not the value calculated by the cumulative multiplier 16 is within the dynamic range when displaying. Correct the value.

On the other hand, if the pixel of interest is located at the periphery of the image, pixel data around the pixel of interest is missing, so correct calculations will not be performed. Therefore, the peripheral part is detected by the peripheral part detection circuit 18 based on the address value of the read address counter 13, data for the peripheral part is read from the peripheral part data ROM 19, and inputted to the data selector 20. Here, the data selector 20 selects the output from the abnormal value correction circuit 17 when the correct arithmetic processing is performed on the pixel of interest, and selects the output from the abnormal value correction circuit 17 when the correct arithmetic processing is not performed on the peripheral part. Data from the ROM 19 is selected and output to the sub-memory 21. The image data recorded in the sub-memory 21 is outputted as an analog video signal by the D/A converter 22.

Note that the matrix for image processing is not limited to the Laplacian (9'X'9) size matrix shown in FIG. 3, but the conventional ( By recording the coefficients of various matrices, such as a Laplacian matrix of 3X 3) size, a single blow differential matrix in the X direction, and a single blow differential matrix in the Y direction, in the coefficient ROM, and by providing several types of matrix counters. , similar image processing using various matrices becomes possible. In addition, Fig. 5 (8) is shown in Fig. 4 [01 (3 x 3
) is a general example of a matrix of size MXN (M, N
is an odd number of 5 or more and a multiple of 3). Also, in Figure 5 (marks indicate (M-3) 72 zeros between each element in the X direction of the matrix of the general example, and (N'-3>72 zeros) between each element in the X direction. An example of the structure of a matrix expanded by arranging zeros is shown.In FIG. 5(B), the helmet and N are odd numbers of 5 or more.

Further, as the matrix used for the above-mentioned arithmetic processing, it may be configured such that an optimal matrix can be selected depending on the viewing angle at the time of observation and the type of lesion.

Furthermore, the arithmetic processing described above does not necessarily need to be performed on the entire image, and may be configured to be performed only on the required area.

FIG. 6 is a block diagram showing a second embodiment of the present invention, and the same members as those in the first embodiment shown in FIG. 2 are denoted by the same reference numerals.

The image processing device of this embodiment includes an A/I) converter 11
and 1 digitized by the A/D converter 11.
A frame memory 31 that records video signals for one frame, a read address counter 13 that is a counter for reading out the pixel of interest on the frame memory 31, and a counter as shown in FIG. 7 for processing the pixel of interest. In addition, there is a line memory address counter 32 that controls the acquisition of 21 lines of data including data in a matrix of size (21×21) centered on the pixel of interest;
In-memory 33-1 to 33-3 that stores data for lines
3-21, a matrix horizontal counter 34 that generates an address in the X direction of a (21×2.1) matrix centered on the pixel of interest as shown in FIG. ) and a coefficient ROM 35 for storing each coefficient of the ° matrix, and the other parts are constructed in the same manner as in the first embodiment.

Next, the operation of the second embodiment configured as described above will be explained. First, a video input signal for image processing is converted from an analog signal to a digital signal by the A/D converter 11, and is recorded in the frame memory 31. . When the readout address counter 13 specifies the address of the pixel of interest, the image data recorded in the frame memory 31 is read out by the line memory address counter 32 so that the image data recorded in the frame memory 31 is centered around the pixel of interest as shown in FIG. By sequentially specifying the addresses of 21 lines including pixels on all four sides, line memory 33-1 is transferred from frame memory 31.
Image data is recorded to 33-21.

The image data recorded in the line memories 33-1 to 33-21 can be read by sequentially specifying pixel addresses in the horizontal direction from the matrix horizontal counter 34 using the read address counter 13.
The pixel data is read from 33-21, and the cumulative multiplier 16
At the same time, coefficients for each pixel position as shown in FIG. 7 are read out from the coefficient ROM 35 and input to the cumulative multiplier 16. The cumulative multiplier 16 receives (21 x 21) pixel data centered around the pixel of interest read out from the line memories 33-1 to 33-21, and R for coefficients.
Each pixel data of interest read out from the OM 35 is multiplied by a corresponding coefficient, cumulative addition is performed, and the resulting value is output to the abnormal value correction circuit 17. The subsequent operations are performed in the same manner as in the first embodiment, and from the D/A converter 22,
An image-processed analog video signal is output.

Also in this example, the type of matrix is not limited to the structure shown in FIG. 7, and the size of the matrix shown in FIG. 4 (8), (B), and (C) In cases where various types of expanded matrices can be used, line memories can be used in accordance with the size of the matrix, and if the matrix size is even larger, expansion is possible.

Each of the above embodiments shows a configuration that processes one signal as an input signal, but it is also possible to provide three such circuits and process each of the R-G-B signals for color display. It may be configured as follows.

In addition, it can be applied not only to R-G-B signals but also to complementary color signals such as cyan, magenta, yellow, etc., and can also be applied to chromaticity coordinate systems such as saturation, hue, brightness, and CIB standards. It is also applicable to signals. Further, it is also possible to use a composite signal such as the NTSC standard for a signal in which the luminance component and color component are separated.

Further, the present invention is not limited to processing by hardware as shown in the first and second embodiments, but also processing by software as shown in the flowchart of FIG. 8, for example. To briefly explain this flowchart, 41
42 is a step of selecting the type and size of the matrix used for the conversion process according to the angle of view at the time of observation of the input image, the type of lesion, etc., 42 is a step of extracting the processing area of the input image, and 43 is a step This is a step of determining whether the area extracted in step 42 is a peripheral area. 44 is a step in which matrix processing is performed using the matrix selected in the matrix selection step 41 for the extraction region determined to be not a peripheral region in the peripheral region detection step 43, and 45 is a step in which the matrix processing is performed on the entire region. This is the step of determining whether or not the 4
6 is after the matrix processing is completed for all areas,
In the step of detecting the maximum and minimum values among the processed pixel values, 47 is a step in which the detected maximum and minimum values are abnormal values outside the dynamic range when displaying - values within the dynamic range. This is the step of converting so that

Note that in the process shown in this flowchart, matrix processing is not performed when a peripheral part is detected in the processing area, but as in the embodiment shown in FIG. 2 or FIG. When the peripheral portion is detected, data may be read from the peripheral portion data ROM and image processing may be performed.

〔Effect of the invention〕

As described above based on the embodiments, the present invention extracts a contour component from at least one of a plurality of video signals constituting an image by using information on pixels that are not adjacent to the pixel of interest. This structure makes it less susceptible to noise components and less sensitive to fine line components, isolated points, etc. Furthermore, in medically important mucous membrane structures, blood vessel images, and the like, detection sensitivity for structural patterns having relatively gentle gradients can also be increased.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram showing a configuration example of an endoscope apparatus to which an endoscopic image processing apparatus according to the present invention is applied, and FIG. 2 is a block diagram showing a first example of the present invention. 3 is a diagram showing the configuration of a (9x9) size matrix used for image processing in the first embodiment, and the fourth diagram is (Bl), (C) is a conventional (3x3) size matrix A diagram showing a configuration example,
Figure 4 (O1]) is a diagram showing a general-purpose example of a (3X3) size matrix, and Figures 5 (C) and (G)) are diagrams showing other matrix configuration examples used in the image processing of the first embodiment. 6 is a block diagram showing the second embodiment of the present invention, and FIG. 7 is a block diagram showing the second embodiment of the present invention (21
X 21) Diagram showing the structure of the size matrix, No. 8
The figure shows a flowchart when contour component extraction is processed by software, FIG. 9 is a block diagram showing a conventional endoscopic image processing device, and FIG.
A diagram showing an example of the configuration of the contour enhancement circuit in the figure, 11th National. (C1 is a diagram showing the pixels of the input and output image, Figure 11 (81
1 is a diagram showing the configuration of a Laplacian matrix used for processing an input image. In the figure, the operator is an endoscope, 2 is an observation device, 3 is a living body, and 4
5 indicates a TV monitor, and 5 indicates an attached device. “Patent applicant: OrigiPath Optical Industry Co., Ltd. Figure 1 Figure 4 (A) (B) (C) (D) Figure 5 (A) Figure 5 (B)

Claims (1)

[Claims] 1. In an endoscopic image processing device that decomposes an image from an endoscope into a plurality of video signals and converts them into electrical signals, data of a certain pixel of interest is converted to at least one of the plurality of video signals, An endoscopic image processing device that converts into a value calculated using at least information on pixels that are not adjacent to the pixel of interest.