JP2007050110A - Image processor and electronic endoscope system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To apply a noise elimination processing to a digital image signals without degrading image quality. <P>SOLUTION: A preceding stage signal processing circuit 42, a noise reduction filter circuit 46 and a stage division circuit 48 are provided inside a processor 30. In the stage division circuit 48, respective pixels are distributed to first to fourth luminance stages corresponding to a luminance level on the basis of the digital image signals and an object image is divided into multiple stages. Then, in the noise reduction filter circuit 46, moving average filtering is executed to the pixels of the respective stages so as to increase the size of a vicinity pixel area from the fourth luminance stage to the first luminance stage. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to image processing of a digital image signal obtained by photographing, and particularly relates to noise removal processing.

In conventional digital image processing, filtering is performed to remove noise. For example, noise is removed by a digital filter such as a spatial frequency processing filter such as a moving average filter or a median filter, or a cyclic filter. For example, in the case of an electronic endoscope apparatus, a noise reduction circuit is provided to remove amplified noise, and noise removal processing is uniformly performed on the entire image (see Patent Document 1). In addition, when removing isolated noise such as white flaw noise that is a pixel defect of the image sensor, filtering is performed on the pixel of interest (see Patent Document 2).
JP-A-10-57312 JP 2004-313413 A

  In an endoscopic image or the like obtained by photographing the inside of a lumen, bright and dark portions are mixed on the same screen. Noise is conspicuous in dark areas, but noise is inconspicuous in bright areas.If noise removal processing is performed uniformly on the entire image, the image of bright areas with less noise is blurred, resulting in overall image quality. Decreases. In addition, in the case of random noise, no effect is obtained even if noise removal processing specifying a pixel is performed.

  The image processing apparatus of the present invention can perform appropriate noise removal processing according to a bright image portion where noise is conspicuous and an image portion where noise is not conspicuous, and can be applied to a photographing device such as a digital camera or an electronic endoscope device. is there. The image processing means includes digitization processing means, luminance stage dividing means, and filtering means. The digitization processing means generates a digital image signal corresponding to the subject image composed of a plurality of pixels. The luminance stage dividing means divides the subject image into a plurality of stages according to the luminance level of each pixel based on the digital image signal. Here, the plurality of stages means that a plurality of ranges (stages) are defined with a predetermined brightness level as a boundary in the range of the entire brightness level, and each pixel is divided into stages to which the brightness level belongs. For example, the range that the brightness level can take may be divided into a plurality of stages such as 4 stages or 8 stages in order of size, and each pixel may be assigned to a stage corresponding to the brightness level.

  The filtering means of the present invention filters the digital image signal at each stage so that the noise removal effect becomes greater in the image portion corresponding to the stage with the lower luminance level. A noise removal process having a relatively high filter strength is applied to a low luminance level, that is, a dark part of an image, and random noise and the like are effectively removed. On the other hand, noise removal processing with relatively low filter strength is performed on bright portions of the image, and the image quality is maintained without blurring the image.

  For example, in the case of a smoothing filter that performs noise removal processing using neighboring pixels, such as a moving average filter or a median filter, the number of neighboring pixels may be increased, or the number of repetitions of filtering may be increased.

  In the case of an acyclic digital filter composed of a delay element and an adder, the corresponding digital image signal may be sequentially delayed and added to the image portion corresponding to the lower luminance level. Alternatively, in the case of a cyclic filter composed of a delay element and an adder, the weighting coefficient of the delayed digital image signal is made larger in the image portion corresponding to the lower luminance level than the weighting coefficient of the corresponding digital image signal. And filter it.

  In order to erase minute image information such as capillaries before classifying the subject image, it is preferable to provide a high frequency filter for removing high frequency components of the digital image signal generated by the digitization processing means. Alternatively, if a color signal corresponding to red (R) in the long wavelength region is used, fine image information is not included so much, so the digitization processing means generates a digital R signal corresponding to red (R). However, the stage division means may perform the stage division based on the digital R signal.

  An electronic endoscope apparatus according to another aspect of the present invention is a digital scope that generates a digital image signal corresponding to a subject image that is formed on a video scope including an image sensor and includes a plurality of pixels. Based on the processing means and the digital image signal, the luminance stage dividing means for dividing the subject image into a plurality of stages according to the luminance level of each pixel, and the image portion corresponding to the stage with the lower luminance level so that the noise removal effect is increased. And filtering means for filtering the digital image signal at each stage. An image processing method, which is another feature of the present invention, generates a digital image signal corresponding to a subject image formed of a plurality of pixels formed on an image sensor, and adjusts the luminance level of each pixel based on the digital image signal. Accordingly, the subject image is divided into a plurality of stages, and the digital image signal is filtered at each stage so that the noise removal effect becomes larger in an image portion corresponding to a stage with a lower luminance level.

  According to the present invention, noise removal processing can be performed on a digital image signal without degrading image quality.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram of an electronic endoscope apparatus according to the first embodiment.

The electronic endoscope apparatus includes a video scope 10 and a processor 30, and a monitor 70 is connected to the processor 30. The video scope 10 is detachably connected to the processor 30.

  The light emitted from the lamp 32 passes through the diaphragm 34 and the condenser lens 36 and enters the incident end 12A of the light guide 12 provided in the video scope 10. The light guide 12 is an optical fiber bundle that transmits light from the lamp 32 to the distal end portion of the scope, and the light that has passed through the light guide 12 is emitted from the exit end 12B of the light guide 12 and passes through the light distribution lens 14. The light is emitted from the tip. Thereby, light is irradiated to an observation site.

  The light reflected at the observation site passes through the objective lens 16 and reaches the CCD 18, whereby a subject image is formed on the light receiving surface of the CCD 18. In the CCD 18, an image signal corresponding to the subject image is generated, and based on the clock pulse signal sent from the CCD driver 38, the analog image signal is read out at a predetermined time interval according to the TV standard such as the NTSC system. The read image signal is sent to the pre-stage signal processing circuit 42 of the processor 30.

  In the pre-stage signal processing circuit 42, various signal processing such as white balance processing and γ processing is performed on the read analog image signal, and the analog image signal is converted into R, G, and B digital image signals. The digital image signal is sent to the noise reduction filter circuit 46 together with the high frequency removal filter 44.

  The high frequency removal filter 44 removes high frequency components of the digital image signal. Then, in the stage division circuit 48, as will be described later, the image signal from which the high frequency component has been removed is staged in accordance with the luminance level. In the noise reduction filter circuit 46, filtering is performed for each of the divided images, and noise removal processing is performed on random noise, white scratch noise, and the like. Here, a moving average filter is used.

  The digital image signal from which noise has been removed is converted into a video signal in accordance with the video standard by the subsequent signal processing circuit 50. The video signal is output to the monitor 70, whereby the observation image is displayed on the monitor 70.

  FIG. 2 is a diagram showing an endoscopic image. FIG. 3 is a diagram showing endoscopic images that are divided according to the luminance level. FIG. 4 is a diagram illustrating a neighboring pixel area of the filtering target pixel.

  The endoscopic image in FIG. 2 is an image obtained by photographing the bronchus, and the left main bronchus and the right main bronchus are projected on the left side and the right side of the endoscopic image, respectively, with the tracheal branch portion interposed therebetween. The tracheal inner wall close to the scope tip and the tracheal branch located in the center of the image are reflected brightly and are therefore brightly projected. On the other hand, since the images of the left main bronchus and the right main bronchus are images inside the lumen, the light is not reflected so much and appears dark.

  In the stage division circuit 48 of the processor 30, the endoscopic image is divided into four stages according to the luminance level. The luminance level of each pixel is represented by, for example, 256 levels, and takes any value from 0 to 255. Then, the range of the luminance level from 0 to 63 is the first luminance stage, the range of the luminance level from 64 to 127 is the second luminance stage, the range of the luminance level from 128 to 191 is the third luminance stage, and the luminance level is from 192 to 255. Is defined as the fourth luminance stage.

  FIG. 3 shows an image when each pixel is divided into four stages according to the luminance level. The image portions near the tracheal branch correspond to the third and fourth luminance stages, and the image portions near the left main bronchus and the right main bronchus correspond to the first and second luminance stages. Here, I1 to I4 represent pixel collection portions corresponding to the first to fourth luminance stages, respectively.

  In the noise reduction filter circuit 46 of the processor 30, each pixel is filtered by a moving average filter. That is, the value of the pixel to be filtered is replaced with the pixel average value of the area including the neighboring pixels of the pixel and output. At this time, as shown in FIG. 4, the neighboring pixel area of the filtering target pixel is defined as an area of (2n + 1) × (2n + 1) (n = 0, 1, 2, 3), and has the above-described luminance level. The value of n varies depending on the corresponding stage. For pixels in the fourth luminance stage, an average value of 9 pixels is calculated by setting the peripheral pixel area to an area of 3 × 3 (n = 1). For the pixels in the third luminance stage, the peripheral area is determined as an area of 5 × 5 (n = 2), and an average value of 25 pixels is determined. Then, the peripheral area of the pixel in the second luminance stage is defined as an area of 7 × 7 (n = 3), and the peripheral area of the pixel in the first luminance stage is 9 × 9 ( n = 4). Such a noise removal process using the moving average filter is executed for each pixel.

  In the noise removal processing by the moving average filter, the effect of smoothing, that is, noise removal becomes larger as the neighborhood pixel area for calculating the average value is increased. If “n” that determines the effect of noise removal, that is, smoothing, is considered as a parameter of the filter strength, a noise removal process having a higher filter strength is applied to a pixel in a darker image portion.

  As described above, according to the present embodiment, the pre-stage signal processing circuit 42, the noise reduction filter circuit 46, and the stage division circuit 48 are provided in the processor 30. In the stage division circuit 48, each pixel is divided into first to fourth luminance stages according to the luminance level based on the digital image signal, and the subject image is divided into multiple stages. In the noise reduction filter circuit 46, the moving average filter is applied to the pixels in each stage so that the size (filter strength) of the neighboring pixel area increases from the fourth luminance stage to the first luminance stage. .

  Note that another spatial filter such as a median filter may be used instead of the moving average filter. Further, the stage division is not limited to four stages, and may be divided into a plurality of stages such as eight stages. Further, the present invention may be applied to a photographing apparatus such as a camera other than the electronic endoscope apparatus.

  Next, the electronic endoscope apparatus according to the second embodiment will be described with reference to FIGS. In the second embodiment, the number of repetitions of filtering is adjusted as the filter strength, and the image is divided into stages based on a signal corresponding to red (R). Other configurations are the same as those in the first embodiment.

  FIG. 5 is a block diagram of an electronic endoscope apparatus according to the second embodiment. FIG. 6 is a diagram illustrating a spatial filter according to the second embodiment.

  The upstream signal processing circuit 42 ′ of the processor 30 ′ separates the R signal from among the R, G, and B digital signals and sends it to the stage division circuit 48. In the second embodiment, no high-frequency removal filter is provided, and filtering is performed based on an R signal with relatively little information having a high spatial frequency such as a capillary.

  In the noise reduction filter circuit 46 ′, the neighborhood complaint area of the pixel to be filtered is fixed, for example, a 3 × 3 pixel area. On the other hand, as shown in FIG. 6, the number m of repetitions of filtering is adjusted. Here, the lower the luminance level, the larger the number of repetitions m is set, and “m” is a parameter for the filter strength. For example, the number of repetitions m = 4, 3, 2, 1 is set for the pixels having the luminance levels of the first, second, third, and fourth luminance stages, respectively. In accordance with the set repetition count m, noise removal processing by the moving average filter is executed.

  Next, an electronic endoscope apparatus according to a third embodiment will be described with reference to FIG. In the third embodiment, a non-recursive filter that is a digital filter is applied for noise removal processing. About another structure, it is the same as 1st Embodiment.

  FIG. 7 is a block diagram of a filter processing unit in the noise reduction / removal filter circuit according to the third embodiment.

The noise reduction filter circuit 46 ″ is composed of four filter processing units according to the first to fourth luminance stages, and one filter processing unit is shown in FIG. , L (l = 1, 2, 3,...) Frame memories K _l , and delay data output from a predetermined frame memory K _l is a frame memory K _l adjacent to the adder TM. _The delay data output from the frame memory K _l is delayed by 1 frame with respect to the original data, and the delay data delayed by 1 to 1 frames from the original data. Are weighted averaged by the adder TM to remove noise.

  As the weighted average of the original data and the delay data is repeated, the noise removal effect increases, and the number of frame memories becomes a filter strength parameter. Here, a filter processing unit composed of four frame memories for the pixels in the first luminance stage, and a filter processing unit composed of three frame memories for the pixels in the second luminance stage The filter processing unit composed of two frame memories for the pixels in the third luminance stage, and the filter processing unit composed of one frame memory for the pixels in the fourth luminance stage. A removal process is performed.

  Next, an electronic endoscope apparatus according to a fourth embodiment will be described with reference to FIG. In the fourth embodiment, a cyclic time filter is applied for noise removal processing. About another structure, it is the same as 1st Embodiment.

  FIG. 8 is a block diagram of a filter processing unit in the noise reduction / removal filter circuit according to the fourth embodiment.

    The noise reduction filter circuit 46 ″ ′ is composed of four filter processing units according to the first to fourth luminance stages, and one filter processing unit is shown in FIG. Is provided with an adder TN and one frame memory M, output data from the adder is input to the frame memory, and delay data delayed by one frame is input to the adder TN.

In the adder TN, the ratio between the original data and the delay data is adjusted based on the following calculation formula.

DATA3 = α × DATA1 + β × DATA2 (1)
However, α + β = 1

Here, DATA3 indicates data output from the adder TN, DATA1 indicates original data, and DATA2 indicates delay data output from the frame memory M. Α and β represent the ratios of the original data and the delay data, respectively.

  By changing the values of α and β, the ratio occupied by the original data and the ratio occupied by the delay data are adjusted. As β is larger than α, the effect of noise removal increases and the strength of the filter increases. Here, β is set to be larger than α in the order of the fourth luminance stage, the third luminance stage, the second luminance stage, and the first luminance stage (for example, the fourth luminance stage, the second luminance stage, Β = 0.8, 0.6, 0.4, and 0.2 in the order of the three luminance steps, the second luminance step, and the first luminance step).

It is a block diagram of the electronic endoscope apparatus which is 1st Embodiment. It is the figure which showed the endoscopic image. It is the figure which showed the endoscopic image divided into steps according to the luminance level. It is the figure which showed the neighborhood pixel area of the pixel for filtering. It is a block diagram of the electronic endoscope apparatus which is 2nd Embodiment. It is the figure which showed the spatial filter in 2nd Embodiment. It is a block diagram of the filter process part in the noise reduction filter circuit in 3rd Embodiment. It is a block diagram of the filter process part in the noise reduction filter circuit in 4th Embodiment.

Explanation of symbols

10 Videoscope 30 Processor 42 Pre-stage signal processing circuit (digitization processing means)
46 Noise reduction filter circuit (filtering means)
46 'noise reduction filter circuit 46 "noise reduction filter circuit 46"' noise reduction filter circuit 48 stage division circuit (luminance stage division means)

Claims (10)

Digitization processing means for generating a digital image signal corresponding to a subject image composed of a plurality of pixels;
Luminance level dividing means for dividing the subject image into a plurality of levels according to the luminance level of each pixel based on the digital image signal;
An image processing apparatus comprising: filtering means for filtering a digital image signal at each stage so that the noise removal effect becomes greater in an image portion corresponding to a stage having a lower luminance level.   2. The filter according to claim 1, wherein the filtering unit includes a smoothing filter that performs a noise removal process using neighboring pixels, and increases the number of neighboring pixels in an image portion corresponding to a lower luminance level. The image processing apparatus described.   2. The filter according to claim 1, wherein the filtering unit includes a smoothing filter that performs a noise removal process using neighboring pixels, and the number of repetitions of filtering is increased for an image portion corresponding to a lower luminance level. The image processing apparatus described.   The image processing apparatus according to claim 1, wherein the smoothing filter is a moving average filter or a median filter.   The filtering means includes an acyclic digital filter composed of a delay element and an adder, and sequentially delays and adds the corresponding digital image signal to an image portion corresponding to a lower luminance level. The image processing apparatus according to claim 1. The filtering means includes a cyclic filter including a delay element and an adder;
2. The image processing apparatus according to claim 1, wherein an image portion corresponding to a stage having a lower luminance level is filtered with a weighting coefficient of the delayed digital image signal larger than a weighting coefficient of the corresponding digital image signal. .   The image processing apparatus according to claim 1, further comprising a high-frequency filter that removes a high-frequency component of the digital image signal generated by the digitization processing unit. The digitization processing means generates a digital R signal corresponding to red (R),
The image processing apparatus according to claim 1, wherein the stage division unit performs stage division based on a digital R signal. A videoscope with an image sensor;
Digitization processing means for generating a digital image signal corresponding to a subject image formed by a plurality of pixels formed on the imaging element;
Luminance level dividing means for dividing the subject image into a plurality of levels according to the luminance level of each pixel based on the digital image signal;
An electronic endoscope apparatus comprising: a filtering unit that filters a digital image signal at each stage so that an image portion corresponding to a stage having a lower luminance level has a greater noise removal effect. A digital image signal corresponding to a subject image formed by a plurality of pixels formed on an image sensor is generated,
Based on the digital image signal, the subject image is divided into a plurality of stages according to the luminance level of each pixel,
An image processing method comprising: filtering a digital image signal at each stage so that an image portion corresponding to a stage having a low luminance level has a greater noise removal effect.

Images