JP3326819B2 - Image processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly to an image processing apparatus suitable for performing noise removal processing on an image displayed on a television monitor in an X-ray television system.

[0002]

2. Description of the Related Art Generally, in an X-ray television system, as shown in FIG.
An X-ray passing through the subject 42 is converted into an image
The obtained visible image is converted into a video signal by a television camera 44, converted into a digital signal by an A / D converter 45, passed through a recursive filter circuit 46, and subjected to image processing. D / A converter 47
The signal is returned to an analog signal and sent to the television monitor device 48 to display an X-ray fluoroscopic image.

[0003] The recursive filter circuit 46 smoothes video data in the time direction, thereby removing image noise.

[0004]

However, in the conventional image processing using such a recursive filter circuit, a noise removing effect is uniformly generated even in a bright or dark place of an image. There is a problem that can not be. That is, the image pickup tube of a television camera has a characteristic that there is a lot of noise in a dark part and the noise is conspicuous. Noise in dark places cannot be removed effectively.

[0005] Another problem is that the image is blurred when there is motion. That is, since the image processing by the recursive filter is the smoothing of the image data in the time direction, if the recursive filter coefficient is increased to increase the noise removing effect, the image of the fast-moving part is blurred.

In view of the above, the present invention enhances the noise removal effect in dark portions of an image and reduces the noise removal effect in bright portions to suppress blurring of the image. It is an object of the present invention to provide an image processing apparatus capable of obtaining an image having a uniform S / N ratio without blurring.

[0007]

In order to achieve the above object, an image processing apparatus according to the present invention includes a frame memory, which multiplies an image signal read from the frame memory by (1-1 / K) times and simultaneously outputs the image signal. A recursive filter means for multiplying the image signal of the frame by 1 / K, adding these, and storing the result in the frame memory; and a recursive filter coefficient K for each pixel from the image data of the current frame. Means for controlling according to the average luminance of the local area; means for performing spatial filtering on the image data of the current frame; and obtaining a motion component for each pixel from the relationship between the image data of the current frame and the image data of the previous frame. Means, and according to the motion component for each pixel, when the motion is large, the image data having passed through the recursive filter It has a means for gradually switched to the image data having undergone the filtering processing means is the distinctive feature.

[0008]

The average luminance in a local area near a certain pixel is obtained, and the recursive filter coefficient K for that pixel is controlled accordingly. Thereby, in a dark part of the image, the recursive filter coefficient K can be increased to increase the noise removing effect. On the other hand, since the noise is originally small in the bright part of the image, the recursive filter coefficient K is reduced to reduce the noise removing effect and prevent the image from being blurred due to motion. At this time, since the recursive filter coefficient K is controlled in accordance with the average luminance in the local area near the pixel of interest, the recursive filter coefficient K can be determined without being affected by luminance variation or noise. Since the recursive filter coefficient K can be determined without being affected by the variation in luminance for each pixel, the recursive filter coefficient K is stable, that is, the noise removal effect is stable, and the image obtained through the recursive filter means is stable. The obtained easy-to-view image can be obtained. Furthermore, when the movement becomes large, the image is switched from recursive filter processing, which is processing in the time direction, in which the image is blurred due to movement, to processing in the spatial direction. To remove noise by spatial processing.

[0009]

An embodiment of the present invention will be described below in detail with reference to the drawings. FIG. 1 shows an image processing apparatus according to one embodiment of the present invention. The image processing apparatus according to this embodiment is used in place of the recursive filter circuit 46 in FIG. It comprises a recursive filter circuit to which a dynamic control function is added every time, a noise removal processing circuit in the spatial direction, and a switching circuit for switching these circuits.

In FIG. 1, digital image data In representing an n-th frame of a video signal of an X-ray fluoroscopic image is input to a local average value calculating circuit 13, and a local average value Mn (x, x, y) at each pixel (x, y) is obtained. y) is calculated. As shown in FIG. 2, the local average value Mn (x, y) is obtained by calculating W × W pixels around the pixel 32 at the position (x, y) on the screen 31 as the center. This is the average value (average luminance) of the pixel data of the local area 33 extracted from the template, and is represented by the following Equation 1.

(Equation 1)

The local average value Mn (x, y) is sent to a recursive filter coefficient calculation circuit 15 to obtain an appropriate recursive filter coefficient Kn (x, y) for each pixel. That is, the recursive filter coefficient calculation circuit 1
5 has a conversion table for obtaining a recursive filter coefficient Kn (x, y) from the local average value Mn (x, y). This conversion table is of a simple inverse proportional type as shown in FIG. It is desirable to select an inverse proportional type with a dead band as shown by B in FIG. 3, an inverse exponential type as shown in C in FIG.

The recursive filter coefficient K thus obtained
From n (x, y) to 1 / {Kn (x, y)} and 1-1 /
{Kn (x, y)} is obtained, and the former is sent to the multiplier 17 and the digital image data In (x,
y), and the latter is sent to the multiplier 18 and multiplied by the processed digital image data I′n−1 (x, y) of the previous frame stored in the frame memory 12.

The outputs of the multipliers 17 and 18 are added to an adder 1
9 and processed digital image data I'n
(X, y) is obtained and is written to the frame memory 12. The recursive filter function is achieved by repeating the above operation for each pixel and each frame.

Thus, the recursive filter coefficient Kn
(X, y) is small when the local average value Mn (x, y) is large and is large when the local average value Mn (x, y) is small. In a dark part, the noise removal effect is large. In general, there is little noise in the bright part and there is much noise in the dark part, so by changing the degree of the noise removal effect according to the brightness in this way, it is possible to respond to the original noise magnitude. As a result, a uniform S / N ratio can be obtained over the entire surface of the image.

On the other hand, the frame memory 11 stores the image data of each frame and outputs the image data of the previous frame. That is, image data In-1 one frame before the current image data In of the n-th frame is obtained from the frame memory 11 with a delay of one frame. This image data In-
1 is input to a local average value calculating circuit 14 having the same configuration as that of the above local average value calculating circuit 13, and the local average value Mn−
1 (x, y) is obtained.

The local average values Mn (x, y) and Mn-1 (x, y) from the local average value calculation circuits 13 and 14 are as follows:
It is sent to the motion coefficient calculation circuit 16. The motion coefficient calculation circuit 16 calculates a motion coefficient ΔMn (x, y) for each pixel according to ΔMn (x, y) = Mn (x, y) −Mn−1 (x, y). The motion coefficient ΔMn (x, y) is calculated based on the pixel (x,
It shows the change over time of the local average value Mn (x, y) for y), and has a strong correlation with the movement of the subject image near the pixel.

The motion coefficient ΔMn (x,
The switching circuit 20 is switched according to y). That is, when the motion coefficient ΔMn (x, y) exceeds a predetermined threshold, the switching circuit 20 is switched from the frame memory 12 to the spatial filter processing circuit 21. The spatial filter processing circuit 21 performs a spatial filter process such as a convolution process using, for example, a template of N × N pixels on the image data In of the current n-th frame, and performs smoothing in the spatial direction in the image of this frame. To obtain the obtained image data.

The image data from the spatial filter processing circuit 21 is sent to the multiplier 18 instead of the image data from the frame memory 12 by switching of the switching circuit 20. Therefore, when the movement becomes larger than a certain level, the processing is switched from the processing in the time direction to the processing in the spatial direction, and noise removal without blurring due to the movement can be performed.

When the movement becomes larger than a certain level, the switching circuit 20 does not completely switch the processing in the time direction to the processing in the space direction. Can be gradually changed. That is, a weighting addition circuit (not shown) for weighting and adding the image data from the frame memory 12 and the image data from the spatial filter processing circuit 21 is provided in place of the switching circuit 20,
The weighting and adding circuit is controlled according to the motion coefficient ΔMn (x, y), and the weight of the image data from the spatial filter processing circuit 21 is increased as the motion coefficient ΔMn (x, y) increases.

Further, although the embodiment applied to the X-ray television system has been described above, the image processing apparatus according to the present invention can be applied to other cases where it is necessary to remove image noise.

[0021]

As described above, according to the image processing apparatus of the present invention, the recursive filter coefficient K for each pixel is dynamically controlled in accordance with the average luminance in the local area near the image of interest to generate a large amount of noise. The noise reduction effect is enhanced in dark areas, and the noise reduction effect is reduced in bright and originally low noise areas to prevent blurring due to motion. , The recursive filter coefficient K can be determined without being affected by variations in luminance for each pixel, and the like.
By stabilizing the noise removal effect, it is possible to obtain a stable and easy-to-view image, and when the movement is even greater, the processing data in the spatial direction is used. It is possible to perform noiseless removal and to obtain an image with a uniform S / N ratio without blurring due to motion on the entire screen.

[Brief description of the drawings]

FIG. 1 is a block diagram of one embodiment of the present invention.

FIG. 2 is a view showing pixels and local regions for explaining the operation of the embodiment.

FIG. 3 is a graph showing conversion characteristics between a local average value M and a recursive filter coefficient K in the embodiment.

FIG. 4 is a block diagram of a conventional example.

[Explanation of symbols]

 11, 12 Frame memory 13, 14 Local average value calculation circuit 15 Recursive filter coefficient calculation circuit 16 Motion coefficient calculation circuit 17, 18 Multiplier 19 Adder 20 Switching circuit 21 Spatial filter processing circuit 41 X-ray tube 42 Subject 43 Image intensity Fire 44 TV camera 45 A / D converter 46 Recursive filter circuit 47 D / A converter 48 TV monitor device

──────────────────────────────────────────────────続 き Continued on front page (58) Field surveyed (Int.Cl. ⁷ , DB name) A61B 6/00

Claims (1)

(57) [Claims] An image signal read from the frame memory is multiplied by (1-1 / K) and an image signal of a current frame is multiplied by 1 / K. A recursive filter means for storing the recursive filter coefficient K for each pixel from the image data of the current frame in accordance with the average luminance of a local region near the pixel; Means for filtering, means for obtaining a motion component for each pixel from the relationship between the image data of the current frame and the image data of the previous frame, and the above-described recursive Means for switching from image data passed through the filter means to image data passed through the spatial filter processing means. Image processing apparatus.