JPH06149998A - Image processing device - Google Patents

Abstract

PURPOSE:To perform filtering processing with optimum frequency characteristic by obtaining an evaluation value from a picture element value, and selecting one of plural templates corresponding to the value when a two-dimensional convolution arithmetic operation is performed. CONSTITUTION:Image data to be processed is stored in an original image memory 11, and it is read out and is sent to an evaluation calculator 12, then, the evaluation value is calculated. A template storage circuit 13 retains three templates, and also, retains the threshold value for the evaluation value, and selects one of the three templates corresponding to relation between the evaluation value sent from the evaluation calculator 12 and the threshold value. One template selected by the template storage circuit 13 is sent to a two-dimensional convolution arithmetic circuit 14. The twodimensional convolution arithmetic circuit 14 performs the two-dimensional convolution arithmetic operation by making the template act on a picture element sent from the original image memory 11.

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 filtering medical X-ray digital images.

[0002]

2. Description of the Related Art A medical X-ray digital image usually has a wide dynamic range of pixel values, but has a feature that the pixel value of an observed region is little changed. For example, in soft tissue, the pixel value changes only slightly.

Therefore, conventionally, the dynamic range of the entire image signal is compressed, while a filtering process is performed to make it easy to see minute changes. This filtering process is performed by applying a template (for example, a template of 3 × 3, 5 × 5, etc.) obtained from the given frequency characteristic to the original image.
By such a filtering process, the high-frequency region of the image can be selectively passed to improve the contrast of the portion to be observed and to clarify the shape of the fine portion.

[0004]

However, the image may be difficult to see if only the filtering process for selectively passing the high frequency band is performed as in the conventional case. That is, in a portion having a high degree of X-ray absorption, noise is emphasized when a high frequency region is selectively passed, resulting in poor image quality and loss of image smoothness.
The fact that the X-ray absorption is high means that the detected X-ray dose is small, the quantum noise is large, and the S / N ratio is low, so that this noise ends up being emphasized.

In view of the above, the present invention makes it possible to change the frequency characteristic of the filter for each part according to the characteristic of the image, thereby performing the filtering process while maintaining the smoothness of the image. An object is to provide a processing device.

[0006]

In order to achieve the above object, the image processing apparatus according to the present invention holds a plurality of templates having different frequency characteristics and having the same DC components. An evaluation value is obtained from the pixel value of the image to be processed, and the value of
It is characterized in that the two-dimensional convolution operation is performed on the pixel values of the image to be processed using the selected template using the selected template.

[0007]

[Operation] A plurality of templates for one image,
A two-dimensional convolution operation is performed using each part. At that time, an evaluation value is obtained from the pixel value, and one of the plurality of templates is selected according to the evaluation value. As a result, it is possible to perform the filtering process of the optimum frequency characteristic for each part, prevent the noise from being emphasized in the part with a lot of quantum noise, prevent the image quality from becoming rough, and maintain the smoothness of the image. You can

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will now be described in detail with reference to the drawings. FIG. 1 shows an image processing apparatus according to an embodiment of the present invention. In FIG. 1, the image data to be processed is stored in the original image memory 11, which is read out and evaluated by the evaluation value calculator 12.
And the evaluation value is calculated. This evaluation value is, for example, the value of the pixel of interest itself, or the vicinity of the pixel of interest (for example, 3 × 3 or 5 × 5 around the pixel of interest).
Pixel) is calculated based on the variance value.

The template storage circuit 13 holds, for example, three templates A, B and C as shown in FIG. 4, and also holds thresholds u and v (see FIG. 3) for evaluation values. However, depending on the relationship between the evaluation value sent from the evaluation value calculator 12 and these threshold values u and v, 3
Select one of the two templates A, B, C.

These three templates A, B and C are made up of a 3 × 3 matrix as shown in FIG. 4, and their respective values have the frequency characteristics shown by the curves A, B and C in FIG. 5, respectively. It is set. As shown in FIG. 5, these frequency characteristics reduce the low range to compress the dynamic range of the image, but pass the middle range to add clear contrast to the fine structure, and the high range causes noise. It is common in that it is attenuated to suppress it, but changes have been made in the degree of passage in the midrange. Curve A has the strongest high-pass characteristics, curve C the weakest, and curve B
Is in the middle. Each value of the template is obtained from this frequency characteristic using a two-dimensional FFT (Fourier transform). The DC component is defined by point p in FIG. 5 and has a common value, but the DC characteristics of each template are different due to the truncation error of the two-dimensional FFT. Here, the DC characteristic of the template is the sum of the values at each point of the template. In this embodiment, since three templates act on one image, the DC characteristics of each template must be exactly matched.

One template selected by the template storage circuit 13 is sent to the two-dimensional convolution operation circuit 14. In the two-dimensional convolution operation circuit 14, the above template is applied to the pixels sent from the original image memory 11 to perform the two-dimensional convolution operation. That is, the pixel of interest, for example, the pixels im and n in FIG. 2 are aligned with the center point of the template, a two-dimensional convolution operation is performed between the value of each pixel and each value of the template, and the value obtained as the operation result. Is the value of the pixel im, n after filtering.

The filtered values thus obtained are stored in the filtered image memory 15 as filtered image data.

Actually, the necessary pixel value is output from the original image memory 11 to the evaluation value calculator 12 to calculate the evaluation value, and the template storage circuit 13 outputs 1 in accordance with this.
When one template is selected and sent to the two-dimensional convolution operation circuit 14, the signal of the pixel corresponding to the template is sent to the two-dimensional convolution operation circuit 14 from the original image memory 11 in synchronization with this. Calculation is performed.

Thus, for example, in one original image, a portion having a high X-ray absorption is subjected to a two-dimensional convolution operation using a template having a frequency characteristic as shown by the curve C in FIG. 5, and the X-ray absorption is low. In the portion, it is possible to perform a two-dimensional convolution operation using a template having a frequency characteristic as shown by the curve A in FIG. Therefore, X-ray absorption is high,
In the portion where the data contains a lot of noise, the pass rate in the high frequency range (which is the middle range in FIG. 5) is made lower than the others,
The degree of emphasis can be weakened so that noise is not emphasized. As a result, it is possible to prevent a noise-emphasized image having a rough image quality from being obtained. Since the template having the frequency characteristic as shown by the curve A in FIG. 5 is applied to the portion where the X-ray absorption is low, a sufficient edge enhancement effect can be obtained by the filtering process, and since noise is small in this portion, the image is rough. There is no such thing. Therefore, the image after the filtering process stored in the filtered image memory 15 has the contrast enhanced while maintaining the smoothness over the entire surface.

Although the original image memory 11 and the filtered image memory 15 are configured as different image memories here, they may be the same memory. Further, the template can be appropriately selected such as 5 × 5 other than 3 × 3. Further, in this embodiment, all the configurations are represented by hardware, but software can be configured as long as they have equivalent functions.

[0016]

As described above, according to the image processing apparatus of the present invention, the frequency characteristic of the filter is changed for each part according to the characteristic of each part of the image while maintaining the smoothness of the image. A filtering process can be performed. As a result, it is possible to perform the filtering process of the optimum frequency characteristic for each part, not to emphasize the noise in the part with a lot of quantum noise, prevent the image quality from becoming rough, and maintain the smoothness of the image. The contrast can be improved by the filtering process.

[Brief description of drawings]

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

FIG. 2 is a diagram showing pixels in an original image.

FIG. 3 is a diagram showing a relationship between an evaluation value and a template.

FIG. 4 is a diagram showing each template.

FIG. 5 is a graph showing frequency characteristics of a filter.

[Explanation of symbols]

 11 original image memory 12 evaluation value calculator 13 template storage circuit 14 two-dimensional convolution operation circuit 15 filtering image memory

Claims (1)

[Claims] 1. A means for obtaining an evaluation value from a pixel value of an image to be processed, and a plurality of templates having different frequency characteristics and having matched DC components are held, and these templates are stored according to the evaluation value. 2. An image processing apparatus comprising: a means for selecting one of the templates and a means for performing a two-dimensional convolution operation on pixel values of an image to be processed using the selected template.