JP5488048B2 - Image processing apparatus and image processing program - Google Patents

Description

  The present invention relates to an image processing apparatus and an image processing program, and more particularly to an apparatus for removing noise from an image.

  In image signal processing, noise removal using a Total Variation filter can efficiently remove noise during low compression. However, when the compression rate is increased, more noise is included in the compressed image, so it is difficult to remove the noise without degrading the image quality.

JP 2009-20605 A

F. Alter, S. Durand and J. Froment, "Adapted total variation for artifact free decompression of JPEG images", J. Mathematical Imaging and Vision, Vol. 23, No. 2, pp. 199-211, Sep. 2005.

  In the image noise removal according to Patent Document 1, mosquito noise can be removed, but removal of block distortion cannot be considered. Further, noise removal using the Total Variation filter of Non-Patent Document 1 has a problem in that block distortion removal becomes insufficient because the total variation with respect to a DCT (Discrete Cosine Transform) coefficient is minimized. In addition, in order to minimize the total variation with respect to the DCT coefficient, there is a problem that a texture component is lost, resulting in a blurred image.

  In view of the above points, an object of the present invention is to appropriately remove block distortion and mosquito noise.

In order to achieve the above object, the invention described in claim 1
Component separation means for separating an input image into a skeleton component and a texture component;
Block distortion removing means for removing block distortion on the texture component separated by the component separating means;
Edge detection means for detecting an edge from the skeleton component separated by the component separation means;
Mosquito noise removing means for removing mosquito noise for pixels around the edge detected by the edge detecting means;
An image processing apparatus comprising: a texture component from which the block distortion and the mosquito noise are removed and a skeleton component are synthesized to obtain a restored image.

The invention according to claim 2
Computer
Component separation means for separating an input image into a skeleton component and a texture component;
Block distortion removing means for removing block distortion on the texture component separated by the component separating means;
Edge detection means for detecting an edge from the skeleton component separated by the component separation means;
Mosquito noise removing means for removing mosquito noise for pixels around the edge detected by the edge detecting means;
An image processing program that combines a texture component from which the block distortion and the mosquito noise have been removed and the skeleton component to function as a means for obtaining a restored image is characterized.

It is a figure which shows the structure of the image processing apparatus which concerns on one Embodiment of this invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments shown in the drawings will be described below. FIG. 1 shows a configuration of an image processing apparatus according to an embodiment of the present invention. This image processing apparatus includes a Total Variation filter 10 that separates a DCT compressed image as an input image into a skeleton component and a texture component of the image, and a DEF (Deblocking Edge Filter) that performs block distortion removal on the texture component thus separated. 12, a Sobel filter 14 for detecting an edge from a skeleton component, a Gaussian filter 16 for removing mosquito noise from pixels around the edge, a texture component and a skeleton component from which block distortion and mosquito noise have been removed, And a component synthesizing unit 18 that obtains a restored image.
This image processing apparatus operates as follows. The DCT compressed image as an input image is separated into a skeleton component and a texture component by the Total Variation filter 10. The skeleton component includes a smooth luminance value change signal and an edge component, and the texture component includes a finely oscillating signal. DCT compression noise is included in the texture component.

  Block distortion is removed by passing the DEF 12 through the separated texture components. This DEF12 is disclosed in ITU-T. “Video coding for low bit rate communication, 1998, which is a non-patent document.

  Since the mosquito noise appears around the edge signal, the edge signal is detected from the skeleton component by the Sobel filter 14, and the pixels around the detected edge signal are filtered by the Gaussian filter 16 to remove the mosquito noise.

  The texture component and the skeleton component from which block distortion and mosquito noise have been removed are synthesized by the component synthesis unit 18 to obtain a restored image.

 According to this embodiment, the block distortion can be effectively removed by performing DEF on the texture component after the component separation by the Total Variation filter, and the mosquito noise generated in the vicinity of the edge is removed using the Gaussian filter. By selecting the pixels to which the Gaussian filter is applied by detecting edges, the removal of texture components can be suppressed as much as possible when removing noise, making it less likely to cause blur in the restored image. Noise mixed in DCT compressed images such as MPEG, MPEG-II, and H.264 can be effectively removed.

  The image processing apparatus shown in FIG. 1 can be realized by software using a computer. In that case, each component shown in FIG. 1 is grasped as a means for realizing each function, and an image processing program is constituted by them. That is, component separation means for separating the input image into skeleton components and texture components, block distortion removal means for removing block distortion from the texture components separated by the component separation means, and component separation by the component separation means Edge detection means for detecting edges from the detected skeleton components, mosquito noise removal means for removing mosquito noise for pixels around the edges detected by the edge detection means, and block distortion and mosquito noise are removed. As a means for synthesizing the texture component and the skeleton component thus obtained to obtain a restored image, the image processing program is configured to cause a computer to function.

  Further, the present invention is not limited to the one described in the above embodiment, and the component separating means, the block distortion removing means, the edge detecting means, and the mosquito noise removing means can realize their respective functions. For example, filters other than the Total Variation filter, DEF, Sobel filter, and Gaussian filter may be used.

10 Total Variation Filter 12 DEF
14 Sobel Filter 16 Gaussian Filter 18 Component Synthesizer

Claims (2)

Component separation means for separating an input image into a skeleton component and a texture component;
Block distortion removing means for removing block distortion on the texture component separated by the component separating means;
Edge detection means for detecting an edge from the skeleton component separated by the component separation means;
Mosquito noise removing means for removing mosquito noise for pixels around the edge detected by the edge detecting means;
An image processing apparatus comprising: a texture component from which the block distortion and the mosquito noise have been removed and a skeleton component to obtain a restored image. Computer
Component separation means for separating an input image into a skeleton component and a texture component;
Block distortion removing means for removing block distortion on the texture component separated by the component separating means;
Edge detection means for detecting an edge from the skeleton component separated by the component separation means;
Mosquito noise removing means for removing mosquito noise for pixels around the edge detected by the edge detecting means;
An image processing program that combines a texture component from which the block distortion and the mosquito noise have been removed and the skeleton component to function as means for obtaining a restored image.