KR0171119B1 - Image signal encoding apparatus using a wavelet transform - Google Patents

Abstract

The present invention relates to an image encoding apparatus using wavelet transform. The present invention relates to a first image signal processing unit (510) for variable quantizing a diagonal edge component of a high frequency, that is, a D region, and a horizontal signal edge component of a high frequency, A second video signal processor 520 for vector quantizing the H region by variable block encoding, a third video signal processor 530 for vector quantizing the vertical signal edge component of the high frequency region, that is, V region, and vector quantization; The low edge diagonal edge components, that is, the LD region, the low frequency horizontal signal edge components, that is, the LH region, and the low frequency vertical signal edge components, that is, the LV region, are vector-quantized by variable block coding, low frequency components having no edge components, That is, the LL region includes a fourth image signal processor 540 which performs discrete cosine transform (DCT) to prevent deterioration of the edge portion and to reduce a small ratio of low frequency components. By assigning it to a video compression more efficient.

Description

Image Coding Device Using Wavelet Transform

1 is a schematic diagram showing an example of an original image for explaining a conventional video encoder.

2 is a diagram illustrating band division of an original image according to FIG. 1;

3 (a) and 3 (b) are schematic diagrams of a conventional wavelet converter embodying FIG.

FIG. 4 is a diagram showing an example of an image obtained by wavelet transforming using the wavelet transform apparatus of FIGS. 3A and 3B.

5 is a block diagram of an image encoding apparatus using wavelet transform according to the present invention.

FIG. 6 is a diagram illustrating an original image divided into variable blocks by an image encoding apparatus using wavelet transform according to the present invention of FIG.

* Explanation of symbols for main parts of the drawings

510-540: Image signal processor

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high quality video system, and more particularly, to an image encoding apparatus using wavelet transform that compresses and blocks a video by using a wavelet transform.

In general, wavelet transform is used to encode an image in a digital system, which is advantageous for interpreting an image signal having an abnormal process because it can represent a signal with locality over time and frequency. The expressed image is similar to the visual characteristics of humans, so it has been widely used in the field of image processing recently.

That is, when the encoding of the video signal is schematically described, first, the wavelet converter divides the received video signal into a plurality of bands according to frequency characteristics and then transmits the received video signal to a discrete cosine converter (DCT).

A discrete cosine converter (DCT) converts a received video signal into a discrete cosine, quantizes it in a quantizer, and then scans and encodes the same to compress a lot of image data and output it to a decoder of an image system.

FIG. 1 is a schematic diagram showing an example of an original image for explaining a conventional video encoder, and FIG. 2 is a diagram showing band division of the original image of FIG. 1 according to a wavelet transform, and FIGS. ) Is a block diagram of a wavelet converter implementing FIG.

In FIG. 2, region D denotes a diagonal edge component of a high frequency signal, that is, a horizontal signal edge component of a high frequency, and a vertical signal edge component of a high frequency.

LH represents a horizontal signal edge component of a low frequency, LV represents a vertical signal edge component of a low frequency, LD represents a diagonal edge component of a low frequency, and LL represents a low frequency component without an edge component among video signals.

In FIG. 3 (a) and (b), reference numeral HF is a high pass filter for filtering high frequency of an input image signal, LF is a low pass filter for filtering low frequency component of an image signal, and the 2: 1 reduction part is a high pass filter. And a 2: 1 reduction unit for reducing the data filtered by the (HF) or the low pass filter (HF) in half.

Based on the above description, the conventional wavelet transform is as follows. That is, the region D of FIG. 2 is a diagonal edge component, and the input image signal is a high pass filter (HF, 301), a 2: 1 reducer (302), and a high pass filter (HF, 303) as shown in FIG. And via the 2: 1 reduction portion 304.

And, the H region is an edge component of the horizontal signal of the image signal, as shown in FIG. 3, the high pass filter HF 301 and the 2: 1 reduction unit 302, the low pass filter LF 305 and 2: After the reduction unit 306, discrete cosine transform (DCT) is performed.

The V region is an edge component of the vertical signal among the image signals, and as shown in FIG. 3, the low pass filter LF 311 and the 2: 1 reduction unit 312, the high pass filter HF 313 and the 2: 1 reduction are shown. Discrete cosine transform (DCT) after passing through portion 314. The L region is a low frequency component of an image signal, and as shown in FIG. 3, the low pass filter LF 311 and the 2: 1 reduction unit 312, the low pass filter LF 315 and the 2: 1 reduction unit ( 316) and then via reference 400 to FIG. 3 (b).

The low frequency region is divided into LD, LH, LV, and LL regions as shown in FIGS. 2 and 3 (b).

That is, the LD region is a diagonal edge component of the low frequency signal, as shown in FIG. 3B, a high pass filter HF 321, a 2: 1 reducer 322, a high pass filter HF 323, After passing through the 2: 1 reduction unit 324, the result is transferred to a discrete cosine converter (DCT) for discrete cosine conversion.

In addition, the LH region is an edge component of the low frequency horizontal signal, as shown in FIG. 3B, a high pass filter HF 321, a 2: 1 reducer 322, a low pass filter HF 325, and After passing through the 2: 1 reduction unit 326, the result is transferred to a discrete cosine converter (DCT) for discrete cosine conversion.

The LV region is a diagonal edge component of the low frequency vertical signal, and the low pass filter LF 331 and the 2: 1 reduction part 332, the low pass filter LF 325 and the 2: 1 reduction are shown in FIG. After passing through section 326, it is transferred to a discrete cosine converter (DCT) for discrete cosine conversion.

The LL region has no low frequency edge components, and as shown in FIG. 3, the low pass filter LF 331 and the 2: 1 reduction part 332, the low pass filter LF 335 and the 2: 1 reduction are shown. It passes to the differential pulse code modulator DPCM that performs differential pulse code modulation after passing through the unit 336.

In the conventional wavelet converter operated as described above, that is, the wavelet transform is performed as shown in FIG. 3 (a), and only the resulting low frequency component is subjected to the wavelet transform as shown in FIG. It is a low frequency component, and the method of implementing it by the differential pulse code modulation (DPCM) rather than the discrete cosine transform (DCT) has an ineffective disadvantage. Also, by not coding the D region, which is a diagonal edge component, the diagonal edge component of the video signal is lost. There is a problem in that high frequency components, i.e., degradation of edge portions and allocation of excessive bits to low frequency components.

The present invention has been made to solve the above-mentioned problems of the prior art, an object of the present invention, without the fixed block coding of the wavelet transform coefficients after wavelet transform the original image, the low-frequency component from 16 × 16 blocks, A high-frequency component, that is, a contour component, is divided into 32 × 32 blocks, and the block encoding the block where the outline is detected is variable block coded up to 4 × 4 blocks to encode only the blocks where the outline is detected. To provide.

A feature of the present invention for achieving the above object is an image using a wavelet transform for outputting a horizontal edge, a vertical signal edge, and a low frequency signal, respectively, a high-frequency diagonal edge obtained by first-order wavelet conversion of the input original image. An encoding apparatus, comprising: a first image signal processor for detecting an outline from a diagonal edge component of a high frequency band, performing variable block coding, and quantizing a vector; A second image signal processor configured to detect an outline component from the high frequency horizontal signal edge component, perform variable block coding, and vector quantize the same; A third image signal processor configured to detect an outline component from the high frequency vertical signal edge component, perform variable block coding, and vector quantize the same; A fourth block in which the low edge diagonal edge component, the low frequency horizontal signal edge component, and the low frequency vertical signal edge component are vector quantized by variable block coding, and the low frequency component having no edge component is discrete cosine transformed (DCT). It comprises a video signal processor.

In addition, the variable block encoding in the first to third image signal processing means 510, 520, and 530 uses 32 × 32 blocks, and the fourth video signal processing means 540 selects blocks whose outline components are detected from 16 × 16 blocks. It is characterized in that only 4 × 4 blocks in which an edge component is detected by variable block up to 4 × 4 blocks are encoded.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

4 is a diagram illustrating an example of an image obtained by wavelet transforming using the wavelet transform apparatus of FIGS. 3A and 3B, and FIG. 5 is a block diagram of an image encoding apparatus using wavelet transform according to the present invention. .

As shown in FIG. 4, after the wavelet transform, the original image is divided into components according to frequency bands, that is, LL, LV, LH, LD, V, H, and D regions.

In the present invention, the contour portion of the image shown in FIG. 4 is detected and variable block coding is performed without fixed block coding.

Accordingly, the outline portion may be variable block encoded using the image encoding apparatus using the wavelet transform shown in FIG. 5.

According to an embodiment of the present invention, an image encoding apparatus using wavelet transform includes a first image signal processing unit 510 for performing variable block encoding on a high frequency diagonal edge component region D and vector quantization, and a high frequency horizontal signal edge component region H. A second image signal processor 520 for vector quantizing the block by variable block coding, a third image signal processor 530 for vector quantization of the high frequency vertical signal edge component region V, and a diagonal of low frequency The edge component region LD, the low frequency horizontal signal edge component region LH, and the low frequency vertical signal edge component region LV are vector-quantized by variable block coding, and the discrete cosine is generated into the low frequency component region LL having no edge component. And a fourth image signal processor 540 for transform (DCT).

The first image signal processor 510 performs variable block encoding by detecting an edge of a high frequency diagonal edge, that is, an outline in a region D shown in FIG. 2.

In more detail, first, by dividing into 32 x 32 blocks, it is determined whether an edge component is detected. At this time, the block 32 × 32 for which no outline component is detected is not transmitted or recorded, and the block 32 × 32 for which the outline is detected is divided into 16 × 16 blocks again to detect whether the outline is included in the block. The detected block 16 × 16 is further divided into 8 × 8 blocks to perform edge detection. In the same way, after block detection is performed to block up to 4x4 blocks, only the blocks (4x4) where the outline is detected are encoded. Next, the first image signal processor 510 vector quantizes the encoded value and outputs the encoded value.

In addition, the second image signal processor 520 may include a high frequency horizontal signal edge component, that is, an outline component of the region H shown in FIG. 2, and the third image signal processor 530 may use a high frequency vertical signal edge component. Variable block coding is performed by detecting an outline component in the region V shown in FIG. In other words, in the same manner as in the first image signal processing unit 510, the block is divided into 32 × 32 blocks to 4 × 4 blocks to encode only the block 4 × 4 where the outline is detected. Thereafter, the encoded value is vector quantized and output.

Finally, the fourth image signal processor 540 may include a low edge diagonal edge component, that is, an LD region, a low frequency horizontal signal edge component, that is, an LH region, and a low frequency vertical signal edge component illustrated in FIG. 2. In the LV region, an edge component is detected by dividing it into 16 × 16 blocks, and the area where the edge is detected is vector-quantized by variable block coding up to 4 × 4 blocks, and a low frequency component without edge components, that is, as shown in FIG. The LL region is discrete cosine transformed (DCT).

6 shows an original image divided into variable blocks by an image encoding apparatus using wavelet transform according to the present invention of FIG.

As shown in FIG. 6, when the image is divided into variable blocks according to the present invention, the original image of FIG. 1 is divided as shown in FIG.

Therefore, since the present invention performs variable block coding after detecting the edge portion of the image after performing the wavelet transformation, it is possible to prevent the degradation of the edge portion and to compress the image more efficiently by allocating small bits to low frequency components. It can be effective.

Claims (2)

An image using a wavelet transform that outputs a diagonal edge (D) of high frequency, a horizontal signal edge (H), a vertical signal edge (V), and a low frequency signal (L) obtained by first-order wavelet transform of an input original image, respectively. An encoding apparatus comprising: first image signal processing means (510) for detecting a contour in the diagonal edge component region (D) of high frequency, performing variable block encoding, and vector quantizing the same; Second image signal processing means (520) for detecting a contour component in the high frequency horizontal signal edge component region (H), performing variable block coding, and quantizing the vector; Third image signal processing means (530) for detecting a contour component in the high frequency vertical signal edge component region (V), performing variable block coding, and quantizing the vector; In the low frequency signal region L, the low edge diagonal edge component LD, the low frequency horizontal signal edge component region LH, and the low frequency vertical signal edge component region LV are vector-quantized by variable block coding and edge components. The low frequency component region LL having no wavelet transform comprises a fourth image signal processing means 540 for performing discrete cosine transform (DCT). The variable block encoding of the first to third image signal processing means (510, 520, 530) is from 32x32 blocks, and the fourth image signal processing means (540) is from 16x16 blocks. An image encoding apparatus using wavelet transform, characterized in that the blocks detected are variably transformed up to 4x4 blocks to encode only 4x4 blocks in which an outline component is detected.