KR100237359B1 - Apparatus and method for shape-adaptive encoding image signal - Google Patents

Abstract

When SA-DCT coefficients of shape-adaptive DCT coefficients are transformed into shape-adaptive DCT coefficients in a device, the SA-DCT coefficients are not scanned for all coefficients in a block when zigzag scanning is performed, Wherein the SAW-DCT coefficients of the SA-CDT transformed block are skipped and scanned when the SAW-DCT coefficients are zigzag-scanned. Therefore, according to the frequency of occurrences in the VLC, It is a breakthrough scanning technology that can reduce coding data by allocating the correct bits.

Description

Apparatus and method for shape adaptive video signal coding

FIG. 1 is an explanatory diagram of a conventional zigzag scan. FIG.

Fig. 2 (a) is a general frame grid explanatory diagram.

Fig. 2 (b) is an explanatory view of a unit block of Fig. 2 (a). Fig.

Figs. 3 (a) to 3 (f) illustrate general shape adaptive DCT transform diagrams. Fig.

FIG. 4 is an explanatory view of a zigzag scan according to the present invention; FIG.

FIG. 5 is a block diagram of a conventional shape adaptive video signal encoding apparatus.

FIG. 6 is a block diagram of a shape adaptive video signal encoding apparatus according to the present invention; FIG.

Figure 7 (a) is an illustration of an object block to be encoded used in the present invention.

7 (b) is an illustration of an SA-DCT coefficient block used in the present invention.

7 (c) is an illustration of an SA-DCT coefficient shape mask used in the present invention.

The present invention relates to a shape adaptation method using a method of scanning only a portion where quantized SA-DCT coefficients exist when zigzag scanning quantized SA-DCT coefficients of an object block after shape adaptive DCT (SA-DCT) Type video signal encoding apparatus and method.

In general, H.261, which provides video phones and video conferencing using ISDN quantities, and H.263, a video phone service using public switched telephone network (PSTN), have been standardized globally.

As described above, H.261 and H.263, which provide a low-speed moving picture such as a video phone, have been degraded in image quality and are demanded to be enhanced.

On the other hand, video-related technologies have been standardized around the world, such as MPEG1 for DSM (Digital Storage Media), MPEG2 for DSM, HDTV, and STV.

Also, when the shape adaptive video signal coding is performed, the video signal can be divided into an object area to be encoded and a non-coding area such as a background.

2 shows a relationship between one frame block 31 and one unit block 33. FIG. 2 (a) and FIG. 2 (b) show a frame grid with a grid start position point x = 0 and Y = b) shows an enlarged view of one unit block, in which one unit block is divided 8 times in the x-axis and 8 times in the y-axis, and the divided particles are referred to as pixels 34 (pixels). The unit block 33 is composed of M × N blocks according to the number of pixels. Therefore, the second (b) block becomes an 8x8 block.

A shape adaptive technique which is conventionally used as a technique to encode an object is disclosed in "Adaptive DCT for Video Coding" by Thomas Sikora and Bela Maka at MPAK 94, July 1994 Shape-Adaptive DCT for Generic Coding of Video) will be described with reference to FIG. Here, the frame grid has an 8x8 block. When all the object areas to be encoded in the block are filled, the coding efficiency is the same as that of the 2D DCT (Discrete Cosine Transform). If all of the object areas are not filled, The one-dimensional DCT is performed in the vertical direction only on the shape portion of the object, and the one-dimensional DCT is further performed in the horizontal direction.

This conventional technique will be described with reference to FIGS. 3 (a) through 3 (f). In Fig. 3 (a), an object area to be coded is shown in an 8x8 block.

In order to perform shape adaptive DCT coding on the object to be encoded, first one-dimensional DCT coding is performed in the vertical direction by filling the upper boundary of the block as shown in FIG. 3 (b).

The black circle in Fig. 3 (c) shows the DC value of the one-dimensional DCT in the vertical direction. In this state, the vertical one-dimensional DCT is completed (Fig. 3 (d)).

When the vertical one-dimensional DCT is completed in this manner, the left boundary of the block is filled again and the one-dimensional DCT coding is performed in the horizontal direction as in the third (e).

Once the horizontal one-dimensional DCT is completed, the vertical direction and horizontal one-dimensional shape adaptive DCT are completed (Fig. 3 (f)).

Zigzag scan is performed as in the order of FIG. 1, with the third (f) diagram formed by the final SA-DCT coefficient.

That is, in the conventional standards, MPEG-1, MPEG-2, JPEG, H.261 and H.263 using block-oriented coding are used for variable length coding after DCT, Lt; / RTI >

The shape adaptive DCT transform and the zigzag scan can be exemplified as the fifth embodiment. As shown in FIG. 7 (a), in which object information to be encoded within an M × N block as a coding unit coexists with a non- The image information a is supplied at the input terminal a1 of the shape adaptive DCT unit 10 and SA-DCT transformed to obtain a transform coefficient as shown in Fig. 7 (b). The output of the shape adaptive DCT unit 10 is supplied to the input terminal b1 of the quantizer 11 through the a2 and quantized by the quantizer 11 is scanned by the scan performing unit 12 in the same order as in the first aspect.

In this case, the scan performing unit 12 sets the portion having no conversion coefficient at the time of scanning to " 0 " and sequentially performs the setting scanning procedure (e.g., 1 to 64) regardless of the presence or absence of the conversion coefficient.

The scanned information is supplied to the next transmission stage which is variable in the variable length coding 13 and stored in the buffer 14. [

At this time, the fullness of the buffer 14 is recognized by the buffer control unit 15, and the buffer control unit 15 controls the control unit b2 of the quantizer 11 to maintain the required transmission and image quality .

The variable length coding unit 13 receives only the SA-DCT transform coefficients output from the scan performing unit 12 and outputs the SAFT coefficients to the VLC unit 13 of the variable length coding unit 13 according to an event EMENT occurring in the input values. And outputs table (VLC TABLE) values.

At this time, the type of the event is the last (LAST) indicating whether or not the last conversion coefficient in the zigzag scan and the last (LAST) indicating whether or not the conversion coefficient value is "0" (RUN) indicating the number of coefficients and a level (LEVEL) indicating a coefficient value that is not "0" appearing after "0".

A small bit is assigned to an event having a high probability of occurrence of the event, and a VLC table value set to a variable-length bit value for each event is allocated to a low-probability event. Output.

For example, some reference to the VLC table used in H.263 is shown in Table 1 below.

[Table 1]

The input value of the variable length encoder is ... × 01 ... (X is a count value that is not "0"). An event with run (RUN) of "1", last (LAST) of "0" and level of LEVEL of "1" And a value of 4 bits is output at 1100.

However, if the scan execution unit 12 scans a portion that does not exist as " 0 ", " × 001 ... or … × 0001 ... And an unnecessary " 0 " are added to the event. At this time, the output of the variable length encoder is 11100, 011010, and 5 bits and 6 bits are allocated.

By adding unnecessary " 0 " value coefficients in this manner, an event in which the run value is increased becomes an unnecessary bit is allocated.

The present invention provides an apparatus and method for scanning only a portion where an SA-DCT coefficient exists when scanning SA-DCT coefficients according to shape adaptive DCT transform for shape adaptive coding .

That is, according to the present invention, in a zigzag scanning apparatus for shape adaptive video signal coding, a transform coefficient output from the shape adaptive DCT unit is provided to divide a coefficient into a non-existent portion and a non-existent portion, A zigzag scanning control means for controlling the scanning in accordance with a zigzag scanning order and skipping the scanning order when it does not exist; And a zigzag scan performing unit for skipping the quantization coefficients obtained by quantizing the transformation coefficients by excluding the skip signals of the zigzag scan control unit from the scan order when the zigzag scan control unit has a skip signal when the zigzag scan is performed, thereby providing a zigzag scan apparatus for shape adaptive video signal encoding will be.

Wherein the zigzag scan control means comprises: an SA-DCT coefficient shape mask forming unit for forming an SA-DCT coefficient shape mask divided into a part in which a transform coefficient exists and a part in which no transform coefficient exists; And a zigzag scan control unit for generating a skip signal so that the zigzag scan performing unit skips the scan order in the scan order when the zigzag scan is performed.

That is, according to the present invention, in the zigzag scan of the SA-DCT coefficients of the transformed block after the shape adaptive DCT, the shape adaptive zigzag scan in which only the portion in which the SA- Method.

The present invention also relates to a method of separating a portion having a SA-DCT coefficient from a portion having no SA-DCT coefficient; And performing a partial scan of only a portion of the SA-DCT coefficient quantized by the segment value according to the classification rule; Wherein the partial scan process includes scanning a portion having quantized SA-DCT coefficients according to a conventional zigzag scan sequence, and skipping a portion having no quantized SA-DCT coefficient. And to provide an adaptive video signal encoding method.

According to another aspect of the present invention, there is provided a shape adaptive image signal coding method for quantizing SA-DCT coefficients and zigzag scanning the quantized SA-DCT coefficients, the method comprising the steps of: Generating a binary mask according to whether a coefficient value is present in the determination process; Quantizing SA-DCT coefficients; Zigzag scanning this; Comparing the scanned coefficient with the binary mask and controlling switching of a portion having no coefficient value to skip a scan output; comparing the scanned coefficient with the binary mask to provide a scan output; A switching step of controlling a switching operation; And performing a variable length coding only on the coefficients scanned according to the switching process.

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

FIG. 6 is a diagram illustrating a zigzag scan apparatus for a shape-adaptive image signal coding apparatus according to the present invention. FIG. A zigzag scan control unit (100) for controlling the existing part to skip according to a conventional zigzag scan order, and skip the existing part if not present, and skip the scan order; And a zigzag scan performing unit 200 skipping the quantization coefficients obtained by quantizing the transform coefficients in a scan order by interlocking with the skip signal of the zigzag scan control means 100 during a zigzag scan.

The zigzag scan control means 100 includes an SA-DCT coefficient mask forming unit 110 for forming an SA-DCT coefficient shape mask divided into a portion in which a transform coefficient exists and a non-exist portion; It is preferable that the zigzag scan control unit 120 generates a skip signal so that the zigzag scan performing unit 200 skips the scan order in the scan order and skips the scan order in the zigzag scan performing unit 200 . At this time, the shape mask information is transmitted to a decoder by shape coding data of a video object plane (VOP) coding process.

The zigzag scan performing unit 200 includes a scan performing unit 12 for zigzag scanning SA-DCT-transformed coefficients and a scan executing unit 12 for performing scan control of the scanned information from the scan performing unit 12 upon input of a skip signal of the zigzag scan control unit 120. [ And a switching unit 210 for skipping provision.

The operation of the present invention constructed as described above will be described.

An M × N block in which object information to be encoded within an M × N block, which is a coding unit, exists together with image information of a portion that is not an object region is shown in a state (an 8 × 8 block) ).

When the object information such as the state shown in FIG. 7 (a) is input to the input terminal of the shape adaptive DCT unit 10 of FIG. 6, the shape adaptive DCT transform is performed and the shape adaptation Type DCT coefficient block is output to the output terminal a2 of the shape adaptive DCT unit 10 as shown in FIG.

The SA-DCT coefficient block thus outputted is input through the input terminal b1 of the quantizer 11 and the quantized coefficient block (quantization process) in the quantizer 11 is input to the scan execution unit 200 of the zigzag scan execution unit 200. [ (Zigzag scanning process) is performed in step S12.

On the other hand, the DCT coefficient block as shown in FIG. 7 (b) corresponds to the portion (black portion) in which coefficients exist in the SA-DCT coefficient mask forming unit 110 of the zigzag scan control means 100 as shown in FIG. And forms an SA-DCT coefficient shape mask indicating a portion (white portion) in which no coefficient exists. This can be formed by dividing the portion where the coefficient exists, for example, by "1" (or "0") and the portion where the coefficient does not exist by "0" Mask formation process).

The SA-DCT coefficient shape mask output by the SA-DCT gain mask forming unit 110 is applied to the zigzag scan control unit 120 through the output stage a3, and the zigzag scan control unit 120 performs the zigzag scan And controls the switching unit 210 of the execution unit 200.

That is, when the output through the output terminal a3 of the SA-DCT coefficient mask forming unit 110 has the same shape as that of the fourth step, the zigzag scan control unit 120 outputs the scan signal at every scan time corresponding to the coefficient block (ai) The switching unit 210 is turned off through the latch a4 to skip the output of the scan coefficient of the scan performing unit 12 (switching process).

That is, in the case of the present invention, the scan scan control unit 120 of the zigzag scan control unit 100 scans a portion of the portion without the coercive force, At the output time point, the switching unit 210 is turned off to cut off the output, so that only the portion having the coefficient is applied to the variable length coding unit 13, whereby the variable length coding is performed with an accurate bit allocation according to the occurrence frequency The amount of data at the time of transmission can be reduced by reducing the amount of data of the unit 13.

That is, according to the present invention, only the part where the transform coefficient is present is skipped and the part where the transform coefficient does not exist is scanned. Thus, the variable length coding unit 13 divides the input coefficient into the last, run, In the present invention, when a portion without a coefficient is skipped in a zigzag scan according to the present invention, a value coefficient of " 0 " is reduced and unnecessary run values are not calculated. As a result, unnecessary bit transmission is suppressed. It is possible to use the extra bits for the low transmission rate (coding process).

As described above, according to the present invention, when a SA-DCT coefficient of a shape adaptive DCT transformed block is zigzag scanned in a shape adaptive DCT unit, a coefficientless portion is skipped and scanned, Thereby reducing the amount of coding data.

Claims (10)

In a shape-adaptive image signal coding apparatus, a transform coefficient output from the shape adaptive DCT unit 10 is divided into a portion in which a transform coefficient exists and a portion in which a transform coefficient exists, A zigzag scanning control unit (100) for controlling the scanning unit to scan in accordance with a zigzag scanning order of the scanning unit; And a zigzag scan performing unit 200 skipping the quantization coefficients obtained by quantizing the transform coefficients in a scan order by interlocking with the skip signal of the zigzag scan control means 100 and skipping the scan quantities during a zigzag scan. Shape adaptive video signal coding apparatus. The apparatus of claim 1, wherein the zigzag scan control means (100) comprises an SA-DCT coefficient mask forming unit (110) for forming an SA-DCT coefficient shape mask divided into a part in which a transform coefficient exists and a part in which no transform coefficient exists, ; And a zigzag scan control unit 120 for generating a skip signal so as to skip the scan flag in the scan order when the zigzag scan performing unit 200 scans the zigzag scan unit, Shaped adaptive video signal encoding apparatus. The apparatus of claim 1, wherein the zigzag scan performing unit includes a scan performing unit for performing a zigzag scan of SA-DCT-converted coefficients, a scan executing unit for performing a skip signal input of the zigzag scan controlling unit, And a switching unit (210) for skipping the provision of the scanned information in the scanning unit (210). In a shape-adaptive image signal coding apparatus, a transform coefficient output from the shape adaptive DCT unit 10 is divided into a portion in which a transform coefficient exists and a portion in which a transform coefficient exists, A zigzag scanning control unit (100) for controlling the scanning unit to scan in accordance with a zigzag scanning order of the scanning unit; A quantizer 11 for quantizing the transform coefficients of the DCT unit 10; A zigzag scan performing unit 200 skipping a quantization coefficient quantized by the quantizer 11 in a scan order in conjunction with a skip signal of the zigzag scan control means 100 during a zigzag scan; A variable length coding unit 13 for performing variable length coding according to the occurrence frequency of the transform coefficient by the scan of the zigzag scan performing unit 200; A buffer 14 for buffering the encoded data of the variable length coding unit 13 and providing the buffering fullness to the buffer control unit 15; And a buffer control unit (15) for controlling the quantizer (11) to maintain the image quality according to the buffering fullness. A method of shape adaptive image signal coding for quantizing an SA-DCT coefficient and zigzag scanning the SA-DCT coefficient, the method comprising scanning only a portion where an SA-DCT coefficient exists. 6. The shape adaptive video signal coding method of claim 5, wherein before scanning only the portion having the quantized SA-DCT coefficients, a step of distinguishing the portion having the SA-DCT coefficient from the portion having no SA-DCT coefficient is performed. 7. The shape adaptive video signal coding method of claim 6, wherein the dividing process uses a binary mask. A step of separating a portion having the SA-DCT coefficient from the portion having no SA-DCT coefficient; Performing a partial scan of only a portion of the SA-DCT coefficient quantized by the segment value according to the segmentation process; The partial scanning process includes scanning the portion having the quantized SA-DCT coefficients according to a conventional zigzag scan sequence, and Wherein the step of skipping the scan includes the step of skipping the scan without the quantized SA-DCT coefficients. 9. The shape adaptive video signal coding method of claim 8, wherein the dividing process uses a binary mask. A shape adaptive image signal coding method for quantizing an SA-DCT coefficient and zigzag scanning the quantized SA-DCT coefficient, the method comprising: determining whether there is a transformed SA-DCT coefficient; Generating a binary mask according to whether a coefficient value is present in the determination process; Quantizing SA-DCT coefficients; Zigzag scanning this; Comparing the scanned coefficient with the binary mask and controlling switching of a portion having no coefficient value to skip a scan output; comparing the scanned coefficient with the binary mask to provide a scan output; A switching step of controlling a switching operation; Wherein the step of performing variable length coding only on coefficients scanned according to the switching process is performed.