JP2005210566A - Image data coding method, and image processing device for coding image data - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image data coding method for reducing a processing time of coding and preventing a deterioration of image data at decoding of the coded image data, and provide an image processing device for encoding the image data by using the coding method. <P>SOLUTION: In the image data coding method and the image processing device for coding the image data by using the coding method, two or more image frames are separated into a prescribed number of pixel blocks and the two or more AC coefficients obtained by quantizing after orthogonal conversion of image data in each pixel block are encoded sequentially from a low frequency side to a high frequency side. When encoding one image frame out of two or more image frames is carried out, a coding stopping condition for stopping the coding of each pixel block based on the consecutive number of invalid AC coefficients out of two or more AC coefficients generated from each pixel block is determined, and after that when coding for another image frame is carried out, the coding for each pixel block is stopped by using the coding stopping condition. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention divides a plurality of image frames into a predetermined number of pixel blocks, orthogonally transforms image data in each pixel block, and then quantizes a plurality of AC coefficients generated from a low frequency side to a high frequency side. The present invention relates to an image data encoding method that sequentially encodes images and an image processing apparatus that encodes image data using the same method.

  Conventionally, a baseline process provided by MPEG (Moving Pictures Experts Group) has been widely used as a compression method by encoding a large amount of image data in a plurality of image frames.

  In the image data encoding method in the baseline process, image data is encoded as follows.

  First, the first image frame is divided into pixel blocks each consisting of 8 pixels × 8 pixels.

  Next, 64 transform coefficients are generated by performing a discrete cosine transform (DCT), which is a typical orthogonal transform, on the image data of the first pixel block.

  Next, quantization is performed by dividing each conversion coefficient by a preset conversion characteristic value to generate one DC coefficient and 63 AC coefficients.

  Next, Huffman coding is performed on the one DC coefficient at the upper left, and then Huffman coding is performed on the 63 AC coefficients by performing zigzag scanning from the low frequency side to the high frequency side. The encoding of the image data ends (see FIG. 3).

  Next, the image data of the first image frame is encoded by performing the same Huffman encoding for all the second and subsequent pixel blocks.

  Then, the image data in each of the pixel blocks is encoded on the image data in all other image frames, thereby encoding the image data in a plurality of image frames.

  As described above, in the image data encoding method based on the MPEG baseline process, it is necessary to perform Huffman encoding in order from the low frequency side to the high frequency side for all AC coefficients.

  However, there are few cases where all AC coefficients are effective coefficients (non-zero coefficients). In particular, there are many invalid coefficients (coefficients that are 0) continuously in the high-frequency AC coefficient.

  Therefore, if Huffman encoding is performed for all AC coefficients, there is a possibility that the time required for data encoding becomes long.

Therefore, in order to shorten the time required for the encoding process, Huffman encoding is not performed for all AC coefficients, but a predetermined number of invalid AC coefficients (coefficients that are 0) are consecutive. In this case, it is assumed that the AC coefficients on the higher frequency side are all invalid coefficients (coefficients that are 0), and the Huffman coding of the AC coefficient is stopped at that time, and the AC coefficient on the higher frequency side than that is stopped. Is provided with an image data encoding method that does not perform encoding (see, for example, Patent Document 1).
Japanese Patent Laid-Open No. 8-214168

  However, when the invalid AC coefficient (coefficient that is 0) is continuously stopped when a predetermined number of predetermined AC coefficients are continuously encoded, the coefficient (effective coefficient for the AC coefficient on the higher frequency side than that) ( If a non-zero coefficient is included, a valid AC coefficient (non-zero coefficient) is not encoded, and when the encoded image data is restored, the image data is degraded. there were.

  That is, the conventional data encoding method sacrifices data reproducibility by giving priority to the processing time required for encoding.

  Therefore, in the present invention according to claim 1, a plurality of AC coefficients generated by dividing each of a plurality of image frames into a predetermined number of pixel blocks, orthogonally transforming the image data in each pixel block, and then quantizing the image data. In the image data encoding method in which encoding is performed in order from the low frequency side to the high frequency side, when encoding is performed on one image frame of a plurality of image frames, a plurality of pixels generated from each pixel block The coding stop condition for stopping the coding for each pixel block is determined based on the continuous number of invalid AC coefficients among the AC coefficients, and then coding for other image frames is performed. When performing the above, the encoding for each pixel block is stopped under the encoding stop condition.

  Further, in the present invention according to claim 2, a plurality of AC coefficients generated by dividing each of a plurality of image frames into a predetermined number of pixel blocks, orthogonally transforming the image data in each pixel block, and then quantizing the image data. In the image data encoding method in which encoding is performed in order from the low frequency side to the high frequency side, when encoding is performed on one image frame of a plurality of image frames, a plurality of pixels generated from each pixel block The coding coefficient is divided into a plurality of frequency bands, and a coding stop condition for stopping the coding for each frequency band is determined based on the continuous number of invalid AC coefficients generated in each frequency band. After that, when encoding another image frame, the AC coefficient of each pixel block is encoded for each frequency band, and the code for each pixel block is encoded. The we decided to stop the encoding stop condition.

  Further, in the present invention according to claim 3, the encoding stop condition is that the invalid AC coefficient that is continuously generated has reached the average value of the continuous number.

  Further, in the present invention according to claim 4, the encoding stop condition is that the invalid AC coefficient generated continuously reaches the minimum value of the continuous number.

  Further, in the present invention according to claim 5, the encoding stop condition is based on the condition that the continuously generated invalid AC coefficient has reached the maximum value of the continuous number.

  Further, in the present invention according to claim 6, orthogonal transform means for generating a predetermined number of transform coefficients by sequentially performing orthogonal transform of image data in a plurality of image frames for each predetermined number of pixel blocks, and this predetermined number of Quantizing means for generating a predetermined number of direct current coefficients and a predetermined number of alternating current coefficients by performing quantization on the transform coefficient, direct current encoding means for encoding the predetermined number of direct current coefficients, and a predetermined number of The storage means for storing the continuous number of invalid AC coefficients among the AC coefficients, and the predetermined number of AC coefficients are sequentially encoded from the low frequency side to the high frequency side, and the storage means stores them. The image processing apparatus has AC encoding means for stopping encoding under the encoding stop condition determined based on the continuous number of invalid AC coefficients.

  And in this invention, there exists an effect as described below.

  In the present invention according to claim 1, a plurality of AC frames generated by dividing each of a plurality of image frames into a predetermined number of pixel blocks, orthogonally transforming the image data in each pixel block, and then quantizing the low-frequency images. In the image data encoding method in which encoding is performed in order from the high frequency side to the high frequency side, a plurality of alternating currents generated from each pixel block when encoding one image frame of the plurality of image frames. An encoding stop condition for stopping encoding for each pixel block is determined based on the continuous number of invalid AC coefficients among the coefficients, and then encoding is performed on other image frames. In this case, since the encoding for each pixel block is stopped under the encoding stop condition, it is not necessary to encode all AC coefficients, so that the processing time required for encoding can be shortened. Furthermore, since the encoding stop condition determined based on the AC coefficient of one image frame is used to encode other image frames, each image is restored when the encoded image data is restored. Deterioration of the image data of the frame can be prevented.

  In the present invention according to claim 2, a plurality of AC coefficients generated by dividing a plurality of image frames into a predetermined number of pixel blocks, orthogonally transforming the image data in each pixel block, and then quantizing the low-frequency images. In the image data encoding method in which encoding is performed in order from the high frequency side to the high frequency side, a plurality of alternating currents generated from each pixel block when encoding one image frame of the plurality of image frames. The coefficient is divided into a plurality of frequency bands, and an encoding stop condition for stopping the encoding for each frequency band is determined based on the continuous number of invalid AC coefficients generated in each frequency band. Then, when encoding another image frame, the AC coefficient of each pixel block is encoded for each frequency band, and the encoding for each pixel block is performed. Since it was decided to stop under the encoding stop condition, not only can the processing time required for encoding be reduced, but in this way, if the encoding stop condition is determined individually for each frequency band of each pixel block, Even when an effective AC coefficient is included on the frequency band side, appropriate image data can be encoded, so that the restored image data can be prevented from being deteriorated thereafter.

  Further, in the present invention according to claim 3, since the encoding stop condition is based on the condition that the continuously generated invalid AC coefficient has reached the average value of the continuous number, the processing required for encoding is performed. The time can be shortened, and deterioration of the image data when the encoded image data is restored can be prevented.

  Further, in the present invention according to claim 4, the encoding stop condition is based on the condition that the continuously generated invalid AC coefficient has reached the minimum value of the continuous number. The time can be further shortened, and deterioration of the image data when the encoded image data is restored can be prevented.

  Further, in the present invention according to claim 5, since the encoding stop condition is based on the condition that the continuously generated invalid AC coefficient has reached the maximum value of the continuous number, the processing required for encoding is performed. The time can be shortened, and deterioration of the image data when the encoded image data is restored can be further prevented.

  Further, in the present invention according to claim 6, orthogonal transform means for generating a predetermined number of transform coefficients by sequentially performing orthogonal transform of image data in a plurality of image frames for each predetermined number of pixel blocks, and this predetermined number of Quantizing means for generating a predetermined number of direct current coefficients and a predetermined number of alternating current coefficients by performing quantization on the transform coefficient, direct current encoding means for encoding the predetermined number of direct current coefficients, and a predetermined number of The storage means for storing the continuous number of invalid AC coefficients among the AC coefficients, and the predetermined number of AC coefficients are sequentially encoded from the low frequency side to the high frequency side, and the storage means stores them. Since the image processing apparatus has AC encoding means that stops encoding under the encoding stop condition determined based on the continuous number of invalid AC coefficients, the processing time required for encoding can be reduced, and Conversion Thus, it is possible to provide an image processing apparatus that can prevent the image data from being degraded when the restored image data is restored.

  An image processing apparatus according to the present invention includes orthogonal transform means for generating a predetermined number of transform coefficients by sequentially orthogonally transforming image data in a plurality of image frames for each predetermined number of pixel blocks, and the predetermined number of transform coefficients. Quantizing means for generating a predetermined number of DC coefficients and AC coefficients by performing quantization, DC encoding means for encoding a predetermined number of DC coefficients, and a predetermined number of AC coefficients AC encoding means for encoding in order from the low frequency side to the high frequency side, and storage means for storing the continuous number of invalid AC coefficients detected when the AC encoding means performs encoding ing.

  In particular, the AC encoding means divides the AC coefficient in one image frame into a plurality of frequency bands for each pixel block, and stores the continuous number of invalid AC coefficients generated for each divided frequency band. The encoding stop condition for each pixel block for the AC coefficient of another image frame is determined based on the continuous number of invalid AC coefficients.

  In addition, when encoding a plurality of AC coefficients in another image frame, the AC encoding means performs encoding up to a frequency band that satisfies the above-described encoding stop condition, and in a frequency band higher than that, Encoding is not performed.

  The image processing apparatus having the above configuration divides a plurality of image frames into pixel blocks each including a predetermined number of 8 pixels × 8 pixels, and orthogonally transforms the image data in each pixel block to obtain a plurality of transform coefficients. After the generation, the plurality of transform coefficients are quantized to generate a predetermined number of DC coefficients and a predetermined number of AC coefficients. After that, the predetermined number of DC coefficients are encoded, and the predetermined number of AC coefficients are Encoding is performed in order from the low frequency side to the high frequency side.

  In particular, when encoding one image frame among a plurality of image frames, each pixel is based on a continuous number of invalid AC coefficients among a plurality of AC coefficients generated from each pixel block. When the encoding stop condition for stopping the encoding for each block is determined and then encoding is performed for another image frame, the encoding for each pixel block is stopped as described above. Stops on condition.

  Since the encoding stop condition is determined in this way, it is not necessary to encode all AC coefficients for image frames other than one image frame, so the processing time required for encoding image data is shortened. be able to.

  Furthermore, since the encoding stop condition is determined based on the image data in one image frame among a plurality of consecutive image frames, image quality degradation such as missing pixels occurs in the restored images. It can be prevented in advance.

  Also, the encoding stop condition is to divide a plurality of AC coefficients generated from each pixel block of one image frame into a plurality of frequency bands and to obtain a continuous number of invalid AC coefficients generated in each frequency band. It can also be conditional on reaching.

  Thus, if the encoding stop condition is determined individually for each frequency band of each pixel block, even if an effective AC coefficient is included on the high frequency band side, appropriate image data Since encoding can be performed, deterioration of the restored image data can be prevented thereafter.

  In addition, the encoding stop condition can also be based on the condition that the invalid AC coefficient that is generated continuously when the encoding is performed for each frequency band has reached the above average value of the continuous number. By doing so, the processing time required for encoding can be shortened, and deterioration of the image data when the encoded image data is restored can be prevented.

  In addition, the encoding stop condition can also be made on condition that the invalid AC coefficient that is continuously generated when encoding for each frequency band has reached the minimum value of the continuous number described above, By doing so, the processing time required for encoding can be further reduced.

  In addition, the encoding stop condition can also be made on condition that the invalid AC coefficient that is continuously generated reaches the maximum value of the continuous number described above when encoding for each frequency band, By doing so, it is possible to further prevent the deterioration of the image data when the encoded image data is restored.

  Hereinafter, specific embodiments according to the present invention will be described with reference to the drawings.

  FIG. 1 shows an image processing apparatus 1 that executes a data encoding method according to the present invention. Such an image processing apparatus is built in a signal processing unit such as a digital camera or a digital video camera, and After the encoded image data is encoded by the image processing apparatus, it is stored in a predetermined storage medium.

  As shown in FIG. 1, the image processing apparatus 1 includes an orthogonal transform unit 2, a quantization unit 3, a storage unit 4, a DC encoding unit 5, and an AC encoding unit 6.

  The orthogonal transform means 2 sequentially reads image data for each image frame from the image data stored in the image memory 7.

  The image data for one image frame is subjected to discrete cosine transform for each pixel block composed of 8 pixels × 8 pixels, thereby generating 64 transform coefficients in one pixel block.

  The quantization means 3 performs quantization by dividing each of the 64 transformation coefficients generated by the orthogonal transformation means 2 by a preset transformation characteristic value in order from the low frequency (DC) side to the high frequency side. As shown in FIG. 2, one DC coefficient DC and 63 AC coefficients AC1 to AC63 are generated.

  The DC encoding means 5 performs Huffman encoding for one DC coefficient DC, and writes the encoded DC component data in a predetermined memory 8.

  The AC encoding means 6 performs a Huffman encoding by zigzag scanning the AC coefficients AC1 to AC63 from the AC coefficient AC1 on the low frequency side to the AC coefficient AC63 on the high frequency side, and the encoded AC component data is predetermined Write to memory 8 (see Figure 3).

  In particular, when the image data in the image frame read first by the image processing apparatus 1 is encoded, the AC encoding means 6 converts the AC coefficients AC1 to AC63 in each pixel block into the low frequency band 9 (AC1 to AC1). It is divided into three frequency bands: AC9), medium frequency band 10 (AC10 to AC35), and high frequency band 11 (AC36 to AC63) (see FIG. 4).

  Then, the number of consecutive occurrences of 0 which is an invalid AC coefficient generated in each frequency band 9, 10, 11 is detected for each frequency band 9, 10, 11 (hereinafter referred to as “0 run length”). This 0 run length is stored in the storage means 4, and the same processing is performed for all the pixel blocks of the image frame to be read first.

  At the same time, the AC encoding means 6 determines the condition for stopping the encoding of the image data in the image frame read from the second time onward, based on the plurality of 0 run lengths stored in the storage means 4.

  The image data encoding method according to the present invention will be specifically described below with reference to the flowcharts shown in FIGS.

  When encoding the image data in the first image frame, first, as shown in the flowchart of FIG. 5, the orthogonal transformation means 2 reads the image data included in the first image frame from the image memory 7 ( Step S1).

  Next, the orthogonal transform unit 2 generates a plurality of transform coefficients by performing discrete cosine transform, which is orthogonal transform, for each pixel block of 8 pixels × 8 pixels with respect to the image data read from the image memory 7 ( Step 2).

  Next, the quantizing means 3 quantizes the plurality of transform coefficients generated by the direct transform means 2 by dividing by a preset transform characteristic value, and one DC coefficient DC and 63 AC coefficients (AC1 to AC63) are generated (step 3).

  Next, the DC encoding means 5 encodes the DC coefficient DC (step S4).

  Next, the AC encoding unit 6 converts the AC coefficients AC1 to AC63 included in one pixel block into a low frequency band 9 (AC1 to AC9), a medium frequency band 10 (AC10 to AC35), and a high frequency band 11 ( It is divided into three frequency bands (AC36 to AC63) (step 5).

  Next, the AC encoding means 6 encodes an AC coefficient for each frequency band 9, 10, and 11.

  At the same time, the 0 run length generated in the low frequency band 9, the 0 run length generated in the medium frequency band 10, and the 0 run length generated in the high frequency band 11 are respectively detected and stored in the storage means 4 ( Step 6).

  Next, if all the pixel blocks have not been encoded, the process returns to step S4 to encode the next pixel block, and when all the pixel blocks have been encoded, the process proceeds to step S8 (step S7).

  Next, the AC encoding unit 6 performs encoding when encoding the image data in the image frame read from the second onward based on the 0 run length stored in the storage unit 4 for each frequency band. An encoding stop condition for stopping is determined (step S8).

  Specifically, using a plurality of 0 run lengths stored in the storage means 4, the average value a of 0 run lengths in each low frequency band 9 and the average value b of 0 run lengths in each medium frequency band 10 The average value c of 0 run lengths in each high frequency band is obtained in advance.

  When the encoding stop condition is determined in this way, the encoding of the image data in the first image frame is finished, and the image data in the next image frame is encoded.

  When encoding the image data in the second and subsequent image frames, as shown in the flowchart of FIG. 6, first, the orthogonal transform means 2 starts the image data included in the second and subsequent image frames from the image memory 8. Are sequentially read frame by frame (step S9).

  Next, the orthogonal transform unit 2 generates a plurality of transform coefficients by performing discrete cosine transform, which is orthogonal transform, for each pixel block of 8 pixels × 8 pixels with respect to the image data read from the image memory 7 ( Step S10).

  Next, the quantizing means 3 quantizes the plurality of transform coefficients generated by the direct transform means 2 by dividing by a preset transform characteristic value, and one DC coefficient DC and 63 AC coefficients (AC1 to AC63) are generated (step S11).

  Next, the DC encoding means 5 encodes the DC coefficient DC (step S12).

  Next, the AC conversion means 6 encodes the AC coefficients AC1 to AC9 in the low frequency band 9 (step S13).

  At this time, if the 0 run length detected in the low frequency band 9 has reached the average value a of 0 run length in the low frequency band 9 determined in step S8, encoding of the AC coefficient is stopped and the process proceeds to step S20. Then, EOB (End Of Block) indicating the end of encoding is inserted after the last effective AC coefficient in the low frequency band 9.

  On the other hand, if the 0 run length detected in the low frequency band 9 does not reach the average value a of 0 run lengths in the low frequency band 9 determined in step S8, the process proceeds to step S15 (step S14).

  Next, the AC encoding means 6 encodes the AC coefficients AC10 to AC35 in the medium frequency band 10 (step S15).

  At this time, if the 0 run length detected in the middle frequency band 10 reaches the average value b of 0 run length in the middle frequency band 10 determined in step S8, the encoding of the AC coefficient is stopped and the process proceeds to step S20. Then, an EOB representing the end of encoding is inserted after the last effective AC coefficient in the medium frequency band 10.

  On the other hand, if the 0 run length detected in the intermediate frequency band 10 does not reach the average value b of the 0 run length in the intermediate frequency band 10 determined in step S8, the process proceeds to step S17 (step S16).

  Next, the AC encoding means 6 encodes the AC coefficients AC36 to AC63 in the high frequency band 11 (step S17).

  At this time, when the 0 run length detected in the high frequency band 11 reaches the average value c of the 0 run length in the high frequency band 11 determined in step S8, the encoding of the AC coefficient is stopped and the process proceeds to step S20. Then, an EOB representing the end of encoding is inserted after the last effective AC coefficient in the high frequency band 11.

  On the other hand, if the 0 run length detected in the high frequency band 11 does not reach the average value c of the 0 run length in the high frequency band 11 determined in step S8, the process proceeds to step S19 (step S18).

  Next, the AC encoding means 6 encodes the AC coefficient of the high frequency band 6 to the end (step S19).

  Next, EOB is inserted after the last effective AC coefficient in the high frequency band 11 (step S20).

  Next, if encoding of all pixel blocks has not been completed, the process returns to step S12, encoding of the next pixel block is performed, and encoding of all pixel blocks is completed, encoding of this image frame is performed. End and start encoding the next image frame.

  Then, the encoding process similar to the above is performed on all the image frames, and the encoding of the image data is completed.

  By encoding image data in this way, the processing time required for encoding can be shortened, and deterioration of the image data when the encoded image data is restored can be prevented.

  In this embodiment, the average value of the 0 run lengths generated in each frequency band of each pixel block constituting the first image frame is used as the 0 run length which is the encoding stop condition. If the minimum value of the 0 run length generated in the above is used, the number of AC coefficients to be encoded is reduced, so that the processing time required for encoding can be further reduced.

  In addition, if the maximum value of the 0 run length generated in each frequency band is used, the number of AC coefficients to be encoded increases, so that deterioration of the image data when the encoded image data is restored is further prevented. be able to.

It is explanatory drawing which shows the image processing apparatus which concerns on this invention. It is explanatory drawing which shows the direct current coefficient and alternating current coefficient in 1 pixel block. It is explanatory drawing which shows a zigzag scan. It is explanatory drawing which shows each frequency band. It is explanatory drawing which shows the image data encoding method of the 1st image frame. It is explanatory drawing which shows the image data encoding method of the 2nd image frame or later.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Orthogonal transformation means 3 Quantization means 4 Memory | storage means 5 DC encoding means 6 AC encoding means 7 Image memory 8 Memory 9 Low frequency band 10 Medium frequency band 11 High frequency band
DC DC coefficient
AC1 to AC63 AC coefficient

Claims (6)

A plurality of AC frames generated by dividing a plurality of image frames into a predetermined number of pixel blocks and quantizing the image data in each pixel block after orthogonal transformation are sequentially encoded from the low frequency side to the high frequency side. In the image data encoding method
When encoding one image frame of the plurality of image frames, each pixel block is based on a continuous number of invalid AC coefficients among a plurality of AC coefficients generated from each pixel block. When the encoding stop condition for stopping the encoding for each image is determined and then encoding is performed on another image frame, the encoding for each pixel block is performed under the encoding stop condition. An image data encoding method characterized by stopping. A plurality of AC frames generated by dividing a plurality of image frames into a predetermined number of pixel blocks and quantizing the image data in each pixel block after orthogonal transformation are sequentially encoded from the low frequency side to the high frequency side. In the image data encoding method
When encoding one image frame among the plurality of image frames, a plurality of AC coefficients generated from each pixel block are divided into a plurality of frequency bands and generated in each frequency band. Based on the consecutive number of invalid AC coefficients that have been determined, the encoding stop condition for stopping the encoding for each frequency band is determined, and then when performing encoding on other image frames, An image data encoding method, wherein an AC coefficient of each pixel block is encoded for each frequency band, and encoding for each pixel block is stopped under the encoding stop condition.   3. The image data encoding method according to claim 1, wherein the encoding stop condition is that the invalid AC coefficient generated continuously reaches an average value of the continuous number. .   3. The image data encoding method according to claim 1, wherein the encoding stop condition is that an invalid AC coefficient that is continuously generated reaches a minimum value of the continuous number. .   The image data encoding method according to claim 1 or 2, wherein the encoding stop condition is that an invalid AC coefficient that is continuously generated reaches a maximum value of the continuous number. . Orthogonal transform means for generating a predetermined number of transform coefficients by orthogonally transforming image data in a plurality of image frames sequentially for each predetermined number of pixel blocks;
Quantization means for generating a predetermined number of direct current coefficients and a predetermined number of alternating current coefficients by performing quantization on the predetermined number of transform coefficients;
DC encoding means for encoding the predetermined number of DC coefficients;
Storage means for storing a continuous number of invalid AC coefficients among the predetermined number of AC coefficients;
The predetermined number of AC coefficients are encoded in order from the low frequency side to the high frequency side, and the encoding stop condition is determined based on the continuous number of invalid AC coefficients stored by the storage means. An image processing apparatus comprising: AC encoding means for stopping encoding.

Images