JP3470763B2 - Image encoding and decoding device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding and decoding apparatus for coding or decoding an image in a facsimile apparatus or the like in order to compress the image signal and transmit it efficiently.

[0002]

2. Description of the Related Art Generally, an image signal has an extremely large amount of information because it is composed of a binary or multi-valued bit stream for a large number of pixels. Therefore, if this is sent to the communication path as it is, in the case of a facsimile machine or the like, the transmission / reception time becomes long. Also, when storing in a memory or the like, a large amount of that memory is used. Therefore, various techniques for compressing this type of signal with high efficiency have been conventionally developed. "Three-dimensional Markov coding of multi-valued image by bit plane configuration" (13th
A technology relating to high-efficiency coding of image signals is also introduced in the Conference on Imaging Technology P219-222. Here, a bit plane coding technique, which is one of the coding methods for multi-valued images, is introduced. This is a method in which a multilevel signal is converted into a plurality of binary signals and a binary image coding method is applied. When a multi-valued image is decomposed and converted into a binary image, since there is a correlation between the binary signals, it is possible to improve coding efficiency by using a three-dimensional Markov model in which other bit planes are included in reference pixels. .

FIG. 2 is a block diagram showing an example of such a conventional image coding apparatus. This apparatus includes a bit plane image generation unit 1 and a bit plane memory 2
And a buffer 3, a coding condition determination unit 4 and a coding unit 5. A still image signal sequence 1A composed of multi-valued image signals is supplied to the bit plane image generation unit 1. The bit-plane image generation unit 1, the digital multi-value image signal, a circuit for converting a plurality of binary bit-plane image obtained by collecting the corresponding bit, respectively.

FIG. 3 shows an explanatory diagram of a bit plane image. As shown in the figure, each pixel of the digital multilevel still image 6 is composed of a multilevel image signal. If this multi-valued image signal is, for example, an image signal of 4-bit structure, the content of a certain image signal is "01" as shown in this figure.
10 ". Such a digital multilevel still image 6
Is four binary bit plane images 7-1, 7-2, 7
-3 and 7-4. The binary bit plane image 7-1 is configured by collecting only the most significant bits of each pixel of the digital multilevel still image 6. The binary bit plane image 7-2 is formed by collecting only the second bit, the binary bit plane image 7-3 is the third bit, and the binary bit plane image 7-4 is the fourth least significant bit. ing.

Such a binary bit plane image is stored in the bit plane memory 2 shown in FIG. The buffer 3 is a circuit for converting a binary bit plane image into a binary image signal series 3A and outputting it to the encoding unit 5. This device achieves high-efficiency encoding of a binary image signal sequence 3A sent from the buffer 3 to the encoding unit 5. For example, the binary bit plane image 7-4 shown in FIG.
When the pixel 9 among them is the encoding target, not only the peripheral pixels but also the pixels of the other bit planes 7-1 to 7-3 are referred to improve the encoding efficiency. The coding condition determination unit 4 is a circuit for estimating such a coding condition. The bit plane memory 2 stores a binary bit plane image as shown in FIG. 3, and outputs a reference image signal 2A to the coding condition determining unit 4 so that a preset pixel therein can be referred to.

The coding condition determination unit 4 estimates the symbol appearance prediction probability from the reference image signal 2A. The symbol appearance prediction probability is “0” for each of the plurality of reference image signals.
It is determined for each fixed combination of "1" and "1". The coding condition determination unit 4 generates a coding parameter 4A based on this symbol appearance prediction probability. Then, the encoding unit 5
Is a binary image signal sequence 3A input from the buffer 3,
Binary entropy coding is performed using the coding parameter 4A, and a coded sequence 5A is output. JP-A-4-35
In Japanese Patent Publication No. 469, the same binary bit plane image is identified by the number of times of coincidence and non-coincidence between the value of the target pixel predicted by a plurality of pixels near the target pixel to be encoded and the actual value of the target pixel A technique for improving image coding efficiency by appropriately switching between a parameter determined so that the prediction probability is high for each image and a parameter for which the prediction probability is almost medium for a general image. Has been introduced. Conventionally, high-efficiency encoding of multi-valued images has been attempted by the above method.

[0007]

By the way, the above-mentioned prior art is generally directed to a monochrome still image having a relatively small number of gradations such as a halftone facsimile.
However, with the improvement in image processing technology, there is an increasing demand for similar high-efficiency encoding in transmission and storage of moving images and color images. In this case, for example, in the case of a moving image, the movement of the image should be taken into consideration, or in the case of a color image, the contents of other color components and the like should be taken into consideration to further improve the coding efficiency. Is.
In this case, it is difficult to perform sufficient high-efficiency encoding by a method of setting a constant reference pixel and encoding the binary bit-plane image obtained from the digital multi-valued still image as in the conventional case. . The present invention has been made paying attention to the above points, and not only ordinary document images but also various types of images such as multi-valued images, moving images, and color images are collectively encoded with high encoding efficiency. It is an object of the present invention to provide an image encoding / decoding device capable of performing the above.

[0008]

The first aspect of the present invention, in order to solve the problems] includes a frame image generating unit for converting the plurality of frame images at successive points were on the axis of the motion picture signal sequence time before Symbol each <br / > A bit plane image generation unit for converting into a plurality of binary bit plane images obtained by assembling corresponding bits of a digital multi-valued image signal forming a frame image, and a description of the plurality of binary bit plane images . A bit plane memory to be stored and the binary bit plane image,
A buffer that converts the binary image signal sequence, the
At least the previous frame image in bitplane memory
A coding condition determination unit that determines a coding condition by referring to a binary bit plane image obtained by conversion, and a binary image signal sequence output from the buffer using the coding condition and output the coded signal sequence. It is characterized by comprising an encoding unit for
It

According to a second aspect of the present invention, a bit plane memory for storing a decoded binary bit plane image of a previous frame image, and decoding with reference to the binary bit plane image in the bit plane memory A decoding condition determining unit that determines the condition; a decoding unit that receives the encoded binary image signal sequence and decodes using the decoding condition;
A buffer for storing a plurality of binary bit plane images obtained from the decoded binary image signal sequence, and a frame image composed of a digital multi-valued image signal are reproduced from the plurality of binary bit plane images. The present invention is characterized by including a bit plane image integration unit and a frame image integration unit that obtains a moving image signal sequence from a plurality of frame images that are continuously arranged on the time axis.

In the image coding apparatus of the first aspect of the invention, a motion detection unit for comparing the frame images continuously arranged on the time axis with each other to detect the motion of the moving image, and In order to shift the reference position of the reference pixel set in the previous frame in the opposite direction to the motion of the moving image,
An encoding condition determination unit that determines the encoding condition may be provided.
Yes. Further, the same image coding apparatus may be provided with a mode setting unit that sets the number of converted frame images per fixed time according to the required received image quality .

The image encoding apparatus may be provided with a mode setting unit for excluding the image on the lower bit side of the binary bit plane image from the object of encoding according to the required received image quality . Further, in the image encoding device ,
A binary coded signal already decoded on the decoding side, and a minimum code amount estimating section for calculating the optimum encoding condition by operating the encoding side and the decoding side using the same arithmetic processing method; A coding condition determination unit that applies the optimum coding condition to the coding condition of the subsequent binary image signal may be provided.

Further , in the image encoding apparatus, a plurality of reference pixel models for performing encoding by referring to any one of the binary bit plane images that have already been encoded are prepared, and an optimum one is selected from them. A coding condition determination unit that selects the reference pixel model and determines the optimum coding condition may be provided.
Yes. In this case, the binary image signal corresponding to the upper bits in the binary bit plane image that has already been encoded is used for selecting the reference pixel model.

Further , in the image encoding apparatus , the binary image signals corresponding to different color components in the already encoded binary bit plane image are used for selecting the reference pixel model.

[0014]

In this device, the frame image generation unit converts the moving image signal sequence into a plurality of continuous frame images on the time axis. Each frame image is converted into a binary bit plane image by the bit plane image generation unit. A fixed amount of binary bit plane images of consecutive frame images are collectively stored in the bit plane memory. Therefore, the encoding condition includes not only the image signals of the peripheral pixels of the image signal to be encoded but also other binary bit planes and pixels of other frames. In the case of a moving image, the coding condition can be determined in consideration of the motion of the moving image. If the number of converted frame images per fixed time is reduced according to the required received image quality, the amount of information about the moving image is sufficiently reduced, and high efficiency can be achieved. For moving images with little motion, such a method has a great effect. Further, the information amount can be compressed by excluding the image on the lower bit side of the binary bit plane image from the encoding target. This is a compression method suitable when the precision of the image is not so required.

It is preferable that the position of the reference image signal and the coding condition are dynamically changed according to the content of the image. In this case, if the coding condition is transferred to the decoding side, the amount of information to be transferred increases. Therefore, the encoding side and the decoding side use the same arithmetic processing method to estimate the optimum encoding condition, and the transmission of this kind of information is omitted. In this case, the optimum coding condition of the already decoded binary image signal is estimated and applied to the coding condition of the subsequent binary image signal, and the coding conditions of the coding side and the decoding side are set. Match. If a plurality of reference pixel models suitable for high-efficiency encoding are prepared in advance and the optimum reference pixel model is selected from among them, the determination of the encoding condition becomes efficient.

Further, by utilizing the higher order bits or the binary image signals of different color components for the selection of the reference pixel model,
The determination of the reference pixel model becomes efficient. Further, if the binary bit plane image is encoded in the direction from the upper bit to the lower bit in order and the binary image signal corresponding to the upper bit is included in the reference pixel, all signals are received on the receiving side. The content of the received image can be predicted before, and the coding efficiency can be improved for images having strong correlation between planes. For color images, if binary image signals corresponding to different color components in a binary bit plane image that has already been encoded are included in reference pixels, a characteristic image signal appears at the same position. The efficiency can be increased.

It is necessary for the encoder to issue a request for performing a complicated encoding process according to various encoding conditions. In this case, rather than newly designing a coding unit having a complicated coding function, it is better to add a coding unit having an additional function to the conventional device and switch the coding unit according to the required function. In some cases, the circuit can be designed at low cost. The switching unit performs an operation of selecting an encoding unit according to an encoding condition to satisfy such a request.

[0018]

The present invention will be described in detail below with reference to the embodiments shown in the drawings. [Image Coding Device] FIG. 1 is a block diagram showing an embodiment of the image coding device of the present invention. This device includes a frame image generation unit 11, a bit plane image generation unit 12, a bit plane memory 13, a buffer 14, an encoding condition determination unit 15, and an encoding unit 16. A characteristic part of the apparatus of the present invention is that the frame image generation unit 11 is provided, as can be seen by comparison with the conventional apparatus shown in FIG. The frame image generation unit 11 is configured to receive the moving image signal series 10A. Further, this apparatus receives the still image signal sequence 10B in the bit plane image generation unit 12 and further directly receives the binary image signal sequence 10C in the buffer 14, so that not only a moving image but also a still image or a general binary image is received. Is also capable of high efficiency encoding.

The frame image generation unit 11 is a circuit for converting the moving image signal series 10A into a plurality of frame images at each continuous point on the time axis. FIG. 4 shows an explanatory diagram of such a frame image. As shown in the figure, the moving image is an image in which the drawn image changes with time, and when viewed in the direction of the time axis T as shown in the figure, each still image is a continuous point on the time axis. Of the nature that exists in. This still image is called a frame image 11A. Conventionally, when performing high-efficiency encoding of images,
Each frame image 11A is individually subjected to a certain encoding process. However, in the present invention, the plurality of frame images 11A are fetched into the bit plane memory 13 via the bit plane image generating unit 12 shown in FIG. 1, and the reference image signal is obtained from any pixel of the plurality of frame images, Determine the coding conditions. The configuration and operation of the bit plane image generation unit 12 are the same as those of the conventional apparatus already described with reference to FIG. The output of the bit plane image generation unit 12 is the bit plane memory 13
And to the buffer 14. The configuration of the buffer 14 and the configuration of the encoding unit 16 are similar to those of the conventional device.

In the present invention, the bit plane memory 1
3 stores signals having the following contents. FIG. 5 shows an explanatory diagram of the bit plane memory configuration. As shown in the figure,
The bit plane memory 13 is configured to continuously store binary bit planes composed of several frame images 11A. For example, in the case of a black and white image with 256 gradations, one image signal is represented by 8 bits. Therefore, eight binary bit plane images are obtained from one frame image. Further, in the case of a full-color image, since there are three color components of red, green, and blue, 24 binary bit plane images, which is three times as large as in the case of a monochrome image, constitutes one frame image. As shown in FIG. 5, the bit-plane memory 13 stores, for example, k binary bit-plane images in the order in which they are input, and can refer to image signals of pixels at arbitrary positions in each plane. Has become.

Returning to FIG. 1 again, the coding condition determining unit 15
Is the reference image signal 13A from the bit plane memory 13
Is a circuit for estimating the symbol appearance prediction probability of the encoding target image signal according to the state of the reference image signal, setting the symbol appearance prediction probability as the encoding parameter 15A, and outputting it to the encoding unit 16. The encoding unit 16 uses this encoding parameter to entropy-encode the binary image signal sequence input from the buffer 14. All of these operations are similar to those of the conventional device, and further detailed description will be omitted.

FIG. 6 is an explanatory view (1) of the reference image signal, FIG. 7 is an explanatory view (2) of the reference image signal in another case, and FIG. 8 is an explanatory view of the reference image signal in another case. (Part 3) is shown. In these figures, reference image signal positions according to the device of the present invention are illustrated. First, FIG.
In the above example, the reference image signal 22-1 corresponding to four pixels in the same frame is taken into the encoding target image signal 21, while the reference corresponding to three pixels of the upper binary bit plane image is taken. The image signal 22-2 is also accepted. In addition, the reference image signal 2 is extracted from each of the three binary bit plane images provided in the frame immediately before the moving image.
It accepts 2-3, 22-4, and 22-5.

In the example of FIG. 7, only the reference image signal 22-1 in the same binary bit-plane image and the reference image signal 22-2 in the binary bit-plane image of the high-order bit generated immediately before it are used. The image signal 21 is received and encoded. As shown in FIG. 6, by referring to the pixel at the same position in the immediately preceding frame image or a pixel in the vicinity thereof, correlation between the frame images can be obtained and the coding efficiency can be improved. . Also in the case of a moving image signal, there is a correlation with other bit planes in the same frame image, and neighboring pixels in the upper plane or the same plane are used as reference image signals.

On the other hand, in case a moving picture of a scene change or large motion such images, since frame correlation is small, as shown in FIG. 7, the upper in the current frame image
The coding efficiency can be improved only by referring to the neighboring pixels of the plane on the bit side . Further, (a) and (b) shown in FIG. 8 show examples of reference image signals in the case of a still image or a binary image. That is, in this example, only the peripheral pixels that have already been encoded are used as the reference image signal 22-1 as in the conventional case.

[Image Decoding Device] FIG. 9 shows a block diagram of an embodiment of the image decoding device of the present invention. This device is a device that receives the coded sequence 16A from the above-described coding device and obtains the moving image signal sequence 36A, the still image signal sequence 35A, or the binary image signal sequence 34A. This apparatus includes a decoding unit 31, a decoding condition determination unit 32, a bit plane memory 33, a buffer 34, a bit plane image integration unit 35, and a frame image integration unit 36.
The decoding unit 31 is a unit that decodes the coded sequence 16A according to the decoding parameter 32A output from the decoding condition determining unit 32. The decoding device is provided with a bit plane memory 33 having the same configuration as the encoding device. The output of the decoding unit 31 is taken out as a binary image signal sequence 34A via the buffer 34.

The bit plane image integration unit 35 performs an operation opposite to that of the bit plane image generation unit 12 of the encoding apparatus shown in FIG. 1, that is, an operation for converting a binary bit plane image into a digital multilevel still image. It is a circuit to perform.
The frame image integration unit 36 is a circuit that obtains a moving image signal sequence 36A by arranging the digital multi-valued still image signal sequence 35A as a frame image on the time axis.

[Encoding Condition and Decoding Condition] In the present invention, the image decoding device as described above performs an operation reverse to that of the encoding device and executes the decoding process. In the present invention, the decoding condition in this case is uniquely determined on the decoding side, not sent from the encoding device. This is because the coding condition may be fixed, but if the switching is performed according to the content of the image, the coding efficiency is further improved. At this time, the encoding condition and the decoding condition dynamically change according to the position of the image signal to be encoded, and therefore the following processing is performed.

FIG. 10 shows an explanatory diagram of a method for determining the coding / decoding conditions. As shown in the drawing, the bit plane memory 13 on the encoding side stores an image signal 22-0 that has been encoded immediately before the image signal 21 to be encoded. The same image signal 22-0 is also stored in the bit plane memory 33 on the decoding side. These are the encoding condition determining unit 15 (FIG. 1) and the decoding condition determining unit 32.
(FIG. 9), the encoding or decoding condition is determined. That is, when the same binary bit plane image is referred to and the encoding condition or the decoding condition is determined by the same arithmetic processing, the encoding side and the decoding side do not need to exchange information about the condition. The same conclusion can be obtained and the encoding and decoding are performed under the same condition.

For example, as a method in this case, for the binary image signal 22-0 which has already been encoded immediately before and the decoding has been completed on the decoding device side, the same condition is set on the encoding side and the decoding side, respectively. The calculation process for estimating the symbol appearance prediction probability is performed. In this case, since the binary bit plane image to be referred to and the image signal to be obtained as a result are clear on the encoding side and the decoding side, respectively, the optimum encoding condition can obtain the same result. Should be. This is applied to the next encoding target image signal 21, and the encoding process is executed. Therefore, the decoding side also applies the decoding condition to the next encoding target image signal. By doing so, it becomes possible to dynamically change the encoding condition and the decoding condition.

[Encoding of Moving Image] FIG. 11 shows a block diagram of an embodiment of the image encoding device of the present invention. The apparatus of this embodiment has a motion detection section 41 and a motion information coding section 42 added to the apparatus shown in FIG. The motion detector 41
The frame image signal is fetched from the frame image generation unit 11 and the respective frame images are compared with each other to detect the motion of the moving image. The motion is output to the motion information coding unit 42 as a motion vector, for example. Motion information coding unit 4
2 receives the output 41A of the motion detection unit 41, encodes the output 41A, and sends it as the motion information 42A to the decoding side.

The output 41A of the motion detecting section 41 is supplied to the coding condition determining section 15 via the bit plane memory 13. This embodiment utilizes the temporal correlation of moving images to achieve high efficiency coding. That is, the motion detector 4
The motion between frames is detected by 1 and the position of the reference pixel one frame before is moved by the size of the detected motion vector to improve the coding efficiency of the moving image. The bit plane memory 13 outputs the output 41A of the motion detection unit 41.
Is accepted, the reference pixel position of the previous frame is changed based on the motion information. For example, when it is detected that the current frame has moved one pixel to the left from the previous frame, the reference pixel position of the previous frame is moved to the right by one pixel. This makes it possible to efficiently extract the temporal correlation of the frame image and improve the coding efficiency. The contents will be described more specifically with reference to the drawings.

FIG. 12 is an explanatory diagram of motion information and a reference image signal. (A) is an example of a reference pixel signal. Four peripheral pixels of the pixel to be encoded and four pixels of the previous frame are selected as reference pixels. When the pixel to be encoded is motion detected, (b)
Consider the case where a motion is detected as in. (C)
Is a reference pixel position when motion information is not used. Since there is a gap between the previous frame and the current frame due to motion,
Even if the reference image signal 22A is selected as the reference pixel of the encoding target image signal 21, the encoding efficiency does not improve so much.
Therefore, as shown in (d), the reference image signal 2 of the previous frame having a strong correlation with the encoding target image signal 21 using the motion information.
2A is selected as a reference pixel. Since such motion information is also required on the decoding side, the motion information encoding unit 4
2 transmits the motion information 42A to the decoding side. Thus
Since each frame image of a moving image is used for encoding, the encoding efficiency can be improved much more than before.

[Mode Setting] FIG. 13 shows a block diagram of an embodiment of the coding apparatus of the present invention. This device is obtained by adding a mode setting unit 43 to the device of the embodiment shown in FIG. The apparatus of the embodiment of FIG. 1 already described encodes the binary bit plane image in the order from the most significant bit side to the least significant bit side, and sends this to the decoding side in order, and the decoding side also produces the image in this order. To play. In general, the upper bit side binary bit plane image has global information of the image.
Therefore, when the decoding side sequentially decodes the binary bit plane image from the upper bits, it becomes possible to grasp what kind of image is reproduced in the initial stage of the decoding.

When the decoding side does not request an image of a certain degree or more, it is possible to improve the transmission amount and the signal rate by omitting or partially thinning out the received image signal. The mode setting unit 43 is a circuit for determining such a transmission mode and outputting the instruction to the frame image generation unit 11 and the bit plane image generation unit 12. For example, when the speed of the transmission path is slow or smooth motion of the moving image is not required, the control signal 43A for limiting the number of converted frame images per fixed time is sent to the frame image generation unit 11.

FIG. 14 shows a mode explanatory diagram in which the number of such converted frame images is reduced. For example, as shown on the left side of the figure, it is assumed that frame images are continuously arranged at a constant density with respect to the time axis T. In this case, the mode setting unit 43 outputs to the frame image generation unit 11 an instruction to pick up every other frame image or every few frames. Frame image generator 1
1 outputs an appropriate number of frame images when outputting the frame images to the bit plane image generation unit 12. As a result, the number of signals to be encoded is reduced, and it is possible to correspond to the transmission speed or the transmission capacity.

Further, as described above, the binary bit plane image has more important information indicating the general nature of the image as the upper plane. The lower plane has information that represents subtle changes that cannot be seen by the human eye. Therefore, its importance becomes lower as it goes to the lower plane. When the decoding side does not require a high-precision image, it may be sufficient to encode a binary bit plane image having only a few upper bits. In such a case, the mode setting unit 43 outputs the control signal 43B to the bit plane image generating unit 12 and controls it.

FIG. 15 shows a mode explanatory diagram excluding the lower bit side. As shown in the figure, only a few binary bit plane images are selected from the upper side and used. FIG.
The bit plane image generation unit 12 shown in FIG.
It outputs to 3 and becomes the object of encoding. For example, assuming that a black and white natural image is represented by 8 bits and 256 gradations, and when only the upper 5 bits and 32 gradations are encoded and transmitted, the SN ratio is about 40 dB, which is the original image for human eyes. It is almost impossible to tell the difference between the decoded image and the decoded image. However, in this case, the code amount is less than half that in the case of encoding all 8 bits. The mode setting as described above is set in the mode setting unit 43 on the encoding side in advance, and, for example, when the operator monitors the received image on the decoding side and a sufficient image quality is obtained, the reception is performed. It can be realized by stopping the transmission or requesting the transmission side to stop the transmission. Therefore, the mode setting unit may be provided on the decoding side.

[Use of Reference Pixel Model] FIG. 16 shows a block diagram of an embodiment of the image coding apparatus of the present invention. The apparatus shown in FIG. 16 is obtained by adding a reference pixel model determination unit 45 and a reference pixel model table 46 to the apparatus shown in FIG. According to the apparatus of the present invention, it is possible to freely and dynamically receive a reference image signal from a wide range according to a pixel to be encoded and perform encoding. In this case, some kinds of reference pixel models corresponding to the characteristics of the image are prepared in advance, and by adaptively switching each time, it becomes possible to perform coding with high efficiency and quickly.

The reference pixel model determination unit 45 is a part that performs state prediction using this reference pixel model and estimates the appearance prediction probability of a coded symbol. At this time, the reference pixel model determination unit 45 selects an optimum reference pixel model for encoding the next pixel based on the specific pixel that has already been encoded. The reference pixel model is used to determine the coding condition for the next pixel. The reference pixel model table 46 includes a memory storing various reference pixel models, and when the reference pixel model information 45A output from the reference pixel model determination unit 45 is input to the coding condition determination unit 15, the corresponding reference pixel models are It is read from the reference pixel model table 46 and the coding condition is determined. The other parts, that is, the configurations and operations of the bit plane memory 13, the buffer 14, the encoding unit 16 and the like are similar to those of the device shown in FIG.

FIG. 17 shows a reference pixel model explanatory view.
For example, the reference pixel model shown in FIG. 17A is a model in which the peripheral 10 pixels encoded immediately before the encoding target image signal 21 are used as the reference image signal 22 and the correlation is large.
The reference pixel model of (b) takes in one pixel adjacent to the encoding target image signal 21 as a reference image signal, and also takes in two corresponding pixels of the upper plane as a reference image signal 22. This is a model with a small correlation in the same plane. For example, when two types of reference pixel models are prepared in this way and the binary bit plane is encoded while adaptively changing the models, for example, only the reference pixel model shown in FIG. 17A is used. The code amount can be reduced by about 25% as compared with the case where there is a case.

The synchronization between the encoding side and the decoding side in this case is also
The procedure is the same as described above. First, for a pixel that has already been encoded, the reference pixel model determination unit 45 estimates what model should be used when encoding the symbol to obtain high encoding efficiency. As a result, the reference pixel model information 45A is sent to the coding condition determination unit 15, and the coding condition for the next pixel is determined. The decoding side also determines the reference pixel model under exactly the same conditions. Therefore, in this case, since the decoding condition is determined using only the pixels that have already been decoded,
The same reference pixel model is selected on the encoding side and the decoding side. This eliminates the need to transmit the selected reference pixel model information from the encoding side to the decoding side. Then, the coding condition determination unit 15 estimates the symbol appearance prediction probability using the reference pixel model, and sends the coding parameter 15A to the coding unit 16. The encoding unit 16 encodes the symbol output from the buffer 14 using the encoding parameter 15A and outputs the encoded sequence 16A.

The reference pixel model selection criteria as described above are as follows when the second bit plane is encoded. First, consider a model having a large correlation with the neighborhood of the coded pixel and a model having a small correlation. In this case, for example, in the case of an image in which the gradation value continuously changes, the first
In a pixel in which the bit plane changes from “0” to “1”, a pixel at the same position in the second bit plane often changes from “0” to “1”. Also, in a model that refers to many neighboring pixels, "0" or "1"
Shows very good coding efficiency when
Coding efficiency decreases when "0" and "1" are alternately inverted. By using this feature, for pixels where “0” and “1” do not change in the first bit plane, a reference pixel model having a large correlation with neighboring pixels is selected, and “0” and “1” are selected in the first bit plane. In the case of a pixel that changes, the reference pixel model that has a small correlation with neighboring pixels is selected. Similarly, when encoding the third and subsequent bit planes, the reference pixel model is selected using the inversion of "0" and "1" of the upper bit plane. The optimum reference pixel model is selected based on the above criteria.

[Use of Selected Model Memory] FIG. 18 is a block diagram showing an embodiment of the image coding apparatus of the present invention.
In this example, the selection model memory 4 is different from the embodiment of FIG.
8 and the minimum code amount estimation unit 47 are added. The minimum code amount estimation unit 47 is a circuit for dynamically estimating the reference pixel model and storing the result in the selection model memory 48. FIG. 19 shows an explanatory diagram of the selection model memory. Figure 1
In the sixth embodiment, the next pixel is encoded with reference to what reference pixel model is used for the pixel already encoded immediately before. On the other hand, in this embodiment, when the minimum code amount estimation unit 47 encodes the image signal of all the pixels that have been encoded in the past, which reference pixel model is used among the plurality of models to perform the encoding. After conversion, it is determined whether or not the code amount is the smallest and stored in the selection model memory 48 each time. The selection may be made in reality, but if the selection is incorrect, it is preferable to correct it and store it in the selection model memory 48.

As shown in FIG. 19, this selection model memory can store information on a plurality of planes similar to the bit plane memory 13. In the example of this figure, 2
A state of determining a reference pixel model for encoding the encoding target image signal 21 of the value bit plane j is shown. The binary bit plane j and the binary bit plane j-1 are two planes in the same frame of the multi-valued image. The binary bit plane i is the bit plane of the previous frame.

The numbers 1 to 4 in this table indicate the numbers of the four types of reference pixel models registered in the reference pixel model table 46. The model with the number 1 is a model in which the correlation between pixels in the same plane is high, and as the numbers increase to 2, 3, and 4, the correlation between pixels becomes smaller. Here, the binary bit plane j
Focusing on the pixel 21 to be coded, the reference model having a small correlation is selected for the pixels around the pixel 21 for which coding has been completed. Therefore, it can be inferred that it is appropriate to select a model having a small correlation for the target pixel this time. It is also possible to perform such model selection by using the correlation between planes or by using the information of the upper plane. The reference pixel model determination unit 45 shown in FIG. 18 determines the reference pixel model with reference to the contents of the selection model memory 48. Therefore, it is possible to select a more appropriate reference pixel model.

The coding condition determining unit 15 of FIG. 18 uses the reference pixel model thus determined to generate a coding parameter 15A and sends it to the coding unit 16. The encoding unit 16 encodes the symbol input from the buffer 14 based on this encoding parameter. Furthermore, the encoding unit 16
In parallel with the encoding process of No. 1, the minimum code amount estimation unit 47 determines that the reference pixel selected from the bit plane memory 13 is 0 or 1 for each of the plurality of reference pixel models registered in the reference pixel model table 46. Based on the coded pixel symbol received from the buffer 14, the model which minimizes the code amount of the symbol is estimated. The estimated reference pixel model number is newly registered in the selected model memory 48, and is used to determine the subsequent reference pixel model.

As described above, if a plurality of fixed reference pixel models are prepared in advance, and the models are adaptively selected according to the characteristics of the image and switched to encode, a local change of the image can be dealt with. Thus, efficient encoding can be performed. In addition, it is possible to improve the coding efficiency of a multi-valued image by switching the model using information of the upper plane and other color planes. Further, by providing the minimum code amount estimation unit 47 and the selection model memory 48, more appropriate model selection can be performed by utilizing the past model selection information, and the coding efficiency is improved.

[Plurality of Encoding Units] FIG. 20 shows a block diagram of an embodiment of the image encoding device of the present invention. In this embodiment, a plurality of coding units 53-1, 53-2, 53-3, 5 are provided.
3-4 ... are provided. The switching unit 52 selects which of these coding units is to be operated. The coding condition determination unit 51 is a circuit for outputting to the switching unit 52 information on which coding unit to select. Each of the above coding units 53-1 and 53-
In order to perform the encoding according to the local feature of the image, 2 prepares the reference pixel model describing the reference pixel position, and executes the encoding using the different models.

Even if different reference pixel models are used,
It is preferable that the encoding can always be executed by one encoding unit. In this case, for example, if the function of the present invention is added to the conventional device as shown in FIG. 2, the encoding unit will be completely modified. However, it may be more economically advantageous to add a new encoding unit having a function for executing the method of the present invention, and add a switching unit to operate the encoding unit. This embodiment can handle such a case.

The device of FIG. 20 operates as follows. Since the optimum reference pixel model differs depending on the local characteristics of the image, the optimum ones are switched and used for the encoding units 53-1, 53-2, .... First, when the first bit plane of a certain frame image is encoded, since there is no plane that has already been encoded, the same constant reference pixel model as in the past is used without using the correlation between bit planes. Do. From the bit plane memory 13, raster scanning is sequentially performed from the upper left pixel of the first bit plane, and the binary image signal sequence 13B is supplied to the switching unit 52. When the coding plane is the first bit plane, the coding condition determining unit 51 does not need to correlate with the bit plane that has been coded.
A selection signal 51A is output to the switching unit 52 so as to select the encoding unit 53-1. The encoding unit 53-1 executes encoding using the reference pixel model as shown in FIG. 17A that utilizes the correlation between neighboring pixels, which is a feature of the first bit plane.

Next, for the second and subsequent bit planes, the information of the first bit plane which has already been encoded can be used. Therefore, the coding model determination unit 51 instructs the switching unit 52 to select the coding unit 53-2. As a result, this time, the encoding unit 53-2 executes the encoding using the reference pixel model as shown in FIG. 17B in consideration of the correlation with the first bit plane.

FIG. 21 is an explanatory view showing an example of still another reference pixel model. This (a) is the reference pixel model of the encoding unit 53-1 shown in FIG. 20, (b) is the reference pixel model of the encoding unit 53-2, and (c) is the encoding unit 53-.
3 reference pixel model. As shown in (a), when the encoding unit 53-1 encodes the encoding target image signal 21 of the first bit plane, the reference image signal 2 around it is encoded.
Use only 2. On the other hand, in the encoding unit 53-2, (b)
As shown in (3), there is a correlation with the surrounding pixels, and the coding is performed using the correlation between the planes. In the example of (c), pixels having a large correlation between planes are encoded. A portion having a change of "0" or "1" in a pixel causes deterioration of encoding efficiency when encoded by the model of (b), for example. Further, a pixel in which “0” or “1” changes in the second bit plane is often a pixel in which “0” or “1” changes in the first bit plane, and therefore, as shown in (c). Coding efficiency can be improved by using a model having a high correlation between planes.

When encoding the second bit plane,
The coding model determination unit 51 determines whether the first pixel of the same position of the coding pixel
It is determined whether the pixel of the bit plane is a black-and-white change point, and if it is not the change point, the model shown in FIG. 21B is selected, and the command for selecting the coding unit 53-2 is switched to the switching unit 52.
Output to. If it is a change point, the model of FIG. 21C is selected, and a signal for selecting the encoding unit 53-3 is output to the switching unit 52.

The reference pixel model shown in FIG.
This is a model that refers to 0 pixel. The total number of combinations of “0” and “1” is 1024, and this number of coding states can be considered. Therefore, the appearance probabilities of "0" and "1" are given for each state. Each coding unit entropy-codes the coded pixel using the appearance probability in the state determined by such a model. And
The appearance probability is updated depending on whether the coded pixel is “0” or “1”. Similarly, regarding the encoding of the third bit plane and thereafter, the encoding efficiency can be improved by adaptively switching the encoding units by using the information of the plurality of upper planes and performing the encoding.

[Reference of image signal of another plane] Regarding the selection of the reference pixel model using the information of the upper plane,
Various methods are possible. In the case of the second bit plane, the upper plane is only the first bit plane,
A plurality of upper planes exist after the third bit plane. Therefore, as in the case of the second bit plane, there is a case where only the information of the one upper plane is used, and a case where the information of a plurality of upper planes is used. For example, in the case of using only the information of the one upper plane, the models may be switched by using the change between “0” and “1” of the adjacent pixels of the upper plane. In addition, when using information of a plurality of upper planes, model selection is performed by making a distinction between the case where adjacent pixels of all the upper planes are the same and the case where even adjacent pixels of at least one of the upper planes are different. Do.

An encoding method of an image composed of a plurality of color components such as a color image is executed as follows. For example, a full-color image in which each of the three primary colors of RBG is represented by 256 gradations is converted into a binary bit plane to be an image of 24 planes. in this case,
The order of conversion and encoding is R1, G1, B1, R
2, G2, B2, R3, G3, B3 ... R8, G8, B8
And so on. Note that R1 means the first bit plane of the red component. In this way, R1 becomes G1
Can be considered as the upper plane of. In the case of a color image, the correlation between the RGB components is large, and features such as the edge portion of the image are likely to appear at the same pixel position for all components. Therefore, if the model is adaptively selected by using the information of the upper plane and the encoding unit is switched to perform the encoding, the encoding efficiency of the color image can be improved.

The present invention is not limited to the above embodiments. With respect to a moving image, a multi-valued still image, a binary image, and various other images, the apparatus of the present invention can achieve high-efficiency encoding with the same configuration. The memory, the frame image generation unit, the bit plane image generation unit, the coding condition determination unit, and the like used for the encoding may be configured by combining circuits of various conventionally known configurations. None of these may be configured by an integrated computer or the like.

[0058]

According to the image coding and decoding apparatus of the present invention described above, a moving image signal sequence is converted into a plurality of frame images, and this is converted into a binary bit plane image, and an arbitrary binary value is converted. Since encoding can be performed with reference to an image signal, high-efficiency encoding can be performed for moving images, multi-valued still images, and binary images by the same encoding device. In addition, since the encoding condition is estimated by the same method on the encoding side and the decoding side and the encoding / decoding condition is set using the image signal that has already been encoded / decoded, the encoding unit The signal can be decoded without accepting conditions and the like from the encoding unit. Further, by detecting the motion of the moving image and shifting the position of the reference image signal in the moving direction of the moving image, highly efficient coding of the moving image can be achieved.

Further, according to the required received image quality, mode setting such as the number of images of a converted frame per constant time or a partial omission of a binary bit plane image is performed to enable coding according to a transmission path. Become. If a binary image signal corresponding to the upper bits in the binary bit plane image that has already been encoded or a binary image signal corresponding to different color components in the color image is used for selecting the reference pixel model, The efficiency of conversion can be increased. Further, by providing a plurality of encoding units having different encoding conditions and selecting these to execute the encoding processing, it is possible to achieve high-efficiency encoding while reducing costs such as the design of the encoding unit. .

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of an image encoding device of the present invention.

FIG. 2 is a block diagram of a conventional image encoding device.

FIG. 3 is an explanatory diagram of a bit plane image.

FIG. 4 is an explanatory diagram of a frame image.

FIG. 5 is an explanatory diagram of a bit plane memory configuration.

FIG. 6 is a reference image signal explanatory diagram (No. 1).

FIG. 7 is a reference image signal explanatory diagram (2).

FIG. 8 is a reference image signal explanatory diagram (3).

FIG. 9 is a block diagram showing an embodiment of an image encoding device of the present invention.

FIG. 10 is an explanatory diagram of an encoding / decoding condition determining method.

FIG. 11 is a block diagram of an embodiment of an image encoding device of the present invention.

FIG. 12 is an explanatory diagram of motion information and a reference image signal.

FIG. 13 is a block diagram of an embodiment of an encoding device of the present invention.

FIG. 14 is an explanatory diagram of a mode in which the number of converted frame images is reduced.

FIG. 15 is an explanatory diagram of modes excluding the lower bit side.

FIG. 16 is a block diagram of an embodiment of an image encoding device according to the present invention.

FIG. 17 is an explanatory diagram of a reference pixel model.

FIG. 18 is a block diagram of an embodiment of an image encoding device of the present invention.

FIG. 19 is an explanatory diagram of a selection model memory.

FIG. 20 is a block diagram of an embodiment of an image encoding device of the present invention.

FIG. 21 is an explanatory diagram showing an example of a reference pixel model.

[Explanation of symbols]

10A video signal sequence 10B still image signal sequence 10C binary image signal series 11 Frame image generator 12-bit plane image generator 13-bit plane memory 14 buffers 15 Encoding condition determination unit 16 Encoding section 16A coded sequence

Claims (10)

(57) [Claims] 1. A frame image generation unit for converting a moving image signal sequence into a plurality of frame images at consecutive points on a time axis, and corresponding bits of a digital multi-valued image signal forming each frame image. A bit plane image generation unit that converts the binary bit plane images into a plurality of binary bit plane images, a bit plane memory that stores the plurality of binary bit plane images, and a binary image of the binary bit plane images. A buffer for converting into a signal sequence and outputting the signal sequence; an encoding condition determining unit for determining an encoding condition with reference to a binary bit plane image obtained by converting at least the previous frame image in the bit plane memory; 2 output from the buffer using the encoding condition
An image encoding device, comprising: an encoding unit that encodes and outputs a value image signal sequence. 2. A bit plane memory for storing a decoded binary bit plane image of a previous frame image, and a decoding for determining a decoding condition by referring to the binary bit plane image in the bit plane memory. A condition determining unit, a decoding unit that receives the encoded binary image signal sequence and decodes the encoded binary image signal sequence using the decoding condition, and a plurality of binary bits obtained from the decoded binary image signal sequence A buffer for storing a plane image, a bit plane image integration unit for reproducing a frame image composed of a digital multi-valued image signal from the plurality of binary bit plane images, and a plurality of consecutively arranged on the time axis. And a frame image integration unit that obtains a moving image signal sequence from the frame image. 3. A frame image generation unit for converting a moving image signal sequence into a plurality of frame images at consecutive points on the time axis, and corresponding bits of a digital multi-valued image signal forming each frame image. A bit plane image generation unit that converts the binary bit plane images into a plurality of binary bit plane images, a bit plane memory that stores the plurality of binary bit plane images, and a binary image of the binary bit plane images. A buffer for converting into a signal sequence and outputting it, and 2 obtained by converting the binary bit plane image and the previous frame image corresponding to the upper bits in the bit plane memory.
A coding condition determining unit that determines a coding condition by referring to the value bit plane image; and 2 that is output from the buffer using the coding condition.
An image encoding device, comprising: an encoding unit that encodes and outputs a value image signal sequence. 4. A motion detection unit for comparing frame images continuously arranged on a time axis with each other to detect a motion of a moving image, and a reference set in a previous frame according to the motion of the moving image. The image coding apparatus according to claim 1, further comprising a coding condition determining unit that determines a coding condition so as to shift a reference position of a pixel in a direction opposite to a direction of the motion of the moving image. 5. The image coding apparatus according to claim 1, further comprising a mode setting unit that sets the number of converted frame images per fixed time in accordance with a required received image quality. 6. The mode setting unit according to claim 1, further comprising: a mode setting unit that excludes an image on a lower bit side of the binary bit plane image from an encoding target according to a required received image quality. Image encoding device. 7. A minimum code amount estimator that performs an operation using the same operation processing method on the encoding side and the decoding side and estimates the optimum encoding condition, and a binary image that follows the optimum encoding condition. The image coding apparatus according to claim 1, further comprising a coding condition determination unit that applies a coding condition of a signal. 8. A plurality of reference pixel models for performing encoding by referring to any one of binary bit plane images that have already been encoded, and selecting an optimum reference pixel model from them. The image coding apparatus according to claim 1, further comprising a coding condition determining unit that determines an optimum coding condition. 9. The image coding apparatus according to claim 8, wherein the binary image signal corresponding to the high-order bit in the already coded binary bit plane image is used for selecting the reference pixel model. . 10. The image coding according to claim 8, wherein binary image signals corresponding to different color components in the already coded binary bit plane image are used for selection of the reference pixel model. apparatus.