JP2011171938A - Method and device for encoding image, method and device for decoding image, solid-state imaging element, digital signal processing element, digital still camera, and monitoring camera - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image encoding method by an adaptive arithmetic encoding where encoding efficiency is further improved. <P>SOLUTION: The image encoding method comprises: a predictive pixel generating step S102 which generates the predicted value of a pixel to be encoded; a change extracting step S103 which generates predicted error information from pixel data and the predicted value of the pixel to be encoded; an arithmetic encoding step S105 which performs arithmetic encoding of the predicted error information bit by bit using predicted probability changing adaptively; and a predicted probability update step 106 (S201 to S203) which updates the predicted probability to decrease the predicted probability to be used for the arithmetic encoding of a first value when the bit subjected to the arithmetic encoding is a first value and so as to increase the predicted probability to be used for the arithmetic encoding of the first value when the bit subjected to the arithmetic encoding is not the first value. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image encoding / decoding method and an apparatus therefor, and in particular, to increase the speed of data transfer by image compression and reduce the amount of memory used in an apparatus that handles images such as a digital still camera and a network camera. The present invention relates to an image encoding / decoding method and apparatus therefor.

  2. Description of the Related Art In recent years, the amount of image data processed by an integrated circuit mounted on an imaging apparatus has increased with the increase in the number of pixels of an imaging element used in an imaging apparatus such as a digital still camera or a digital video camera. In order to handle a large amount of image data, it is possible to increase the operating frequency and increase the capacity of the memory in order to secure the bus width of the data transfer in the integrated circuit, but this directly leads to an increase in cost. .

  Thus, in recent years, various processing techniques for reducing the amount of image data by removing redundant signal components included in an image signal have been proposed. Note that such processing for reducing the amount of image data is called high-efficiency encoding of an image, or simply image encoding. In addition, this technique is sometimes called data compression coding, redundancy suppression coding, bit rate reduction coding, or the like.

  One of the most typical coding methods is arithmetic coding. This assigns a short code to an output level with a high occurrence frequency (that is, appearance frequency) and a long code to an output level with a low occurrence frequency. As a result, the average code length of the output code can be shortened.

  Hereinafter, arithmetic coding (more specifically, arithmetic coding called a range coder) will be described.

  In arithmetic encoding, the data to be encoded is replaced with the occurrence frequency (specifically, the occurrence probability) by dividing the number line 0 to 1 by the occurrence frequency of the data (event) to be encoded. For example, when the input value “11001...” Is encoded as shown in FIG. 23, the prediction probability P (0 ≦ P ≦ 1) where “1” appears and the prediction probability where “0” appears when each bit is sequentially encoded. By replacing the occurrence frequency with Q (Q = 1−P), this is the division ratio on the number line specified by the origin C and the width A shown in FIG. 24 (that is, the section on the number line). Represented by Specifically, two variables C (origin) and A (width) are provided on the number line, and C = 0 and A = 1 are initialized as shown in FIG. Next, the first bit is inspected, and when the bit is a value “1”, the number line is divided without changing the origin C, and the prediction probability P newly indicated by the division (1) in FIG. The width A of the number line is updated to the size of. Specifically, when the code “1” corresponding to the prediction probability P comes, the width A of the number line is changed to the prediction probability P (the section of the prediction probability P indicated by the division (1)), and the origin C is Do not change as is. Subsequently, when the code “0” corresponding to the prediction probability Q comes, the position of the origin C in FIG. 24 is moved to the right by the prediction probability P (the section of the prediction probability P indicated by the division (2)) to increase the width A. Change to the section of the prediction probability Q in the division (2).

  According to the above-described method, the bits constituting the input value “11001...” Shown in FIG. 23 are continuously encoded (that is, the position of the origin C is left as it is, or the width A is narrowed or the origin C is moved. When the process of narrowing the width A) is continued, the origin C converges to a point on a number line of 0 to 1. Thereby, the original input value can be expressed by the converged position of the origin C (or the position C and the width A of the origin). Such coding is called arithmetic coding.

  Therefore, when the code “1” corresponding to the prediction probability P comes, the number line is divided, but the position of the origin C does not change. Therefore, as long as the code “1” corresponding to the prediction probability P continues, The position of the origin C does not change only by dividing the number line (that is, the width A changes). That is, as the number of codes (bit values) corresponding to the prediction probability P increases, the number of arithmetic codes necessary for expressing the origin C does not increase even if the number line division proceeds (the prediction probability Q is increased). The position of the origin moves only when the corresponding code “0” comes).

  On the other hand, the arithmetic code decoding process is as shown in FIG. First, as in the case of encoding, a number line from 0 to 1 is prepared. Now, it is assumed that a value to be decoded (encoded value “encoded value”) is a position indicated by a one-dot chain line in FIG. If the encoded value is within the range of the prediction probability P on the number line, it is decoded as “1”. Subsequently, as in the encoding, the width A (here, the section of the prediction probability P in the first number line in FIG. 25) is further changed to the prediction probability P and the prediction probability Q (the prediction in the second number line in FIG. 25). It is determined whether the code to be decoded is “0” or “1” depending on which range the encoded value belongs to. Here, since the encoded value belongs to the interval of the prediction probability P in the second number line in FIG. 25, it is decoded as “1”. By repeating this, decoding is performed.

  Here, the arithmetic coding described above has the following drawbacks. That is, it is necessary to know the occurrence frequency in advance in order to obtain the prediction probability P at which “1” appears and the prediction probability Q at which “0” appears. For this reason, it is necessary to read all data to be encoded before performing arithmetic encoding.

  There is adaptive arithmetic coding as a solution to such a drawback of arithmetic coding. This is the same as the arithmetic coding described above in that it is divided by the prediction probability of the data to be encoded on the number line from 0 to 1, but the prediction probabilities P and Q are read in advance and all the data are read in advance. Is not calculated. Specifically, in adaptive arithmetic coding, since the occurrence frequency is unknown at first, the prediction probabilities P and Q are set to 0.5 temporarily, and this prediction probability is adaptively set as the code is input. Change.

  For example, assuming that the prediction probability P = 0.5, if there are many “1” s in the incoming code, the prediction probability P is changed in a direction larger than 0.5. Conversely, when “0” increases, the prediction probability is lowered, and the prediction probability P is made smaller than 0.5 (P <0.5). Thus, by changing the prediction probability adaptively, it is not necessary to determine the prediction probability in advance by reading all data in advance.

  Various techniques have been proposed as techniques for compressing image data using such adaptive arithmetic coding (see, for example, Patent Document 1). The technique of Patent Document 1 uses the above-described conventional technique to apply image data compression processing to pixel signals (RAW data) input from an image sensor, thereby increasing the number of pixels in the image sensor. The purpose is to reduce the bus bandwidth required for writing to and reading from the memory and to enable high-speed operation even when the processing load increases. In addition, by extracting the prediction error between the compression target pixel value and the prediction value by exclusive OR, and using this as the compression target information, the variable length using adaptive arithmetic coding to increase the compression efficiency Encoding method is adopted. For the implementation method, the prediction error is calculated based on the prediction value predicted from the values of the encoding target pixel and the surrounding pixels, and the arithmetic code division ratio is adapted by changing this prediction error to a predetermined variable. In this case, encoding is performed using a fixed division ratio without controlling the division ratio in arithmetic coding when the surrounding pixels have a specific pattern. Yes.

Japanese Patent No. 3127513

  However, such a conventional coding method using adaptive arithmetic coding does not consider the occurrence frequency bias inherent in image data, and therefore does not perform efficient coding and compression. There is a problem of inefficiency.

  Here, since the occurrence frequency inherent in the image data is highly correlated with the pixel to be encoded and its neighboring pixels, the prediction error that is the difference between them is the MSB (Most Significant Bit). ) “0” occurs more frequently on the side, and “1” occurs on the LSB (Least Significant Bit) side. Since the above conventional technique does not consider this, when encoding in the direction from MSB to LSB, since “0” is frequently input at the start of encoding, as encoding is repeated. The frequency of occurrence of “0” becomes higher. However, as the encoding process proceeds, the frequency of occurrence of “1” increases as it goes to the LSB side, resulting in a difference between the updated frequency of occurrence and the frequency of occurrence of data to be encoded. As a result, when "1" is encoded, a long code amount is output, and there is a problem that the code amount finally output increases.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide an image coding method or the like by adaptive arithmetic coding with further improved coding efficiency.

  In order to solve the above problem, one aspect of an image encoding method according to the present invention is an image encoding method for compressing pixel data constituting image data, and is at least one positioned around a pixel to be encoded. A prediction pixel generation step of generating a prediction value of the encoding target pixel from pixel data of the pixels, and extracting the bit change between the pixel data of the encoding target pixel and the prediction value. A change extraction step for generating prediction error information, and a prediction error information generated by the change extraction step is arithmetically encoded using a prediction probability that adaptively changes for each bit constituting the prediction error information. An arithmetic coding step, and a prediction probability used for arithmetic coding of the first value when the bit coded in the arithmetic coding step is the first value A prediction probability update step of updating the prediction probability so as to increase the prediction probability used for arithmetic encoding of the first value when the bit encoded in the arithmetic encoding step is not the first value In the arithmetic encoding step, the arithmetic encoding is performed using the prediction probability updated in the prediction probability update step for each bit constituting the prediction error information.

  As a result, in the arithmetic coding for each bit using the prediction probability, the prediction probability is updated in the direction of decreasing the prediction probability for the event that has occurred (the value of the bit that is the target of arithmetic coding). Therefore, the prediction error information subject to arithmetic coding goes from the MSB to the LSB, so that the state inversion such as the transition from the high occurrence frequency of “0” to the high occurrence frequency of “1” is supported. The prediction probability is updated, and accordingly, the prediction probability is adaptively updated in consideration of the deviation of the occurrence frequency of bits in the image data, and the coding efficiency of the image data is improved.

  The present invention can be realized not only as such an image encoding method, but also as an image decoding method corresponding to the image encoding method, or an LSI or the like provided with steps constituting those methods as an electronic circuit. It can also be realized as a device, realized as a program for causing a computer to execute the steps constituting those methods, or realized as a recording medium such as a computer-readable CD-ROM on which the program is recorded.

  According to the present invention, adaptive arithmetic coding is performed in consideration of occurrence frequency deviation in image data, so that the coding efficiency of image data is improved.

  Therefore, according to the present invention, image data obtained by a digital camera or the like is compressed to a smaller size and stored in a memory. Therefore, the practical significance of the present invention in particular where digital cameras have become widespread is extremely high.

FIG. 1 is a block diagram showing a configuration of an image encoding device and an image decoding device according to Embodiment 1 of the present invention. (A) And (b) is a flowchart which shows the image coding process and decoding process in an image coding apparatus and an image decoding apparatus, respectively. A flowchart showing a method for updating a prediction probability in arithmetic coding The figure explaining the prediction formula in a prediction pixel production | generation part The figure explaining the input order of the prediction error information input into an arithmetic coding part The figure explaining the input order of the prediction error information input into an arithmetic coding part The figure explaining the update of the prediction probability used for arithmetic coding Diagram showing input value (specific example) for arithmetic coding (A) And (b) is a figure which shows the input order (specific example) of the prediction error information arithmetically encoded in Embodiment 1. The figure which shows the update (specific example) of the prediction probability used for arithmetic coding in Embodiment 1 The figure which shows an encoding process result (specific example) The figure which shows the numerical formula for explaining arithmetic coding The figure for demonstrating the specific example of the arithmetic coding in this Embodiment A diagram for explaining a specific example of arithmetic coding in the conventional method The figure which shows an encoding process result (specific example) (A) And (b) is the flowchart in the fixed-length encoding of the image coding apparatus and the image decoding apparatus in Embodiment 2 of this invention, respectively. (A) And (b) is a figure which shows the input order of the prediction error information arithmetically encoded in Embodiment 2. (A)-(d) is a figure which shows the input order of the prediction error information arithmetically encoded in Embodiment 2. Block diagram showing a configuration of a digital still camera according to Embodiment 3 of the present invention Block diagram showing the configuration of a digital still camera according to Embodiment 4 of the present invention. FIG. 6 is a block diagram illustrating a configuration of a surveillance camera according to a fifth embodiment. FIG. 7 is a block diagram illustrating a configuration of a surveillance camera according to a sixth embodiment. The figure which shows the input value and prediction probability for demonstrating the conventional arithmetic coding Diagram for explaining conventional arithmetic coding Diagram for explaining conventional arithmetic decoding The figure which shows the update of the prediction probability used for arithmetic coding in the conventional method

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description of each embodiment and each modification, components having the same functions as those described once will be given the same reference numerals and description thereof will be omitted.

<< Embodiment 1 >>
FIG. 1 is a block diagram showing the configuration of image encoding device 300 and image decoding device 310 according to Embodiment 1 of the present invention. FIG. 2A and FIG. 2B are flowcharts showing the procedures of the image encoding method and the image decoding method, respectively, according to Embodiment 1 of the present invention. FIG. 3 shows the update of prediction probabilities used in the arithmetic coding and arithmetic decoding shown in FIGS. 2A and 2B (step S106 in FIG. 2A and S115 in FIG. 2B). It is a flowchart which shows a detail.

  As shown in FIG. 1, the image encoding device 300 is a device that compresses pixel data constituting image data in units of M pixels (hereinafter also referred to as “M pixels”), and is a processing target pixel. A value input unit 301, a change extraction unit 302, a prediction pixel generation unit 303, an arithmetic coding unit 304, a prediction probability update unit 305, and a packing unit 306 are provided.

  The processing target pixel value input unit 301 receives pixel data of the encoding target pixel. If the received pixel data is initial pixel value data, the processing target pixel value input unit 301 outputs the initial pixel value data to the packing unit 306 and receives the received pixel data. When the data is not initial pixel value data, the processing unit outputs the pixel data to the change extraction unit 302 and the predicted pixel generation unit 303. Note that the initial pixel value data is pixel data of the first pixel in M pixels that is a unit of compression (that is, arithmetic coding).

  The prediction pixel generation unit 303 generates a prediction value of the encoding target pixel from pixel data of at least one pixel located around the encoding target pixel according to a predetermined prediction procedure, and the change extraction unit 302 It is a processing part which outputs to.

  The change extraction unit 302 calculates a bit change between the pixel data of the encoding target pixel sent from the processing target pixel value input unit 301 and the prediction value of the encoding target pixel sent from the prediction pixel generation unit 303. This is a processing unit that generates prediction error information indicating the bit change by extraction and outputs the prediction error information to the arithmetic coding unit 304.

  The arithmetic coding unit 304 is a processing unit that performs arithmetic coding on the prediction error information sent from the change extraction unit using a prediction probability that adaptively changes for each bit constituting the prediction error information. In the arithmetic coding, arithmetic coding is performed using the prediction probability updated by the prediction probability update unit 305 for each bit constituting the prediction error information.

  The prediction probability update unit 305 determines in advance a prediction probability P used for arithmetic coding of the first value when the bit encoded by the arithmetic coding unit 304 is the first value (“1”). Is reduced by the specified α (that is, the prediction probability Q used for the arithmetic encoding of the second value (“0”) is increased by a predetermined α), while being encoded by the arithmetic encoding unit 304 If the bit is not the first value, the prediction probability used for arithmetic coding of the first value is increased (that is, the prediction probability Q used for arithmetic coding of the second value (“0”) is set in advance. It is a processing unit that updates the prediction probabilities P and Q so as to reduce the value by a predetermined α).

  That is, the prediction probability update unit 305 updates the prediction probability in a direction to lower the prediction probability for the event that occurred (the value of the bit that is the target of arithmetic coding) in the arithmetic coding for each bit using the prediction probability. To go. This corresponds to the inversion of the state such as the transition from the high occurrence frequency of “0” to the high occurrence frequency of “1” because the prediction error information to be subjected to arithmetic coding goes from MSB to LSB. This is to update the predicted probability.

  The packing unit 306 receives at least one pixel or more initial pixel value data sent from the processing target pixel value input unit 301 and at least one bit or more arithmetic code sent from the arithmetic coding unit 304. It is an output unit that performs packing that combines them and outputs the encoded data to the outside of the image encoding device 300.

  On the other hand, as shown in FIG. 1, the image decoding device 310 is a device that decodes pixel data from the arithmetic code obtained by the image encoding device 300, and includes an unpacking unit 311 and an arithmetic decoding unit 312. , A prediction probability update unit 313, a change extraction unit 314, a prediction pixel generation unit 315, and an output unit 316.

  The unpacking unit 311 receives encoded data to be subjected to arithmetic decoding from the outside, and outputs the initial pixel value data to the output unit 316 when the received encoded data is initial pixel value data. When the received data is not initial pixel value data (it is an arithmetic code), the processing unit outputs the data to the arithmetic decoding unit 312.

  The arithmetic decoding unit 312 predicts the arithmetic code sent from the unpacking unit 311 by performing arithmetic decoding using the adaptively changing prediction probability for each bit constituting the arithmetic code. A processing unit that generates error information and outputs the error information to the change extraction unit 314, the prediction probability update unit 313, and the prediction pixel generation unit 315. In the arithmetic decoding, for each bit constituting the arithmetic code, the prediction probability is updated. Arithmetic decoding is performed using the prediction probability updated by the unit 313.

  The prediction probability update unit 313 has the same function as the prediction probability update unit 305 included in the image encoding device 300, and the bit decoded by the arithmetic decoding unit 312 is the first value (“1”). In this case, the prediction probability P used for arithmetic decoding of the first value is lowered by a predetermined α (that is, the prediction probability Q used for arithmetic decoding of the second value (“0”) is set in advance. On the other hand, when the bit decoded by the arithmetic decoding unit 312 is not the first value, the prediction probability used for arithmetic decoding of the first value is increased (that is, the first This is a processing unit that updates the prediction probabilities P and Q so that the prediction probability Q used for arithmetic decoding of the value of 2 (“0”) is reduced by a predetermined α).

  The predicted pixel generation unit 315 uses the initial pixel value data sent from the output unit 316 or the already decoded pixel value data, with the same processing logic as the predicted pixel generation unit 303 of the image encoding device 300. This is a processing unit that generates a predicted value of a pixel that is a processing target in the change extraction unit 314 (that is, a decoding target pixel).

  The change extraction unit 314 performs a predetermined operation (here, exclusive OR) on the prediction error information generated by the arithmetic decoding unit 312 and the prediction value generated by the prediction pixel generation unit 315. Thus, the processing unit generates pixel data of the decoding target pixel.

  The output unit 316 outputs the initial pixel value data sent from the unpacking unit 311 and the pixel data restored by the change extraction unit 314 to the external and predicted pixel generation unit 315 as decoded pixel data. It is a processing unit.

  Note that all or part of the image encoding device 300 and the image decoding device 310 in the present embodiment may be realized by hardware such as an LSI (Large Scale Integration), a CPU (Central Processing Unit), or the like. It may be realized by a program executed by. The same applies to the following embodiments.

  First, a process for encoding an image (hereinafter referred to as an image encoding process) performed by the image encoding apparatus 300 according to the present invention will be described with reference to FIGS. FIG. 2A is a flowchart of the image encoding process. The M pixel to be encoded is input to the processing target pixel value input unit 301. In the present embodiment, each pixel data is digital data having an N-bit length. In addition, the packing unit 306 packs the initial pixel value data of at least one pixel and the arithmetic code corresponding to the plurality of pixel data, so that the encoded data (packing data) corresponding to the M pixel is used as an image. The data is output from the encoding device 300. Here, the natural numbers M and N are determined in advance. The pixel data input to the processing target pixel value input unit 301 is output to the prediction pixel generation unit 303 at an appropriate timing. However, when the encoding target pixel of interest is input as initial pixel value data (FIG. 2A: YES in step S101), it is input directly to the packing unit 306.

  When the encoding target pixel of interest is not initial pixel data (FIG. 2A: NO in step S101), the pixel data is input to the predicted pixel generation unit 303. The pixel data input to the prediction pixel generation unit 303 is the initial pixel value data input before the target encoding target pixel, the previous encoding target pixel, or the previous encoding target pixel. , One of the decoded pixel data. The predicted pixel generation unit 303 generates a predicted value of the focused pixel data using the input pixel data (FIG. 2A: step S102).

  Here, this image coding apparatus 300 employs predictive coding as a coding method for pixel data. Predictive coding is a method of generating a prediction value for a pixel to be encoded and encoding a difference value between the pixel to be encoded and the prediction value. As for the predicted value, in the case of pixel data, the value of the target pixel to be encoded is predicted from the neighboring pixel data based on the fact that the values of adjacent pixels are the same or are likely to be close. By doing so, the amount of change is kept as small as possible. FIG. 4 is an explanatory diagram showing an arrangement of adjacent pixels used for calculation of a predicted value. In the figure, “x” indicates the pixel value of the target pixel. “A”, “b”, and “c” are pixel values of adjacent pixels for obtaining the predicted value “y” of the target pixel. The prediction formulas (1) to (7) that are generally used for calculating the predicted value are shown below.

y = a (1)
y = b (2)
y = c (3)
y = a + b-c (4)
y = a + (bc) / 2 (5)
y = b + (ac) / 2 (6)
y = (a + b) / 2 (7)

  The image encoding device 300 thus obtains the predicted value “y” of the target pixel using the pixel values “a”, “b”, and “c” of the neighboring pixels of the target pixel, and this predicted value “y”. And a prediction error Δ (= y−x) between the encoding target pixel “x” and the prediction error Δ are encoded. For this purpose, the prediction pixel generation unit 303 calculates a prediction value using any one of the prediction expressions (1) to (7) used in the prediction encoding described above. Note that not only the prediction formulas (1) to (7) described above, but if an internal memory buffer can be secured in the compression process, peripheral pixels other than the pixel adjacent to the pixel of interest are also held in the memory buffer. It is also possible to improve prediction accuracy by using it for prediction. In the present embodiment, the prediction formula (1) is used in step S102.

  The change extraction unit 302 performs an exclusive OR operation on the pixel data of the encoding target pixel expressed in N bits and the prediction value, calculates prediction error information E having an N-bit length, and the arithmetic encoding unit 304 And it outputs to the prediction probability update part 305 (Fig.2 (a): step S103). Here, the calculation of the prediction error information E is repeated for M pixels (FIG. 2A: step S104).

The arithmetic encoding unit 304 performs arithmetic encoding on the input prediction error information E for M pixels (FIG. 2A: step S105). Here, an example of the processing order in the arithmetic coding is shown in FIGS. FIG. 5 shows a case in which encoding processing is sequentially performed from the MSB to the LSB so that all the digits of the prediction error information E are scanned across the prediction error information E for M pixels. The processing order is shown. That is, in the order shown in FIG. 5, _first , the most significant bit of the prediction error information E ₁ is encoded, then the most significant bit of the prediction error information E ₂ is encoded, and further, the most significant bit of the prediction error information E ₃ is encoded. When the most significant bits of all the prediction error information are encoded as shown in FIG. 5, the second bit from the most significant bit of the prediction error information E ₁ is encoded, and the most significant bit of the prediction error information E _{2 is} further encoded. The encoding process is performed with priority on the MSB, such as encoding the second bit from the top. On the other hand, FIG. 6 shows a processing order in a case where the encoding process is sequentially performed in accordance with the input of the prediction error information E (for each prediction error information E). That is, in the order shown in FIG. 6, the encoding process is performed in the order of MSB to LSB of the prediction error information of the pixel of interest, and the encoding is subsequently performed in the order of MSB to LSB of the prediction error information of another pixel of interest. We will continue the process.

  The prediction probability update unit 305 determines the prediction error information E input from the change extraction unit 302 and updates the prediction probability (FIG. 2A: step S106). Specifically, when the input information is “1” (FIG. 3: YES in step S201), the prediction probability update unit 305 increases the prediction probability Q of “0” by α, The prediction probability is updated by lowering the prediction probability P of "" by α (FIG. 3: step S202). When the input information is “0” (FIG. 3: NO in step S201), the prediction probability Q of “0” is decreased by α, while the prediction probability P of “1” is increased by α. The probability is updated (FIG. 3: step S203).

  Such update of the prediction probability will be described in detail with reference to FIG. The frequency of occurrence of “1” is P (real number of 0 ≦ P ≦ 1), the frequency of occurrence of “0” is Q (= 1−P), and the reference value before the start of encoding is P = Pbase ( 0 <Pbase <1 real number), Q = 1−P. When “1” is encoded (FIG. 3: YES in step S201), the prediction probability P is subtracted by α (a real number where 0 <α <1) so as to lower the prediction probability P of “1” (FIG. 3). 3: Step S202). On the other hand, when “0” is encoded (FIG. 3: NO in step S201), the prediction probability P is added by α (a real number where 0 <α <1) so as to increase the prediction probability P of “1”. (FIG. 3: Step S203). When the updated prediction probability P reaches Pmax which is the maximum value of P (0 ≦ Pmax ≦ 1 and Pbase <Pmax), the prediction probability P is updated to Pmax which is the maximum value. Further, when Pmin which is the minimum value of the prediction probability P (0 ≦ Pmin ≦ 1 and Pbase> Pmin) is reached, the prediction probability P is updated to Pmin which is the minimum value.

  The packing unit 306 performs packing by combining initial pixel value data of at least one pixel and an arithmetic code of at least one bit, and the encoded data after packing (that is, packing data) is stored in a memory such as an SDRAM or a system LSI. Or the like is output to the outside of the image encoding apparatus 300.

  Next, a process for decoding an image (hereinafter referred to as an image decoding process) performed by the image decoding apparatus 310 according to the present invention will be described with reference to FIGS. 1 and 2B. FIG. 2B is a flowchart of the image decoding process. The decoding target data is encoded data (that is, packing data) input from the image encoding device 300. The packing data is separated into at least initial pixel value data and an arithmetic code of at least one bit by the unpacking unit 311 (FIG. 2B: step S111).

  If the information separated by the unpacking unit 311 is initial pixel value data (FIG. 2B: YES in step S112), the information is output to the output unit 316 (FIG. 2B: step S113). On the other hand, when the information separated by the unpacking unit 311 is an arithmetic code (FIG. 2B: NO in step S112), the information is output to the arithmetic decoding unit 312.

  The arithmetic decoding unit 312 restores the prediction error information by decoding the arithmetically encoded code (that is, the arithmetic code) sent from the unpacking unit 311 (FIG. 2B: step S114). ). The result decoded by the arithmetic decoding unit 312 is output to the prediction probability update unit 313 and the change extraction unit 314.

  The prediction probability update unit 313 updates the prediction probability in units of bits based on the decoding result input from the arithmetic decoding unit 312 (FIG. 2B: step S115). Note that the processing in the prediction probability update unit 313 overlaps with the processing in the prediction probability update unit 305 included in the image encoding device 300, and thus will not be described in detail. Here, the reference value Pbase of the prediction probability, the update step α, the maximum value Pmax of P, and the minimum value Pmin of P are the same as those used in the encoding process.

  The predicted pixel generation unit 315 uses the initial pixel value data or the previously decoded pixel data input before the target decoding target pixel, and the predicted pixel of the image encoding device 300. A predicted value of the pixel data of interest is generated with the same processing logic as that of the generation unit 303 (FIG. 2B: step S116).

  The change extraction unit 314 performs an exclusive OR between the decoded prediction error information E expressed by N bits sent from the arithmetic decoding unit 312 and the N-bit prediction value input from the prediction pixel generation unit 315. The pixel data is restored by performing the calculation (FIG. 2B: step S117). The restored pixel data is output to the output unit 316.

  The output unit 316 outputs the restored pixel data to the outside of the image decoding device 310 such as a memory such as an SDRAM or a system LSI.

  Next, the image encoding process in the present embodiment will be described in detail using a specific data example.

  FIGS. 8 to 15 and FIG. 26 are diagrams for explaining a specific example of the image encoding process in the present embodiment.

  Here, it is assumed that the processing target pixel value input unit 301 sequentially receives pixel data of fixed bit width (N bits) data. The amount of pixel data received by the processing target pixel value input unit 301 is 4 pixels (M = 4), and each pixel is 6 bits (N = 6). That is, it is assumed that the dynamic range of pixel data is 6 bits.

  FIG. 8 shows four pixel data input to the processing target pixel value input unit 301 as an example. It is assumed that 6-bit pixel data corresponding to each pixel is input to the processing target pixel value input unit 301 in the order of the pixel P801, the pixel P802, the pixel P803, and the pixel P804. The numerical value of the pixel data is a signal level (for example, luminance information) indicated by the corresponding pixel data. Note that the pixel data corresponding to the pixel P801 is initial pixel value data.

  In the present embodiment, it is assumed that the prediction value of the encoding target pixel is calculated by the above-described prediction expression (1) as an example. In this case, the calculated predicted value of the encoding target pixel is the value of the pixel adjacent to the left of the encoding target pixel. That is, it is predicted that the pixel value of the encoding target pixel is highly likely to be the same pixel value (level) as the pixel input immediately before.

  In step S101 of the image encoding process shown in FIG. 2A, the processing target pixel value input unit 301 determines whether or not the input pixel data is initial pixel value data. If the determination result is YES, the processing target pixel value input unit 301 stores the received pixel data in an internal buffer, and transmits the input pixel data to the packing unit 306. And a process transfers to step S108 mentioned later. On the other hand, if the result of the determination in step S101 is NO, the process proceeds to step S102.

  Here, it is assumed that the processing target pixel value input unit 301 has received pixel data as initial pixel value data corresponding to the pixel P801. In this case, the processing target pixel value input unit 301 stores the input pixel data in an internal buffer, and the processing target pixel value input unit 301 transmits the received pixel data to the packing unit 306. When pixel data is stored in the buffer, the processing target pixel value input unit 301 overwrites and stores the received pixel data in the internal buffer.

  Subsequently, it is assumed that the pixel P802 is an encoding target pixel. In this case, it is assumed that the processing target pixel value input unit 301 has received pixel data (encoding target pixel data) corresponding to the pixel P802. The pixel value indicated by the encoding target pixel data is “36” (see FIG. 8). In this case, since the received pixel data is not initial pixel value data (NO in step S101), the processing target pixel value input unit 301 transmits the received pixel data to the change extraction unit 302.

  When it is determined NO in step S <b> 101, the processing target pixel value input unit 301 transmits the pixel data stored in the internal buffer to the predicted pixel generation unit 303. Here, the transmitted pixel data is the pixel value “43” of the pixel P801 (see FIG. 8).

  The processing target pixel value input unit 301 overwrites and stores the received pixel data in an internal buffer. The process proceeds to step S102.

  In step S102, the predicted pixel generation unit 303 calculates a predicted value of the encoding target pixel. Specifically, the predicted pixel generation unit 303 calculates a predicted value using the prediction formula (1). In this case, the pixel value (“43”) indicated by the pixel data received by the predicted pixel generation unit 303 from the processing target pixel value input unit 301 is calculated as the predicted value.

  In step S103, the change extraction unit 302 performs exclusive OR operation processing. Specifically, the change extraction unit 302 calculates a prediction error by calculating an exclusive OR of the pixel data input from the processing target pixel value input unit 301 and the prediction value input from the prediction pixel generation unit 303. Information E is calculated. Here, it is assumed that the received pixel data is “100100” and the predicted value is “101011”. The change extraction unit 302 calculates the prediction error information E “001111” by calculating the exclusive OR of the pixel data (“100100”) and the predicted value (“101011”). The change extraction unit 302 performs this process on the four pixels to be encoded, and transmits three pieces of prediction error information to the arithmetic encoding unit 304 and the prediction probability update unit 305 (step S104).

  In step S105, the arithmetic encoding unit 304 performs arithmetic encoding. Specifically, the arithmetic encoding unit 304 sequentially performs arithmetic encoding on the prediction error information E in units of bits. The order of prediction error information that is arithmetically encoded will be described with reference to FIGS. Now, as shown in FIG. 9A, the prediction error information E to be arithmetically encoded is three, E1, E2, and E3. The arithmetic coding unit 304 performs arithmetic coding with priority on the same bit in the direction from MSB to LSB of each prediction error information. Here, since the prediction error information is E1 (“001111”), E2 (“001101”), and E3 (“000101”), the arithmetic encoding unit 304 is as shown in FIG. 9B. , Input order: encoding is performed in the order of “0, 0, 0, 0, 0, 0, 1, 1, 0...”.

  In step S106, the prediction probability updating unit 305 updates the prediction probability used for arithmetic coding. Specifically, using the information encoded in step S105 as an input, the prediction probability update unit 305 updates the prediction probability used for arithmetic encoding. The update of the prediction probability will be described with reference to FIG. As an initial value with a prediction probability corresponding to “1” of 0.5 (Pbase = 0.5) and a “0” prediction probability of 0.5 (Q = 1−0.5), “maximum of prediction probability P” The value is 0.9 (Pmax = 0.9), the minimum value of the prediction probability P is 0.1 (Pmin = 0.1), and the update step is 0.1 (α = 0.1). Here, when “0” is arithmetically encoded, the prediction probability updating unit 305 predicts the prediction probability of “0” from the prediction probability P = 0.5 and the prediction probability Q = 0.5. Decrease the probability Q by α (that is, increase the prediction probability P by α) to update the prediction probability P = 0.6 and the prediction probability Q = 0.4, followed by arithmetic coding of “0” In this case, the prediction probability update unit 305 updates the prediction probability P = 0.7 and the prediction probability Q = 0.3 in the same manner as before, for example, the prediction probability P = 0.9, the prediction probability Q = 0. When “1” is arithmetically coded in the state of 1, the prediction probability is updated to P = 0.8 and the prediction probability Q = 0.2 The bit length (M pixels × N) to which this process is input When three pieces of prediction error information E1, E2, and E3 are newly input, the same processing is performed with the prediction probability P = 0.5 and the prediction probability Q = 0.5. When “1” is arithmetically encoded, the prediction probability may be lowered by multiplying the prediction probability P by β (0 <β <1 real number).

  Here, the update of the prediction probability in the conventional method will be described with reference to FIG. The initial value, maximum value, minimum value, and update step of the prediction probability are the same as described above, and Pbase = 0.5, Pmax = 0.9, Pmin = 0.1, and α = 0.1. When “0” is arithmetically encoded, the prediction probability of Q is increased by α to increase the prediction probability of “0” from the prediction probability P = 0.5 and the prediction probability Q = 0.5 (that is, The prediction probability P is lowered by α), and the prediction probability P = 0.4 and the prediction probability Q = 0.6 are updated. Subsequently, when “0” is arithmetically encoded, the prediction probability is updated to P = 0.3 and the prediction probability Q = 0.7. For example, when “1” is arithmetically coded in the state of the prediction probability P = 0.1 and the prediction probability Q = 0.9, the prediction probability P = 0.2 and the prediction probability Q = 0.8 are updated. Is done. This process is repeated for the input bit length (M pixels × N bit length). Thus, unlike the prediction probability update unit 305 in the present embodiment, the conventional method updates the prediction probability in a direction to increase the prediction probability of the event that has occurred.

  In step S107, the arithmetic encoding unit 304 determines whether or not the processing has been completed for the M pixels to be arithmetic encoded. If NO in step S107, the process proceeds to step S105, and arithmetic coding by the arithmetic coding unit 304 is performed. If YES in step S107, the process proceeds to S108.

  In step S108, the packing unit 306 performs a packing process for generating a single output data by combining a plurality of arithmetic codes. In the packing process, the packing unit 306 stores the pixel P801 in the buffer memory as the initial pixel value data received from the processing target pixel value input unit 301, and thereafter sequentially receives the code received from the arithmetic coding unit 304 in the buffer. Packing processing is performed by storing in the memory. When the encoding target pixel packing process ends, the process proceeds to step S109.

  In step S <b> 109, the packing unit 306 determines whether or not the image encoding process for M pixels that are units to be packed is completed. If NO in step S109, the process proceeds to step S101, and at least one process from step S101 to step S108 is performed on the pixel data received by the processing target pixel value input unit 301 next. The processing from step S101 to step S108 is repeatedly performed on the pixel P803 and the pixel P804, and sequentially stored in the buffer memory. On the other hand, if YES in step S109, the packing unit 306 outputs the encoded data in the buffer memory from the image encoding device 300 to the outside. Thereby, a process transfers to step S110.

  In step S110, the image encoding device 300 determines whether all the encoding processes for one image have been completed based on the encoded pixel data that has just been output. If YES, the encoding process ends and NO. If there is, the process proceeds to step S101, and at least one process from step S101 to step S109 is executed.

  As a result of executing the above processing and calculation, the calculated code and its length are shown as “present method” in FIG. Furthermore, the result of encoding according to the conventional technique based on the update of the conventional prediction probability shown in FIG. 26 is shown as “conventional technique” in FIG. FIG. 12 shows an update equation on the number line of arithmetic coding, and FIGS. 13 and 14 show arithmetic coding in “present method” and “conventional method” according to the mathematical formula shown in FIG. 12, respectively. Indicates. As shown in FIG. 13, the encoded value is “0.9999858317728” as the calculation processing result of “this method”, and the required resolution is “0.00000004115059”. From this, “0.99998583177728 ÷ 0.00000004115059≈2429626”, and when the result of this decimal number is expressed in binary, the code of “101010001001010111010” of 22 bits shown in FIG. 11 is obtained. Similarly, as shown in FIG. 14, the encoded value is “0.9245554707392” as the calculation result of the “conventional method”, and the required resolution is “0.0000000522547”. When the calculation result in the same manner as in “this method” is expressed in binary, the code “1000011011111101000101101” of 25-bit length shown in FIG. 11 is obtained. As a result, the code length is “25 (bits)” in the conventional method, whereas the code length can be shortened to “22 (bits)” in this method, so that efficient coding is performed. I understand that. Further, the above specific example is an example in which the occurrence frequency of 0 and 1 has almost no deviation. However, since there is usually this bias, the code length is almost smaller than 18 bits, and there is a compression effect by this method.

  In other words, according to the present embodiment, in the arithmetic coding for each bit using the prediction probability, the prediction probability is updated in the direction of decreasing the prediction probability for the event that has occurred (the value of the bit subject to arithmetic coding). To go. Therefore, the prediction error information subject to arithmetic coding goes from the MSB to the LSB, so that the state inversion such as the transition from the high occurrence frequency of “0” to the high occurrence frequency of “1” is supported. The prediction probability is updated, and accordingly, the prediction probability is adaptively updated in consideration of the deviation of the occurrence frequency of bits in the image data, and the coding efficiency of the image data is improved.

  FIG. 15 shows codes and lengths calculated as a result of executing processing and calculation when the initial value of the prediction probability is Pbase = 0.7. Here, the reason why the prediction probability of Pbase is high is that the prediction error information to be encoded is more frequently “0” as the head data to be encoded is. As a result, the code length in the conventional method is “24 (bits)”, whereas the code length in the “present method” is “17 (bits)”, indicating that more efficient encoding is performed. .

<< Embodiment 2 >>
In the second embodiment, a modified example of the image coding apparatus 300 described in the first embodiment will be described.

  FIGS. 16A and 16B are flowcharts illustrating image encoding processing and decoding processing in the image encoding device and the image decoding device according to the second embodiment, respectively. As shown in FIGS. 16A and 16B, steps S1301, S1302, and S1311 are changed from the first embodiment.

  Functions corresponding to those in FIG. 2 are denoted by the same reference numerals, and description thereof is omitted.

  In step S1301, the arithmetic encoding unit 304 according to the present embodiment, for each encoding target pixel, total bits of prediction error information to be arithmetically encoded for each predetermined group (group pixels for M pixels). By limiting the number to a preset fixed bit S (S is a natural number), the output data is packed into S bits. If the packing data does not reach S bits even after all the bit information to be encoded is arithmetically encoded, the packing unit 306 fills up the dummy data (that is, padding) to obtain S-bit packing data. Output.

  In step S1302, under the control of the image coding apparatus 300 in the present embodiment, the arithmetic coding unit 304 ends the arithmetic coding in the group pixel when the packed data reaches S bits, The packing unit 306 outputs the packing data to the outside. As a specific example, as shown in FIGS. 17A and 17B, the arithmetic coding unit 304 performs arithmetic coding of S (here, 13) bits out of prediction error information of group pixels to be coded. When the process ends, the arithmetic coding for the group pixel ends. In other words, when the generated code amount reaches S bits, the arithmetic encoding unit 304 performs arithmetic encoding with the input of arithmetic encoding as “00000011101111” (S bits) only, and then performs arithmetic encoding of “00111”. Do not implement. In this way, the amount of code generated by arithmetic coding for the group pixels is fixed length S bits.

18A to 18D, the prediction error information to be encoded is prioritized from the LSB side and excluded from the encoding target, thereby reducing the encoding target information, and the generated code amount becomes S bits or less. It is a figure explaining the process in the case of performing arithmetic coding in this way. Here, in the case where arithmetic coding is performed by removing the lower 1 bit of the prediction error information E from the encoding target, the arithmetic coding unit 304 predicts the prediction error information when the generated code amount exceeds S bits. The lower 2 bits of E are removed from the encoding target, and arithmetic encoding is performed again. The encoding target code is reduced and arithmetic encoding is repeated until the encoding amount generated in this way becomes S bits or less. FIGS. 18A and 18C are examples in which the lower 1 bit is excluded from the encoding target, and shows prediction error information E _{1 to} E ₃ when only the input of arithmetic encoding is “001110011000010”. . FIGS. 18B and 18D are examples in which the lower 2 bits are excluded from the encoding target, and show prediction error information E _{1 to} E ₃ when the input of arithmetic encoding is only “001100110001”. . Also, the prediction probability reference value Pbase, the update step α, the maximum value Pmax of P, and the minimum value Pmin of P used in FIGS. 17 and 18 are common to the image encoding device 300 and the image decoding device 310. If so, the value may be changed according to the pattern of the input order.

  In step S1311 on the decoding process side, the image decoding apparatus 310 according to the present embodiment determines whether or not the decoding process has been completed for the S-bit arithmetic code to be decoded. If the S-bit decoding process has not been completed (NO in step S1311), the process proceeds to step S114. If the S-bit decoding process has been completed (YES in step S1311), the process proceeds to step S119. In addition, since the arithmetic decoding unit 312 has known M pixels and N bits, the arithmetic decoding unit 312 calculates the bit length that can be restored with the S bit length from the bit length that should be encoded (= M pixels × N bit length). Prediction error information is restored by supplementing “0” as dummy data for the subtracted amount. The dummy data may be “1”.

  By realizing such fixed-length encoding, when the generated fixed-length encoded data is stored in, for example, a memory or the like, encoded data corresponding to a pixel at a specific location in the image is stored. Can be easily identified. As a result, random access to the encoded data can be performed. That is, according to the present embodiment, random accessibility to the memory can be realized. Furthermore, since the bus width can guarantee a fixed length, when data access to certain compressed pixel data is required, it is only necessary to access the packed data packed for each bus width.

<< Embodiment 3 >>
In Embodiment 3, an example of a digital still camera including the image encoding device 300 and the image decoding device 310 described in Embodiment 1 will be described.

  FIG. 19 is a block diagram showing a configuration of a digital still camera 1600 according to the third embodiment. As shown in FIG. 19, the digital still camera 1600 includes an image encoding device 300 and an image decoding device 310. Since the configurations and functions of image coding apparatus 300 and image decoding apparatus 310 have been described in the first embodiment, detailed description thereof will not be repeated.

  The digital still camera 1600 further includes an imaging unit 1610, an image processing unit 1620, a display unit 1630, a compression conversion unit 1640, a recording storage unit 1650, and an SDRAM 1660.

  The imaging unit 1610 images a subject and outputs digital image data corresponding to the image. In this example, the imaging unit 1610 includes an optical system 1611, an imaging element 1612, an analog front end 1613 (abbreviated as AFE in the drawing), and a timing generator 1614 (abbreviated as TG in the drawing). The optical system 1611 is composed of a lens or the like and forms an image of a subject on the image sensor 1612. The image sensor 1612 converts light incident from the optical system 1611 into an electrical signal. As the imaging device 1612, various imaging devices such as an imaging device using a CCD (Charge Coupled Device) and an imaging device using a CMOS can be adopted. The analog front end 1613 performs signal processing such as noise removal, signal amplification, and A / D conversion on the analog signal output from the image sensor 1612, and outputs the result as image data. The timing generator 1614 supplies a clock signal serving as a reference for the operation timing of the image sensor 1612 and the analog front end 1613 to them.

  The image processing unit 1620 is an LSI or the like that performs predetermined image processing on pixel data (RAW data) input from the imaging unit 1610 and outputs the processed image data to the image encoding device 300. In general, as shown in FIG. 19, a white balance circuit 1621 (abbreviated as WB in the figure), a luminance signal generation circuit 1622, a color separation circuit 1623, an aperture correction processing circuit 1624 (abbreviated as AP in the figure), A matrix processing circuit 1625 and a zoom circuit 1626 (abbreviated as ZOM in the drawing) for enlarging / reducing an image are provided. The white balance circuit 1621 is a circuit that corrects the color component by the color filter of the image sensor 1612 at a correct ratio so that a white subject is photographed white under any light source. The luminance signal generation circuit 1622 generates a luminance signal (Y signal) from the RAW data. The color separation circuit 1623 generates a color difference signal (Cr / Cb signal) from the RAW data. The aperture correction processing circuit 1624 performs processing for adding a high frequency component to the luminance signal generated by the luminance signal generation circuit 1622 to make the resolution appear high. The matrix processing circuit 1625 adjusts the spectral characteristics of the image sensor and the hue balance that has been corrupted by image processing on the output of the color separation circuit 1623.

  In general, the image processing unit 1620 temporarily stores pixel data to be processed in a memory such as an SDRAM, performs predetermined image processing, YC signal generation, zoom processing, and the like on the temporarily stored data. Often the data is temporarily stored in the SDRAM again. Therefore, it is conceivable that the image processing unit 1620 is used for both output to the image encoding device 300 and input from the image decoding device 310.

  The display unit 1630 is an LCD or the like that displays an output (image data after image decoding) from the image decoding device 310.

  The compression conversion unit 1640 outputs image data obtained by compressing and converting the output from the image decoding device 310 according to a predetermined standard such as JPEG to the recording storage unit 1650. The compression conversion unit 1640 decompresses and converts the image data read by the recording / storing unit 1650 and outputs the image data to the image encoding apparatus 300. That is, the compression conversion unit 1640 is a processing unit that can process data based on the JPEG standard. Such a compression conversion unit 1640 is generally mounted on a digital still camera.

  The recording storage unit 1650 receives the compressed image data and records it on a recording medium (for example, a non-volatile memory). The recording storage unit 1650 reads compressed image data recorded on the recording medium and outputs the compressed image data to the compression conversion unit 1640.

  In addition, the image coding apparatus 300 and the image decoding apparatus 310 in this Embodiment are not limited to RAW data as an input signal. For example, the data to be processed by the image encoding device 300 and the image decoding device 310 is YC signal (luminance signal or color difference signal) data generated from the RAW data by the image processing unit 1620, temporarily compressed to JPEG, etc. It may be data (luminance signal or color difference signal data) obtained by decompressing the data of the JPEG image that has been processed.

  Here, when the input signals of the image encoding device 300 and the image decoding device 310 are RAW data, the prediction pixel generation unit 303 may generate a prediction value from adjacent pixels having the same color component. For example, it is assumed that the pixel array of the encoding target pixel is Bayer array RAW data. In this case, the RAW data can be divided into an R (red) component, a G (green) component, and a B (blue) component. When the prediction formula (1) is used, the RAW data is adjacent to the encoding target pixel. Instead of the pixel, the pixel on the left side of the same color may be used.

  Thereby, for example, when an image with a large change in R component is input, if RGB is confused in the same group, the MSB in the prediction error information E is small even though the change in G component and B component is small. The frequency of occurrence of “1” on the side increases, and efficient coding cannot be performed. Therefore, by forming a group for each color component, it is possible to perform efficient compression according to the change of each color without being affected by other color components, so the correlation between RGB components This is effective for images with low image quality.

  As described above, the digital still camera 1600 according to the present embodiment is not limited to the compression conversion unit 1640 that is generally mounted on a digital still camera, and the image encoding apparatus 300 and the image for processing RAW data and YC signals. A decoding device 310 is provided. As a result, the digital still camera 1600 according to the present embodiment can perform a high-speed imaging operation that increases the number of continuous shots having the same resolution with the same memory capacity of SDRAM or the like. In addition, the digital still camera 1600 can increase the resolution of a moving image stored in a memory such as an SDRAM having the same capacity.

  The configuration of the digital still camera 1600 shown in the third embodiment is the same as that of the digital still camera 1600. The digital video camera includes an imaging unit, an image processing unit, a display unit, a compression conversion unit, a recording storage unit, and an SDRAM. It is also possible to apply to this configuration.

<< Embodiment 4 >>
In this embodiment, an example of the configuration of a digital still camera when an image sensor provided in the digital still camera includes an image encoding device 300 will be described.

  FIG. 20 is a block diagram illustrating a configuration of a digital still camera 1700 according to the fourth embodiment. As shown in FIG. 20, the digital still camera 1700 includes an imaging unit 1610A instead of the imaging unit 1610, and an image processing unit instead of the image processing unit 1620, as compared with the digital still camera 1600 of FIG. It differs from the point provided with 1620A. Since the other configuration is the same as that of the digital still camera 1600, detailed description will not be repeated.

  The imaging unit 1610A is different from the imaging unit 1610 of FIG. 19 in that the imaging unit 1610A includes an imaging device 1612A instead of the imaging device 1612, and the analog front end 1613 is omitted. Other than that, it is the same as the imaging unit 1610, and therefore, detailed description will not be repeated. The image sensor 1612A is an example of a solid-state image sensor including an image sensor such as a CMOS sensor and the image encoding device 300. The image sensor 1612A includes an amplifier and an A / D conversion circuit in a pixel, and outputs an image for outputting a digital signal. This is a one-chip LSI on which the encoding device 300 is mounted.

  Also, the image processing unit 1620A is different from the image processing unit 1620 of FIG. 19 in that it further includes an image decoding device 310. Since the other configuration is the same as that of image processing unit 1620, detailed description will not be repeated. This image processing unit 1620A is an example of a digital signal processing element incorporating an image decoding device 310a having the same function as the image decoding device 310 in the first embodiment, and is realized as a one-chip LSI.

  The image encoding device 300 included in the image sensor 1612A encodes the pixel signal imaged by the image sensor 1612A, and transmits the data obtained by the encoding to the image decoding device 310 in the image processing unit 1620A.

  The image decoding device 310 in the image processing unit 1620A decodes the data received from the image encoding device 300. With this processing, it is possible to improve the data transfer efficiency between the image sensor 1612A and the image processing unit 1620A in the integrated circuit.

  Therefore, the digital still camera 1700 according to the present embodiment has the same memory capacity as the digital still camera 1600 according to the third embodiment, increases the number of continuous shots with the same resolution, increases the resolution of moving images, and the like. A high-speed imaging operation can be realized.

  The configuration of the digital still camera 1700 shown in the fourth embodiment is the same as that of the digital still camera 1700. The digital video camera includes an imaging unit, an image processing unit, a display unit, a compression conversion unit, a recording storage unit, and an SDRAM. It is also possible to apply to this configuration.

<< Embodiment 5 >>
In the present embodiment, as shown in FIG. 21, an example of the configuration of the monitoring camera when image data received by the monitoring camera is output from the image encoding device 300 will be described.

  Usually, in a surveillance camera device, image data is encrypted in order to ensure security on the transmission path so that image data transmitted from the surveillance camera is not stolen on the transmission path by a third party. In general, image data that has been subjected to predetermined image processing by the image processing unit 1801 in the surveillance camera signal processing unit 1810 is converted into JPEG, MPEG4, H.264 by the compression conversion unit 1802. Personal privacy protection is performed by compressing and converting in accordance with a predetermined standard such as H.264, further encrypting by an encryption unit 1803, and transmitting it from the communication unit 1804 to the Internet. The surveillance camera signal processing unit 1810 is an example of a digital signal processing element including the image decoding device 310, and is realized as a one-chip LSI.

  Therefore, like the monitoring camera 1800 shown in FIG. 21, the output from the imaging unit 1610A including the above-described image encoding device 300 is input to the monitoring camera signal processing unit 1810, and is input into the monitoring camera signal processing unit 1810. Since the image data captured by the imaging unit 1610A can be pseudo-encrypted by decoding the encoded data by the installed image decoding device 310, the imaging unit 1610A and the surveillance camera signal processing unit 1810 The security on the transmission line is ensured, and the security can be further improved as compared with the prior art.

<< Embodiment 6 >>
As a method for realizing the monitoring camera device, as in the monitoring camera 1900 shown in FIG. 22, (1) an image processing unit 1901 that performs predetermined camera image processing on an input image from the imaging unit 1610, and (2) A monitoring camera signal processing unit equipped with a signal input unit 1902 that receives image data transmitted by the image processing unit 1901, performs compression conversion, encrypts the image data, and transmits the image data from the communication unit 1804 to the Internet. There is a form in which 1910 is realized by a separate LSI.

  In this embodiment, the image encoding device 300 is mounted on the image processing unit 1901, and the image decoding device 310 is mounted on the surveillance camera signal processing unit 1910, whereby the image data transmitted by the image processing unit 1901 is simulated. Therefore, security on the transmission path between the image processing unit 1901 and the surveillance camera signal processing unit 1910 is ensured, and it is possible to further improve the security as compared with the prior art. The image processing unit 1901 is an example of a digital signal processing element including the image encoding device 300, and is realized as a one-chip LSI.

  Therefore, according to the present embodiment, the data transfer efficiency of the surveillance camera is improved, and it is possible to realize a high-speed imaging operation such as increasing the resolution of the moving image, and further, by pseudo-encrypting the image data It is possible to improve security such as preventing leakage of image data and protecting privacy.

  As described above, according to the image encoding / decoding device and the method thereof in the present embodiment, in the arithmetic coding for each bit using the prediction probability, an event that has occurred (the target of arithmetic coding) The prediction probability is updated in the direction of lowering the prediction probability with respect to the bit value). Therefore, the prediction error information subject to arithmetic coding goes from the MSB to the LSB, so that the state inversion such as the transition from the high occurrence frequency of “0” to the high occurrence frequency of “1” is supported. The prediction probability is updated, and accordingly, the prediction probability is adaptively updated in consideration of the deviation of the occurrence frequency of bits in the image data, and the coding efficiency of the image data is improved.

  As described above, Embodiments 1 to 6 are used for the image encoding method, the image encoding device, the decoding method, the image decoding device, the solid-state imaging device, the digital signal processing device, the digital still camera, and the surveillance camera according to the present invention. However, the present invention is not limited to these embodiments.

  For example, without departing from the spirit of the present invention, the embodiments obtained by subjecting these embodiments to various modifications conceived by those skilled in the art, and any combination of the components in each embodiment are realized. Forms are also included in the present invention.

  Each functional block in each block diagram is typically realized as an LSI which is an integrated circuit. Each of these functional blocks may be realized as an individual LSI chip, or may be integrated into one chip so as to include a part or all of them.

  Here, although LSI was mentioned as an integrated circuit, it may be an integrated circuit called IC, system LSI, super LSI, or ultra LSI depending on the degree of integration. Further, the method of circuit integration is not limited to LSI or the like, but may be realized by a dedicated circuit or a general-purpose processor. Even if the present invention is realized by using an FPGA (Field Programmable Gate Array) that can be programmed after manufacturing the LSI, or a reconfigurable processor that can reconfigure the connection and setting of the circuit cells in the LSI. Good.

  In each of the above embodiments, each component may be configured by dedicated hardware, or a component that can be realized by software may be realized by executing a program. For example, each component can be realized by a program execution unit such as a CPU reading and executing a software program recorded on a recording medium such as a hard disk or a semiconductor memory.

  The present invention can be used as an encoder that compresses an image, a decoder that decompresses a compressed image, a solid-state imaging device, a digital signal processing device, a digital still camera, and a surveillance camera. In particular, since the image encoding / decoding device according to the present invention can improve the encoding efficiency of image data, it can encode image data with a shorter code length than the conventional method. . Accordingly, an image handling apparatus such as a digital still camera or a network camera can encode / decode image data while preventing deterioration of image quality. Therefore, the present invention is useful for catching up with an increase in the amount of image data processing in recent years.

300 Image Encoding Device 301 Processing Target Pixel Value Input Unit 302 Change Extraction Unit 303 Predictive Pixel Generation Unit 304 Arithmetic Coding Unit 305 Prediction Probability Update Unit 306 Packing Unit 310, 310a Image Decoding Device 311 Unpacking Unit 312 Arithmetic Decoding Unit 313 Prediction probability update unit 314 Change extraction unit 315 Predictive pixel generation unit 316 Output unit 1600 Digital still camera 1610, 1610A Imaging unit 1611 Optical system 1612, 1612A Imaging element 1613 Analog front end 1614 Timing generator 1620, 1620A Image processing unit 1621 White balance Circuit 1622 Luminance signal generation circuit 1623 Color separation circuit 1624 Aperture correction processing circuit 1625 Matrix processing circuit 1626 Zoom circuit 1630 Display unit 1640 Compression conversion unit 165 Record storage unit 1660 SDRAM
1700 Digital Still Camera 1800 Monitoring Camera 1801 Image Processing Unit 1802 Compression Conversion Unit 1803 Encryption Unit 1804 Communication Unit 1810 Monitoring Camera Signal Processing Unit 1900 Monitoring Camera 1901 Image Processing Unit 1902 Signal Input Unit 1910 Monitoring Camera Signal Processing Unit

Claims (12)

An image encoding method for compressing pixel data constituting image data,
A prediction pixel generation step of generating a prediction value of the encoding target pixel from pixel data of at least one pixel located around the encoding target pixel;
A change extracting step of generating prediction error information indicating the bit change by extracting a bit change between the pixel data of the encoding target pixel and the prediction value;
An arithmetic encoding step for performing arithmetic encoding of the prediction error information generated by the change extraction step using an adaptively changing prediction probability for each bit constituting the prediction error information;
When the bit encoded in the arithmetic encoding step is the first value, the prediction probability used for arithmetic encoding of the first value is lowered, and the bit encoded in the arithmetic encoding step is A prediction probability update step of updating the prediction probability so as to increase the prediction probability used for arithmetic coding of the first value when the value is not 1,
An image encoding method in which the arithmetic encoding step performs the arithmetic encoding using the prediction probability updated in the prediction probability update step for each bit constituting the prediction error information. The image encoding method according to claim 1, comprising:
In the arithmetic coding step, for each group pixel which is a plurality of predetermined pixels, a predetermined number smaller than the total number of bits of the plurality of prediction error information among the plurality of prediction error information corresponding to the group pixel is determined. When the arithmetic coding is performed only for the fixed number of bits S and the number of bits obtained by arithmetic coding of the prediction error information corresponding to the group pixel is less than the S bits, the arithmetic coding is performed. An image encoding method for generating an S-bit fixed-length arithmetic code by padding dummy data into bits obtained in step 1 above. An image encoding method according to claim 2, wherein
In the arithmetic coding step, the fixed-length arithmetic code is generated by ending arithmetic coding when a code amount generated by the arithmetic coding reaches the S bits. An image decoding method for decoding pixel data from an arithmetic code obtained by the image encoding method according to claim 1,
An arithmetic decoding step for generating the prediction error information by performing arithmetic decoding using an adaptively changing prediction probability for each bit constituting the arithmetic code;
When the bit decoded in the arithmetic decoding step is the first value, the prediction probability used for the arithmetic decoding of the first value is lowered, and the bit decoded in the arithmetic decoding step is changed to the first value. A prediction probability update step of updating the prediction probability so as to increase the prediction probability used for arithmetic decoding of the first value when the value is not 1,
A predicted pixel generation step of generating a predicted value of the decoding target pixel;
Change extraction for generating pixel data of a pixel to be encoded by performing a predetermined calculation on the prediction error information generated in the arithmetic decoding step and the prediction value generated in the prediction pixel generation step Including steps,
The image decoding method, wherein, in the arithmetic decoding step, the arithmetic decoding is performed using the prediction probability updated in the prediction probability updating step for each bit constituting the arithmetic code. An image encoding device for compressing pixel data constituting image data,
A prediction pixel generation unit that generates a prediction value of the encoding target pixel from pixel data of at least one pixel located around the encoding target pixel;
A change extracting unit that generates prediction error information indicating the bit change by extracting a bit change between pixel data of the encoding target pixel and the predicted value;
An arithmetic encoding unit that performs arithmetic encoding on the prediction error information generated by the change extraction unit using a prediction probability that adaptively changes for each bit that constitutes the prediction error information;
When the bit encoded by the arithmetic encoding unit is the first value, the prediction probability used for arithmetic encoding of the first value is lowered, and the bit encoded by the arithmetic encoding unit is A prediction probability update unit that updates the prediction probability so as to increase the prediction probability used for arithmetic coding of the first value when the value is not 1,
The image coding apparatus, wherein the arithmetic coding unit performs the arithmetic coding using the prediction probability updated by the prediction probability update unit for each bit constituting the prediction error information. An image decoding apparatus for decoding pixel data from an arithmetic code obtained by the image encoding method according to claim 1,
An arithmetic decoding unit that generates the prediction error information by performing arithmetic decoding using an adaptively changing prediction probability for each bit constituting the arithmetic code;
When the bit decoded by the arithmetic decoding unit is the first value, the prediction probability used for arithmetic decoding of the first value is lowered, and the bit decoded by the arithmetic decoding unit is changed to the first value. A prediction probability updating unit that updates the prediction probability so as to increase the prediction probability used for arithmetic decoding of the first value when the value is not 1,
A prediction pixel generation unit that generates a prediction value of a decoding target pixel;
Change extraction for generating pixel data of a pixel to be decoded by performing a predetermined calculation on the prediction error information generated by the arithmetic decoding unit and the prediction value generated by the prediction pixel generation unit With
The image decoding device, wherein the arithmetic decoding unit performs the arithmetic decoding using the prediction probability updated by the prediction probability update unit for each bit constituting the arithmetic code.   A solid-state imaging device comprising the image encoding device according to claim 5.   A digital signal processing element comprising the image encoding device according to claim 5.   A digital signal processing element comprising the image decoding device according to claim 6.   A digital still camera comprising at least one of the solid-state imaging device according to claim 7, the digital signal processing device according to claim 8, and the digital signal processing device according to claim 9.   A digital video camera comprising at least one of the solid-state imaging device according to claim 7, the digital signal processing device according to claim 8, and the digital signal processing device according to claim 9.   A surveillance camera comprising at least one of the solid-state imaging device according to claim 7, the digital signal processing device according to claim 8, and the digital signal processing device according to claim 9.

Images