JP6632752B2 - Image encoding apparatus and method, program and storage medium - Google Patents

Description

  The present invention relates to an image encoding technique for encoding image data.

  JPEG-LS is one of the image data encoding methods. In JPEG-LS, lossless coding or near lossless coding can be selected, and a high compression ratio is realized by switching between predictive coding and run-length coding based on the state of pixels surrounding the pixel to be coded. The prediction coding predicts a pixel value of a coding target pixel from surrounding pixels, and codes a prediction error. In run-length encoding in JPEG-LS, a run length (the number of continuations) of the same pixel value is encoded. As another method using prediction encoding and run-length encoding, a method of selecting and encoding identification information of a plurality of prediction methods, their run lengths (continuous values), and prediction errors is also known (for example, , Patent Document 1).

Japanese Patent No. 3888513

  When predictive coding and run-length coding are compared, it is clear that the more continuous the same data, the overwhelmingly higher coding efficiency of the latter over the former. However, in the conventional encoding techniques, which encoding is used for the subband obtained by the wavelet transform based on the same uniform criterion, there is still room for improvement.

  In view of the above, an object of the present invention is to provide a technique for realizing higher coding efficiency while using a wavelet transform.

In order to solve this problem, for example, an image encoding device of the present invention has the following configuration. That is,
An image encoding device that encodes image data,
Transform means for performing a wavelet transform on the image data to be encoded to generate a plurality of subbands;
A first coding for predictively encoding the coefficient of interest by inputting coefficients constituting the subband in raster scan order and obtaining a predicted value of the coefficient of interest from encoded coefficients located around the coefficient of interest Means,
Second encoding means for inputting the coefficients in the subband in the raster scan order, and obtaining and encoding a run representing the number of consecutive same coefficients;
A first determination as to whether or not the subband of interest to which the target coefficient belongs is a subband LL, and a second determination to determine the relationship between the values of a plurality of coded coefficients located around the target coefficient. And determining means for determining whether the coefficient of interest is to be encoded by the first encoding means or to be encoded by the second encoding means using the coefficient of interest as a starting point of a run, based on You .

  According to the present invention, it is possible to realize higher coding efficiency while using the wavelet transform.

FIG. 2 is a diagram illustrating an example of a configuration of an image encoding device. The figure which shows an example of a structure of an encoding part. FIG. 2 is a diagram illustrating a hardware configuration of an information processing unit. FIG. 3 is a diagram illustrating an example of a configuration of an imaging unit of the camera. 5 is a flowchart illustrating processing of an encoding unit. 9 is a flowchart illustrating processing of a subband encoding unit. 5 is a flowchart illustrating processing of a first mode determination unit. 9 is a flowchart illustrating a process of a second mode determination unit. FIG. 4 is a diagram illustrating subband division by wavelet transform. The figure which shows the positional relationship of an attention coefficient and a peripheral coefficient. The figure explaining Golomb-Rice encoding. 9 is a flowchart illustrating processing of a second mode determination unit according to the second embodiment.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. The embodiment described below shows an example in which the present invention is specifically implemented, and is one of the specific embodiments having the configuration described in the claims.

[First Embodiment]
The first embodiment is applied to an image encoding device that divides image data into a plurality of subbands by wavelet transform, and encodes the subband-divided wavelet coefficients by predictive encoding and run-length encoding. You. Hereinafter, the image encoding device according to the present embodiment will be described.

  In the first embodiment, an example in which the image encoding device is applied to a camera will be described. FIG. 1 is a block diagram of the camera 100. The camera 100 includes an operation unit 101, an imaging unit 102, an information processing unit 109, a storage unit 107, and an I / O interface 108. The information processing unit 109 includes an acquisition unit 103, an image processing unit 104, an encoding unit 105, and an output unit 106.

  FIG. 4A is a front view of the camera 100. As shown in FIG. 4A, the camera 100 includes an imaging unit 102. FIG. 4B is a diagram illustrating an internal configuration of the imaging unit 102. The imaging unit 102 includes a zoom lens 401, focus lenses 402 and 403, an aperture stop 404, and a shutter 405. Further, it has an optical low-pass filter 406, an iR cut filter 407, a color filter 408, an image sensor 409, and an A / D converter 410. The user can adjust the amount of incident light incident on the imaging unit 102 by adjusting the aperture 404. The imaging element 409 is a light receiving element such as a CMOS or a CCD. When the light amount of the subject is detected by the imaging element 409, the detected light amount is converted into a digital value by the A / D conversion unit 410 and output to the information processing unit 109 as digital data.

  FIG. 3 is a diagram showing the internal configuration of the information processing unit 109. The information processing unit 109 has a CPU 301, a RAM 302, and a ROM 303, and the components are mutually connected by a system bus 304. The ROM 303 stores programs executed by the CPU 301 and various data, and the RAM 302 is used as a work area and various buffers when the CPU 301 executes the programs. The CPU 301 executes a program stored in the ROM 303, and uses the RAM 302 as a work area and a buffer. This functions as the acquisition unit 103, the image processing unit 104, the encoding unit 105, and the output unit 106 shown in FIG. 1 and controls the entire apparatus. Here, each component is realized by the CPU 301 executing the component, but may be realized by a dedicated circuit or the like that performs the same processing.

  The operation unit 101 is an input device such as a button, a dial, and a touch panel provided in the camera body, and can be operated by a user to perform commands such as start and stop of shooting and setting of shooting conditions. The storage unit 107 is a non-volatile storage medium such as a memory card that can store the RAW image data acquired by the imaging unit 102 and the image data. The I / O interface 108 can utilize a serial bus connection implemented by a universal serial bus (USB) and has a corresponding USB connector. Of course, a LAN connection or a wireless connection using an optical fiber may be used.

  The display unit 305 displays captured images and characters. A liquid crystal display is generally used for the display unit 305. In addition, a touch panel function may be provided. In that case, a user instruction using the touch panel can be treated as an input of the operation unit 101.

  Hereinafter, processing of the information processing unit 109 according to the present embodiment will be described. The acquisition unit 103 acquires the RAW image data output from the imaging unit 102 and outputs the RAW image data to the image processing unit 104 based on a user's imaging instruction to the operation unit 101. The image processing unit 104 performs a demosaic process on the RAW image data to generate image data in which one normal pixel includes three components of RGB, and outputs the image data to the encoding unit 105. The encoding unit 105 performs an encoding process on the image data. In the present embodiment, one pixel is represented by 8 bits for each of RGB, and a case in which a color component is encoded as a monochrome image will be described as an example. That is, the monochrome multi-valued image of the R component, the monochrome multi-valued image of the G component, and the monochrome multi-valued image of the B component are independently encoded. The image data to be encoded is not limited to the RGB format, but may be another color space or RAW image data. For example, when one pixel is composed of Y, Cb, and Cr, each of which is composed of 8 bits, each component may be encoded as a monochrome image. When applied to RAW image data, the imaging elements of the imaging unit 102 have a Bayer array (array of a set of 2 × 2 imaging elements corresponding to R, G0, G1, and B colors). The signal from each image sensor is represented by 14 bits. In this case, a monochrome multi-valued image composed of R components, a monochrome multi-valued image composed of G0 components, a monochrome multi-valued image composed of G1 components, and a monochrome multi-valued image composed of B components are respectively encoded. I do. Therefore, the present application is not limited by the type of the color space of the image, the type of the color component constituting the image, and the number of bits. It should be understood that the reason why the image to be encoded is composed of three components of R, G, and B and has 8 bits (256 gradations) for the sake of easy understanding.

  The processing of the encoding unit 105 according to the present embodiment will be described. FIG. 2 shows an internal configuration of the encoding unit 105, and FIG. 5 shows an encoding process performed on image data of one color component (in the embodiment, any of R, G, and B) performed by the encoding unit 105. It is a flowchart. That is, the process of FIG. 5 is actually executed three times. Then, the code data obtained as a result of performing the operations three times is combined to form one file.

  The color component input unit 201 inputs image data of one component in the image data output from the image processing unit 104 as a monochrome multivalued image (step S501).

  The wavelet transform unit 202 performs a wavelet transform using the monochrome image data supplied from the color component input unit 201 as an input. The image data is divided into a plurality of decomposition levels by a wavelet transform, and each decomposition level is composed of a plurality of subbands. The decomposition level indicates the number of times the low-frequency component is recursively divided into two in the horizontal and vertical directions. When the decomposition level increases by one, the horizontal and vertical resolutions become half.

  FIG. 9 shows an example in which decomposition is performed three times by the wavelet transform. The decomposition level 1 subbands are represented as LL1, HL1, LH1, and HH1, the decomposition level 2 subbands are represented as LL2, HL2, LH2, HH2, and the decomposition level 3 subbands are represented as LL3, HL3, LH3, and HH3. In the second and subsequent wavelet transforms, the subbands LL and LL2 obtained in the last wavelet transform are omitted because the subbands LL obtained in the immediately preceding wavelet transform are to be converted. Will remain. For example, the horizontal and vertical resolution of HL2 is half of HL1. The subband LH indicates a horizontal frequency characteristic (horizontal component) of the local region to which the wavelet transform has been applied. The sub-band HL indicates frequency characteristics in the vertical direction (vertical component), and the sub-band HH indicates frequency characteristics in the oblique direction (oblique component). The sub-band LL is a low frequency component. In the present embodiment, a real number type 9/7 filter which is also used in JPEG2000 (ISO / IEC15444 | ITU-T T.800) which is an international standard is used, but the present invention is not limited to this. Other wavelet transform filters such as an integer type 5/3 filter may be used. Further, the processing unit of the wavelet transform may be a processing of a line or a processing of an image.

  The wavelet transform unit 202 outputs the wavelet coefficients of the respective subbands subjected to the wavelet transform to the quantization unit 203 (Step S502).

  The quantization unit 203 receives the wavelet coefficient of each subband supplied from the wavelet transform unit 202 and quantizes the subband with a different step width for each subband. Further, another quantization method may be used, such as performing quantization with a different step width for each sub-band line. The quantization unit 203 outputs the quantized wavelet coefficients to the subband encoding unit 216 for each subband (step 503).

  Hereinafter, the process of the subband encoding unit 216 is performed. As will be described later in detail, the sub-band encoding unit 216 encodes each sub-band.

  First, in step S504, the number of times of decomposition by the wavelet transform is set to a variable i. In the present embodiment, an example in which wavelet transform is performed three times will be described, and a variable i is set to 3. Note that the number of wavelet transforms can be appropriately changed by the user from the operation unit 101. Further, in the flowchart of FIG. 5, the subbands LL (i), HL (i), LH (i), and HH (i) are generalized using the variable i. For example, HL (3) in FIG. It should be understood that it shows the same subband as HL3.

  In step S505, the subband encoding unit 216 performs subband encoding of the subband LL (3) (LL3 in FIG. 9). Therefore, the sub-band encoding unit 216 receives the quantized wavelet transform coefficients of the sub-band LL3 from the quantization unit 203, and performs encoding. The subband encoding unit 216 stores the encoded data obtained by this encoding process in a buffer area reserved in the RAM 302 in advance.

  Thereafter, in step S506, subband encoding section 216 determines whether variable i is 0 or not. Here, it is assumed that i = 3 at the initial stage. Therefore, the process proceeds to step S507.

  In step S507, the sub-band encoding unit 216 receives the quantized wavelet transform coefficient of the sub-band HL (3) from the quantization unit 203, performs sub-band encoding, and converts the obtained encoded data. Store in the above buffer area. In step S508, the sub-band encoding unit 216 receives the quantized wavelet transform coefficients of the sub-band LH (3) from the quantization unit 203, performs sub-band encoding, and converts the obtained encoded data into the above-described data. In the buffer area. Then, in step S509, the sub-band encoding unit 216 receives the quantized wavelet transform coefficient of the sub-band HH (3) from the quantization unit 203, performs sub-band encoding, and obtains encoded data. Is stored in the buffer area.

  Thereafter, in step S510, the number i of decompositions using wavelet coefficients is reduced by one. Here, as a result, the variable i = 2. Then, the process returns to step S506. The above processing is performed until it is determined in step S506 that the variable i = 0.

  If it is determined that the variable i has become 0, LL (3), HL (3), LH (3), HH (3), HL (2),..., HH (1) are stored in the buffer area of the RAM 302. Coded data of a total of 10 subbands are stored (step S511).

  Although the above is an encoding process of one color component, as described above, in the embodiment, since each of the R, G, and B color components is encoded, the encoding unit 105 performs the above-described process by three. Will be executed several times. Returning to FIG. 1, the encoding unit 105 sends encoded data of all color components to the output unit 106. The output unit 106 forms an image file by adding an appropriate header to the supplied encoded data, outputs the compressed image data file to the storage unit 107, and stores the compressed image data file. The stored compressed image data file may be output to a device outside the camera via the I / O interface 108.

  The overall processing flow when the image data is divided into sub-bands by the wavelet transform and encoded for each sub-band has been described. Next, the details of the processing of the subband encoding unit 216 that encodes the wavelet coefficients of each subband will be described.

  FIG. 2 shows the internal configuration of the subband encoding unit 216, and FIG. 6 is a flowchart showing the processing performed by the subband encoding unit 216.

  First, in step S601, the sub-band encoder 216 inputs the sub-band number S. The subband number S is a number for identifying a subband, and is used to identify whether the subband to be encoded is any of the LL component, the HL component, the LH component, and the HH component. For example, identification numbers such as S = 1 for the LL component, S = 2 for the HL component, S = 3 for the LH component, and S = 4 for the HH component are set. Note that when outputting the quantized wavelet transform coefficients to the subband encoding unit 216 in units of subbands, the quantization unit 203 adds a subband number S to the beginning.

  In step S602, the subband encoding unit 216 determines whether all the wavelet coefficients in the subband have been encoded. If all of the coefficients have been coded, the coding process ends. If there are uncoded coefficients, the process proceeds to step S603. In step S603, the subband encoding unit 216 inputs the wavelet coefficients of the subband of interest in line units. Then, in step S604, the sub-band encoding unit 216 sets a wavelet coefficient located at the left end of the input line as a target coefficient to be encoded.

  FIG. 10 shows the positional relationship between the coefficient of interest and its surrounding coefficients. Here, x is a target coefficient to be encoded. As shown in the figure, the reference range is a coefficient around the coefficient of interest x in two lines including the line where the coefficient of interest x is located and the line immediately before the line. Further, since the coefficients are encoded in the raster scan order, the left coefficient a, the upper left coefficient c, the upper coefficient b, the upper right coefficient d, the coefficient e located two to the left from the target coefficient x, and three from the target coefficient x The coefficient f located to the left by the amount is a coded coefficient. When the coefficient of interest x is located at the left end of the first line of the subband, the coefficients a to f do not exist. When the coefficient of interest x is located at the left end of the second and subsequent lines of the subband, the coefficients a, c, e, and f do not exist. Such a coefficient that does not exist is regarded as having a predetermined value (for example, 0). In addition, in order to refer to the coefficient at the already coded position, the subband coding unit 216 has, for example, a buffer memory for two lines.

  In step S606, subband encoding section 216 determines whether all wavelet coefficients in the line have been encoded. In other words, it is determined whether or not the coding of the coefficient at the end of the line (right end) has been completed. If it is determined that the encoding of all the coefficients in the line has been completed, the process moves to the next line in step S605. If there are uncoded coefficients, the process moves to step S607.

  In step S607, the mode selection unit 204 selects an encoding mode based on the encoding mode determination according to the subband. The coding mode to be selected includes a predictive coding mode and a run-length mode. When the coefficient at the left end of the line is set as the target coefficient, the run length RL = 0 is initially set. When the run length RL is other than 0, or when the state of the surrounding coefficient satisfies the run length coding condition, the coding mode is set to the run length mode. When the run length RL is 0, it is determined whether the run length encoding condition is satisfied by the following processing.

  The mode selection unit 204 selects the first mode determination unit 205 and the second mode determination unit 206. Based on the subband number S input in step S601, if the subband is an LL component, an HL component, or an HH component, the first mode determination unit 205 is selected. If the subband is an LH component, the second mode determination unit 206 is selected. When the coefficient at the left end of the line is set as the target coefficient, the second mode determination unit 206 initializes the horizontal run count HR = 0.

  Here, the determination process performed by the first mode determination unit 205 (the determination process when the encoding target subband is LL, HL, or HH) will be described with reference to the flowchart of FIG.

  In step S701, a target coefficient x is obtained. Then, in step S702, a determination of the run length mode is performed. As shown in FIG. 10, a left coefficient a, an upper left coefficient c, an upper coefficient b, and an upper right coefficient d are referred to. When the subband LL is to be encoded, it is determined that the coefficient of interest x satisfies the run-length encoding condition when a = b = c = d. In the case of subband HL or HH, it is determined that the run-length encoding condition is satisfied when a = b = c = d = 0. This is because, in the high-frequency component, many 0s appear in the wavelet coefficient value, and thus the coding efficiency is increased by using the run-length mode determination limited to the coefficient value 0. The run-length mode determination in step S702 refers to not only the line of the target coefficient x but also the coefficients of the immediately preceding line (the upper left coefficient is c, the upper coefficient is b, and the upper right coefficient is d). Call.

  In step S703, it is determined whether a run-length encoding condition is satisfied. If the run-length encoding condition is satisfied, a run-length mode is set in step S705. If the run-length encoding condition is not satisfied, the prediction encoding mode is set in step S704. The processing performed by the first mode determination unit 205 is the same processing as JPEG-LS.

  Next, FIG. 8 is a flowchart showing the processing performed by the second mode determination unit 206 (determination processing when the encoding target subband is LH).

  In step S801, the second mode determination unit 206 acquires a target coefficient x. In step S802, the second mode determination unit 206 determines a run length mode. As shown in FIG. 10, a left coefficient a, an upper left coefficient c, an upper coefficient b, and an upper right coefficient d are referred to. When a = b = c = d = 0, the second mode determining unit 206 determines that the coefficient of interest x satisfies the run-length encoding condition. Similar to step S702, the run-length mode determination in step S802 is referred to as a two-line (multiple-line) reference mode determination because the line of the target coefficient x and the coefficient of the line immediately before are referred to.

  In step S803, the second mode determination unit 206 receives the result of step S801 and determines whether or not the run-length encoding condition is satisfied. If the run-length encoding condition is satisfied, a run-length mode is set in step S808. If the run-length encoding condition is not satisfied, the process moves to step S804.

  In step S804, a run length mode different from that in step S802 is determined. Only the same line as the coefficient x of interest is referred to, and the number of continuations of the coefficient value 0 (horizontal run count) is used as a determination condition. When the horizontal run count HR is equal to or greater than a predetermined number (threshold), a run length encoding condition is satisfied. Here, the threshold value is set to 3. Explaining with reference to FIG. 10, when a left coefficient a, a two-left coefficient e, and a three-left coefficient f are set, a = e = 0, and if f is non-zero, the horizontal run count HR is 2. . The run-length mode determination in step S804 refers to only the coefficient on the same line with respect to the target coefficient x, and is therefore referred to as one-line reference mode determination.

  In step S805, the second mode determination unit 206 receives the result of step S804, and determines whether or not the coefficient of interest x satisfies the run-length encoding condition. If the run-length encoding condition is satisfied, a run-length mode is set in step S808. If the run-length encoding condition is not satisfied, the prediction encoding mode is set in step S806.

  If the prediction encoding mode has been set in step S806, the horizontal run count HR is updated in step S807. When the coefficient of interest x is 0, 1 is added to the horizontal run count HR. If the coefficient of interest x is a coefficient value other than 0, the horizontal run count HR is initialized to 0.

  If the run length mode has been set in step S808, the horizontal run count HR is initialized to 0 in step S809. This is because the coefficients a, e, and f located to the left of the coefficient of interest x have already been coded, and the coefficient of interest x is used as the starting point of the run.

  The above is the processing of step S607. The sub-band LL uses a two-line reference mode determination not limited to the coefficient value 0, and the sub-bands HL and HH use a two-line reference mode determination limited to the coefficient value 0. The sub-band LH indicates a horizontal frequency characteristic (horizontal component) of the local region to which the wavelet transform is applied, and thus the correlation between the wavelet coefficient lines is low. Therefore, in addition to the two-line reference mode determination limited to the coefficient value 0 and the one-line reference mode determination that does not refer to the upper line, the determination accuracy of the run-length mode is improved, and the coding efficiency can be improved. Become. In addition, the one-line reference mode can be determined by simply updating the horizontal run count HR, so that it can be realized by a simple calculation. Here, the one-line reference mode determination is added to the sub-band LH, but the same processing may be performed for the sub-bands HL and HH.

  Returning to the description of the processing in FIG. If the run-length mode has been set for the coefficient of interest x in step S608, the process proceeds to step S612, and the run-length encoding process by the run-length encoding unit 213 starts.

  Once the processing is started in step S612, the run-length encoding unit 213 increases the run length RL by 1 as long as the coefficient of interest x has the same value as the coefficient a on the left (a predicted value for the coefficient of interest). , The coefficient on the right side is continuously read as a new coefficient of interest x. However, when the noted coefficient x has a value different from the immediately preceding coefficient a, or when the noted coefficient x is the same as the coefficient a on the left, but reaches the end (right end) position of the line. Outputs the code word of the run length RL counted up to that time, and initializes the run length RL. Then, the process returns to step S606. Since the run-length encoding is the same as that of JPEG-LS, the detailed description is omitted here.

  On the other hand, if the process has proceeded to step S609, the prediction conversion selection unit 207 selects the first prediction conversion unit 208, the second prediction conversion unit 209, the third prediction conversion unit 210, and the fourth prediction conversion unit 211. Do. Based on the subband number S input in step S601, if the subband is an LL component, the first predictive conversion unit 208 is selected. If the subband is the HL component, the second prediction conversion unit 209 is selected. If the subband is the LH component, the third prediction conversion unit 210 is selected. If the subband is the HH component, the fourth prediction conversion unit 211 is selected.

ここで予測誤差（差分）は以下のようにして算出する。
予測誤差 ＝ 着目係数ｘの係数値 ― 予測値ｐ The first predictive conversion unit 208 calculates a predictive value p using MED (Median Edge Detection) prediction for predictive conversion and a prediction error using the predictive value p. The formula for calculating the predicted value p using the peripheral coefficients in FIG. 10 is as follows.

Here, the prediction error (difference) is calculated as follows.
Prediction error = coefficient value of coefficient of interest x−prediction value p

The second prediction conversion unit 209 uses prediction in the vertical direction for prediction conversion. Since the HL component has frequency characteristics in the vertical direction, vertical prediction is effective for improving coding efficiency. The prediction value p and the prediction error using the peripheral coefficients in FIG. 10 are as follows.
p = b
Prediction error = coefficient value of coefficient of interest x−prediction value p

The third predictive conversion unit 210 performs horizontal prediction for predictive conversion. Since the LH component has frequency characteristics in the horizontal direction, horizontal prediction is effective for improving coding efficiency. The prediction value p and the prediction error using the peripheral coefficients in FIG. 10 are as follows.
p = a
Prediction error = coefficient value of coefficient of interest x−prediction value p

  The fourth predictive conversion unit 211 does not perform predictive conversion. That is, the prediction error is as follows. Prediction error = coefficient value of the coefficient of interest x

Returning to FIG. 2, the first prediction conversion unit 208, the second prediction conversion unit 209, the third prediction conversion unit 210, and the fourth prediction conversion unit 211 output the calculated prediction errors to the Golomb encoding unit 212. The Golomb encoding unit 212 performs Golomb-Rice encoding of the prediction error (step S610). The Golomb encoding unit 212 first converts the prediction error (Diff) into a non-negative integer value (MV). The conversion formula is as follows.

Next, the Golomb encoding unit 212 performs Golomb-Rice encoding of the non-negative integer value (MV) using the parameter k. The procedure of Golomb-Rice coding is as follows.
(1) The MV is expressed in a binary number, and the value 0 obtained by shifting the MV to the right by k bits is arranged, and 1 is added thereafter.
(2) The lower k bits of the MV are taken out and added to the end of the bit string generated above.

  FIG. 11 shows the relationship between a parameter k of Golomb-Rice coding, a non-negative integer value (MV), and a codeword. The configuration of Golomb-Rice coding is not limited to this. For example, a code may be configured by reversing 0 and 1, and the order of (1) and (2) described in the above procedure may be changed. The codes may be replaced with each other. Here, the method for determining the encoding parameter k is not particularly specified, but it is sufficient that the same parameter can be used on the encoding side and the decoding side. For example, a method of selecting a suitable parameter k for each subband in advance may be used, or the parameter k may be dynamically changed during the encoding process.

  For example, consider the case where the parameter k = 0. When the prediction error (Diff) is -2, the non-negative integer value (MV) is 3, and when the prediction error (Diff) is +2, the non-negative integer value (MV) is 4. When the non-negative integer value (MV) is 3, the code word becomes 0001 and is represented by 4 bits. When the non-negative integer value (MV) is 4, the code word becomes 00001 and is represented by 5 bits.

  When the Golomb encoding unit 212 finishes outputting the codeword for the coefficient of interest x as described above, the process sets the coefficient on the right of the coefficient of interest x as a new coefficient of interest in step S611, and returns to step S606.

  Returning to FIG. 2, code generation section 214 outputs code data of wavelet coefficients of subbands.

  As described above, according to the present embodiment, it is possible to reduce the code amount by using different run-length mode determination processing for the sub-band divided wavelet coefficients according to the sub-band. Specifically, a one-line reference mode determination that does not refer to the immediately preceding line is added to the sub-band LH indicating the horizontal frequency characteristic (horizontal component). As a result, it is easy to shift to run-length encoding under the situation where run-length encoding is estimated to be efficient, and the encoding efficiency can be improved with a simple operation.

[Second embodiment]
The first embodiment is directed to an image encoding device that divides image data into a plurality of subbands by wavelet transform, and predictively encodes the subband-divided wavelet coefficients by run-length encoding. Applied. It has been described that the coding efficiency can be improved by a simple calculation by adding the one-line reference mode determination that does not refer to the upper line to the subband LH indicating the frequency characteristics (horizontal components) in the horizontal direction.

  The second embodiment is also applied to an image encoding device that divides image data into a plurality of subbands by wavelet transform, and predictively encodes the subband-divided wavelet coefficients by run-length encoding. You. In the second embodiment, a method for reducing the memory used by a simple calculation while maintaining the coding efficiency by the coding mode determination according to the subband will be described.

  The device configuration in the second embodiment is the same as that in FIGS. 1 to 3 of the first embodiment, and the basic processing is also the same. The difference between the second embodiment and the first embodiment lies in the processing content of step S607 in FIG. That is, the criterion for selecting the first mode determination unit 205 and the second mode determination unit 206 in the mode selection unit 204 in the second embodiment is different from that in the first embodiment.

  In step S601 of FIG. 6, mode selection section 204 selects first mode determination section 205 based on the input subband number S if the subband is an LL component. In addition, the mode selection unit 204 selects the second mode determination unit 206 if the subband of interest is any of LH, HL, and HH other than the subband LL. When the coefficient at the left end of the line is set as the target coefficient, the second mode determination unit 206 initializes the horizontal run count HR = 0.

  The first mode determination unit 205 is common to the first embodiment, and a description thereof will not be repeated. Thus, the processing performed by the second mode determination unit 206 will be described with reference to the flowchart of FIG.

  In step S1201, the second mode determination unit 206 obtains the coefficient of interest x. Then, the second mode determination unit 206 determines the run length mode in step S1202. Only the same line as the target coefficient x is referred to, and the number of continuous coefficient values 0 (horizontal run count) is set as a determination condition. When the horizontal run count HR is equal to or larger than the threshold, the run length encoding condition is satisfied. Here, the threshold value is set to 3. To explain with reference to FIG. 10, when a left coefficient a, a two-left coefficient e, and a three-left coefficient f, (a = e = 0, f is a coefficient value other than 0), the horizontal run count HR is 2. Here, the number of consecutive coefficient values 0 is used as the determination condition, but the coefficient value is not limited to 0, such as the number of consecutive coefficient values equal to the left coefficient a is used as the determination condition.

  In the run-length mode determination in step S1202, only the coefficient on the same line as the target coefficient x is referred to, so that this is referred to as one-line reference mode determination.

  In step S1203, the second mode determining unit 205 receives the determination result in step S1202, and determines whether the run-length encoding condition is satisfied. If the run-length encoding condition is satisfied, a run-length mode is set in step S1206. If the run-length encoding condition is not satisfied, the mode is set to the predictive encoding mode in step S1204.

  If the prediction encoding mode has been set in step S1204, the second mode determination unit 205 updates the horizontal run count HR in step S1205. When the coefficient of interest x has a coefficient value of 0, 1 is added to the horizontal run count HR (counting up). If the coefficient of interest x is a coefficient value other than 0, the horizontal run count HR is initialized to 0 (reset).

  If the run length mode has been set in step S1206, the first mode determination unit 205 initializes the horizontal run count HR to 0 in step S1207.

  The above is the processing of step S607 in the second embodiment. Step S607 and subsequent steps are the same as those in the first embodiment. The LH component, HL component, and HH component of the subband are high frequency components, and the coefficient value 0 increases at a high compression ratio. Therefore, run-length encoding is applied to many regions. Therefore, even if the mode determination is performed only by the one-line reference mode determination, the coding efficiency can be maintained. In addition, in the one-line reference mode determination, since it is not necessary to refer to the immediately preceding line, there is no need to hold the coefficient information of the immediately preceding line, and the memory usage can be reduced. Further, since the determination can be made only by updating the horizontal run count HR, it can be realized by a simple calculation.

  According to the second embodiment by the above-described encoding processing, a different run-length mode determination processing is performed on the wavelet coefficients divided into sub-bands according to the sub-bands, thereby maintaining the encoding efficiency and maintaining the memory efficiency. Usage can be reduced. Specifically, the mode determination is performed only for the one-line reference mode determination for the LH component, the HL component, and the HH component that are the high frequency components of the subband. In addition, by using the predictive conversion that does not use the coefficient information of the upper line, it is possible to maintain the coding efficiency with a simple operation and reduce the memory usage.

(Other Examples)
In the embodiment, an example in which the image encoding apparatus is applied to an encoding unit mounted on a camera has been described. However, the target apparatus is not limited to a camera. Regardless, it is applicable.

  The present invention is also realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program and reads the program. This is the process to be performed.

100 camera, 101 operating unit, 102 imaging unit, 103 obtaining unit, 104 image processing unit, 105 encoding unit, 106 output unit, 107 storage unit, 108 I / O, 201 color Component input unit, 202: Wavelet transform unit, 203: Quantization unit, 204: Mode selection unit, 205: First mode determination unit, 206: Second mode determination unit, 207: Predictive transformation selection unit, 208: First prediction Transformer, 209: second predictive converter, 210: third predictor, 211: fourth predictor, 212: Golomb encoder, 213: run-length encoder, 214: code generator, 215 ... Sign output section

Claims (8)

An image encoding device that encodes image data,
Transform means for performing a wavelet transform on the image data to be encoded to generate a plurality of subbands;
A first coding for predictively encoding the coefficient of interest by inputting coefficients constituting the subband in raster scan order and obtaining a predicted value of the coefficient of interest from encoded coefficients located around the coefficient of interest Means,
Second encoding means for inputting the coefficients in the sub-band in the raster scan order, obtaining and encoding a run indicating the number of continuous same coefficients,
A first determination as to whether or not the subband of interest to which the target coefficient belongs is a subband LL, and a second determination to determine the relationship between the values of a plurality of coded coefficients located around the target coefficient. A determination unit for determining whether to encode the coefficient of interest by the first encoding unit or to encode the coefficient of interest by the second encoding unit as a starting point of a run, based on
Image encoding device characterized by Ru with a. The determining means includes:
If the first determination is that the subband of interest is a subband LL and the second determination is that the values of the plurality of coefficients are the same, the second encoding unit It is determined that encoding is performed, and when the second determination determines that the values of the plurality of coefficients are not the same, it is determined that encoding is performed by the first encoding unit.
If the first determination is a case where the subband of interest is a subband other than the subband LL, and if the second determination is that all the values of the plurality of coefficients are 0, the second determination is performed. When the second encoding unit determines that encoding is performed, and when the second determination determines that all of the plurality of coefficient values are not 0, the first encoding unit determines that encoding is performed.
The image encoding device according to claim 1, wherein: The image encoding apparatus according to claim 1, wherein the first encoding unit performs Golomb encoding of the prediction error. 4. The image code according to claim 1, wherein in the second determination, the determination is performed with reference to values of four encoded coefficients located around the coefficient of interest. 5. Device. By computer executes read, the computer program for causing to function as each means included in the apparatus according to any one of claims 1 to 4. A computer-readable storage medium storing the program according to claim 5 . An image encoding method for encoding image data,
A transforming unit that performs a wavelet transform on the image data to be encoded and generates a plurality of subbands;
The first encoding means inputs the coefficients constituting the sub-band in raster scan order, and obtains the predicted value of the coefficient of interest from the coded coefficients located around the coefficient of interest, thereby predicting the coefficient of interest. A first encoding step of encoding;
A second encoding step in which the second encoding means inputs the coefficients in the sub-band in raster scan order, and obtains and encodes a run indicating the number of continuous same coefficients;
Determining means for determining whether or not a subband of interest to which the coefficient of interest belongs is a subband LL and a relationship between values of a plurality of encoded coefficients located around the coefficient of interest; A determination step of determining whether to encode the noted coefficient in the first encoding step or to encode in the second encoding step with the noted coefficient as a starting point of a run based on a second determination When,
An image encoding method comprising: An image encoding device that encodes image data,
Transform means for performing a wavelet transform on the image data to be encoded to generate a plurality of subbands;
A first coding for predictively coding the coefficient of interest by inputting coefficients constituting the subband in raster scan order and obtaining a predicted value of the coefficient of interest from coded coefficients located around the coefficient of interest; Means,
Second encoding means for inputting the coefficients in the sub-band in the raster scan order, obtaining and encoding a run indicating the number of continuous same coefficients,
Based on the type of the subband of interest and the values of a plurality of encoded coefficients located around the target coefficient, the target coefficient may be encoded by the first encoding unit, or the target coefficient may be defined as a starting point of a run. Determination means for determining whether to perform encoding by the second encoding means as
An image encoding device comprising:

Images