JP2012054875A - Image encoding method, image decoding method, image encoder and image decoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the encoding efficiency of an entire image.SOLUTION: An image encoding method includes: a first procedure of separating an encoding object image into at least a first image and a second image corresponding to pixel values of respective pixels of the image; a second procedure of calculating an encoding cost of the separated first image; a third procedure of calculating an encoding cost of the separated second image; a fourth procedure of repeating the first procedure, the second procedure and the third procedure, changing a separation method in the first procedure, and determining the separation method to be used on the basis of the encoding costs of the first image and the encoding costs of the second image calculated for the respective separation methods; a fifth procedure of encoding the separated first image on the basis of the separation method to be used; a sixth procedure of encoding the second image separated on the basis of the separation method to be used; and a seventh procedure of generating an encoded stream on the basis of the encoded first image, the encoded second image, and information on the separation method to be used.

Description

  The present invention relates to an image encoding method and an image decoding method, and more particularly to a technique used for an image photographing apparatus such as a medical image diagnostic apparatus, an image recording apparatus, CT (Computed Tomography), and MRI (Magnetic Resonance Imaging).

  Techniques such as CT and MRI are used in medical diagnostic imaging apparatuses. CT is a technique that uses X-rays or other radiation, and MRI is a technique that uses a nuclear magnetic resonance phenomenon to image a tomogram of the human body. The captured image is recorded as digital data according to a standard system called DICOM (Digital Imaging and Communication in Medicine). Especially in the case of medical images, it is necessary to save the original data of the image. Therefore, the compression method such as the JPEG-LS standard (ISO-14495-1 / ITU-T T.87) is a lossless compression method. Compressed. Since such an image has a wide dynamic range, it is often recorded with a wide bit width of about 16 bits.

  When lossless compression is performed on an image recorded with such a wide bit width, the influence of noise is very large. In general, in a compression method such as JPEG-LS, the compression rate is increased by predicting the pixel value of a pixel of interest using the pixel values of adjacent neighboring pixels. Therefore, when the image to be compressed includes a lot of noise, there is a problem that the compression rate is lowered.

  In order to solve such a problem, a method is disclosed in which an image is separated (divided) into a plurality of bit plane sets, and each separated bit plane set is encoded separately (Patent Documents 1 and 2). reference). In the methods disclosed in Patent Documents 1 and 2, an image is set as a bit plane set on the upper bit side based on a specified fixed cutting level or a cutting level determined only by the encoding condition on the upper bit side. It is separated into a bit plane set on the lower bit side.

Japanese Patent Laid-Open No. 08-322050 JP 2000-244922 A

  However, in the encoding methods disclosed in Patent Documents 1 and 2, even if the separated upper bit side bit plane set and the lower bit side bit plane set are encoded separately, the preferred separation method differs depending on the image. For this reason, there is a problem that the coding efficiency of the entire image is lowered depending on the image. Further, a separation method for efficiently encoding has not been studied.

  An object of the present invention is to provide an image encoding method, an image decoding method, an image encoding device, and an image decoding device that improve the encoding efficiency of the entire image.

  In order to solve the above problems, for example, the configuration described in the claims is adopted.

  The present application includes a plurality of means for solving the above-described problems. To give an example, an image encoding method for encoding an encoding target image, in which the encoding target image is represented by the encoding target image. A first procedure for separating at least a first image and a second image according to a pixel value of each pixel, a second procedure for calculating an encoding cost of the separated first image, and separation The third procedure for calculating the encoding cost of the second image, the first procedure, the second procedure, and the three procedures are repeated by changing the separation method in the first procedure, and for each separation method. And a fourth procedure for determining the separation method to be used based on the encoding cost of the first image and the encoding cost of the second image calculated in the above, and separation based on the separation method to be used. A fifth procedure for encoding the first image; A sixth procedure for encoding the second image separated based on the separation method to be used; the encoded first image; the encoded second image; And a seventh procedure for generating an encoded stream based on the information on the power separation method.

  According to the embodiment of the present invention, the coding efficiency of the entire image can be improved.

  Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments.

It is a block diagram which shows an example of the image coding apparatus of 1st embodiment of this invention. It is a block diagram which shows an example of the image decoding apparatus of 1st embodiment of this invention. It is a flowchart which shows the image coding method of 1st embodiment of this invention. It is a flowchart which shows the image decoding method of 1st embodiment of this invention. It is a block diagram which shows an example of the image coding apparatus of 2nd embodiment of this invention. It is a flowchart which shows the image coding method of 2nd embodiment of this invention. It is a block diagram which shows an example of the image coding apparatus of 3rd embodiment of this invention.

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

(First embodiment)
FIG. 1 is a block diagram showing an example of an image encoding device 100 according to the first embodiment of the present invention. The image coding apparatus 100 includes a pixel value separation unit 101, an image coding cost calculation unit 102, a noise coding cost calculation unit 103, an optimum value adjustment unit 104, an image coding unit 105, a noise coding unit 106, and a stream combination unit. 107.

  In the following, a case where the encoding target image is a monochrome still image such as CT or MRI and this monochrome still image is lossless lossless encoded will be described as an example. However, the present invention is not limited to this case. If the input image is a color image, for example, the same processing may be performed for each of the R, G, and B planes. If the input image is a moving image, the same processing may be performed for each frame of the moving image.

  The pixel value separation unit 101 separates the input image into a plurality of bit plane sets. That is, by separating each pixel value of the input image into a plurality of values at bit boundaries, a plurality of bit plane sets that are held in the format of the image are generated. The pixel value separation unit 101 may generate a plurality of images by separating each pixel value of the input image using a specific pixel value as a threshold value.

  The pixel value separation unit 101 first, based on information on preset separation points (which bits are used as bit boundaries), an image composed of a bit plane set on the upper bit side and bits on the lower bit side Separated into noise information consisting of a set of planes. After the second time, the input image is separated into two bit plane sets based on adjustment separation points set by an optimum value adjustment unit 104 described later. Finally, based on the optimum separation point (separation point to be used) set by the optimum value adjusting unit 104 described later, the input image is separated into two bit plane sets.

  The separation points may be switched in image units, block units, pixel units, or a plurality of these units. Further, the separation point information may be determined in advance according to a certain rule, or may be separately transmitted to the pixel value separation unit 101 as header information.

  The pixel value separation unit 101 sends the separated upper bit side image to the image coding cost calculation unit 102, and sends the separated lower bit side noise information to the noise coding cost calculation unit 103.

  The image coding cost calculation unit 102 calculates the coding cost of the image sent from the pixel value separation unit 101. The encoding cost is an index indicating the encoding efficiency of an image, and is a code amount when an image is actually encoded and a prediction amount of the code amount. It shows that the encoding efficiency of an image is so high that the value of this encoding cost is small. The image encoding cost calculation unit 102 may obtain a code amount when an image is actually encoded. Even if the image is not actually encoded, an approximate code amount may be estimated based on a value such as a prediction error (an error between the input image and the predicted image). For example, when encoding an image using a pixel-by-pixel prediction method such as JPEG-LS, the sum of absolute values of prediction errors (errors of input pixel values and predicted pixel values) occurring in each pixel is calculated. The code amount may be estimated based on the sum of the absolute values of the calculated prediction errors.

  The noise coding cost calculation unit 103 calculates the coding cost of noise information sent from the pixel value separation unit 101. The noise encoding cost calculation unit 103 may obtain a code amount when noise information is actually encoded. Also, an approximate code amount may be estimated based on the appearance probability of each noise value.

  The optimum value adjustment unit 104 encodes the higher-bit side image encoding cost calculated by the image encoding cost calculation unit 102 and the lower-bit side noise information encoding cost calculated by the noise encoding cost calculation unit 103. Based on the above, an optimum separation point (optimum value) to be used by the pixel value separation unit 101 is set.

  Specifically, the optimum value adjustment unit 104 sends information about the separation point (separation point for adjustment) obtained by changing the separation point used by the pixel value separation unit 101 to the pixel value separation unit 101. The pixel value separation unit 101, the image coding cost calculation unit 102, and the noise coding cost calculation unit 103 operate sequentially based on the information on the adjustment separation points, and the codes when the adjustment separation points are used. Calculation cost. The above process is repeated a certain number of times or until the encoding cost optimization is completed. Then, the optimum value adjustment unit 104 is a separation point at which the sum of the encoding cost calculated by the image encoding cost calculation unit 102 and the encoding cost calculated by the noise encoding cost calculation unit 103 is minimized. Is set as the optimum separation point. Information about the set optimum separation point is sent to the pixel value separation unit 101.

  For example, a case will be described in which an image on the upper bit side is encoded using the JPEG-LS method, and noise information on the lower bit side is encoded using a method based on the LZ77 method such as LZH or 7-zip. When the dynamic range of the input image is 12 bits, the upper 8 bits of the image are encoded using the JPEG-LS method compared to the encoding cost when all bits of each pixel of the input image are encoded using JPEG-LS. In some cases, the total encoding cost when the lower 4 bits of noise information is encoded by the LZ77 method is smaller. Therefore, the optimum value adjustment unit 104 adjusts the separation point according to the input image, such as whether the upper 8 bits and the lower 4 bits are separated or whether the upper 9 bits and the lower 3 bits are separated. The optimum value adjustment unit 104 separately transmits information on adjustment separation points to the pixel value separation unit 101 by recording the information in a bit stream. When adjusting the dynamic range of the input image, information related to the adjustment is also transmitted.

  The image encoding unit 105 converts the upper bit side image separated by the pixel value separation unit 101 based on the optimum separation point into a lossless encoding method such as JPEG-LS, lossless JPEG, JPEG2000, or JPEG, JPEG2000, MPEG. And encoding using a lossy encoding method such as H.264 / AVC.

  The image encoding unit 105 may perform encoding for each image, or may perform encoding by performing inter-image prediction using a plurality of images. Further, the image on the upper bit side may be encoded after being shifted to the right by the amount of the separated lower bits, or may be encoded as it is without being shifted to the right. Furthermore, in order to adjust the dynamic range of the input image, the pixel values of all the pixels may be increased or decreased by a certain value. In this case, the image encoding unit 105 separately transmits the increase / decrease value related to the adjustment. The image encoding unit 105 outputs the generated encoded stream to the stream combining unit 107.

  The noise encoding unit 106 uses the low-order bit side noise information separated by the pixel value separation unit 101 based on the optimum separation point as a general dictionary method based on LZ77 methods such as LZH, ZIP, 7-ZIP, etc. Encoding is performed using a so-called encoding method. This type of encoding is used because the noise information on the lower bit side is generally random noise, and the encoding method such as the LZH method is better than the encoding method such as the JPEG-LS method. This is because there are many cases where the compression can be efficiently performed. However, the noise information on the lower bit side may include information correlated with surrounding information. Therefore, the noise encoding unit 106 may use the JPEG-LS method in the same manner as the image encoding unit 105.

  In addition, this noise encoding part 106 is encoded using a bit pack, when the noise information per pixel is smaller than 8 bits. That is, instead of inserting noise information of one pixel into 1 byte (= 8 bits), noise information of a plurality of pixels is inserted into 1 byte. For example, when the noise information per pixel is 4 bits, noise information for 2 pixels is inserted into 1 byte. On the other hand, if the noise information per pixel is 3 bits, the noise information for 2 pixels is inserted into 1 byte and the 2 bits are used as an empty bit, or the noise information for 3 pixels is inserted into 1 byte in order. The last 1 bit of the noise information of the pixel is inserted into another 1 byte. The noise encoding unit 106 outputs the generated encoded stream to the stream combining unit 107.

  The stream combining unit 107 combines the encoded stream of the image generated by the image encoding unit 105 and the encoded stream of noise information generated by the noise encoding unit 106, and outputs the combined stream as one encoded stream. . The stream combining unit 107 multiplexes information on optimum separation points in the header of the output encoded stream.

  With the configuration described above, the image encoding device 100 can efficiently compress losslessly by separating an input image into image information and noise information. Also, the image separation method can be made variable for each input image.

  Note that the series of processing may be executed in units of images, or may be executed in units of a plurality of frames. In other words, the optimum value adjustment unit 104 may set an optimum separation point for each of all the frame images, or may set one common optimum separation point for a plurality of frame images. Also good. In particular, when a series of processes is executed in units of a plurality of frames, an optimal separation point set in a predetermined frame may be applied to the next frame. Thereby, when the change between frames is small, it is possible to save the trouble of calculating the optimum separation point for each frame, and it is not necessary to transmit the optimum separation point for each frame.

  In the above description, the optimum value adjustment unit 104 sets the optimum separation point by calculating the repetitive coding cost. However, the present invention is not limited to this case. For example, an optimal separation point may be set by preparing a plurality of separation point candidates in advance and calculating the coding cost based on each candidate in parallel. In this case, it is necessary to prepare a plurality of separation point candidates in advance, but since it is parallel processing, it can be processed at high speed.

  FIG. 2 is a block diagram illustrating an example of the image decoding device 200 according to the first embodiment of this invention. The image decoding device 200 includes a data separation unit 201, a header analysis unit 202, an image decoding unit 203, a noise decoding unit 204, and an image synthesis unit 205.

  The data separation unit 201 separates the input encoded stream into a header, an encoded stream of an image on the upper bit side, and an encoded stream of noise information on the lower bit side. The separated header, image encoded stream, and noise information encoded stream are sent to the header analysis unit 202, the image decoding unit 203, and the noise decoding unit 204, respectively.

  The header analysis unit 202 analyzes the header sent from the data separation unit 201 to acquire the information on the optimum separation point. The information on the optimum separation point is used for synthesizing the upper bit side image and the lower bit side noise information to restore the original image. The header analysis unit 202 sends the separation point information to the image decoding unit 203, the noise decoding unit 204, and the image synthesis unit 205.

  The image decoding unit 203 decodes the encoded stream of the image sent from the data separation unit 201. The decoding of the image is performed by a predetermined method. For example, when the encoded stream is encoded by the JPEG-LS method, the encoded stream is decoded by the JPEG-LS method. If the bit shift or dynamic range is adjusted before encoding, this adjustment is restored. The image decoding unit 203 uses a decoding method corresponding to the encoding method used by the image encoding unit 105 in FIG.

  The noise decoding unit 204 decodes the encoded stream of noise information sent from the data separation unit 201. The decoding of noise information is performed by a predetermined method. For example, when the encoded stream is encoded by the LZH method, the encoded stream is decoded by the LZH method. In addition, when encoding is performed using a bit pack, it is restored to noise information in units of pixels. The noise decoding unit 204 uses a decoding method corresponding to the encoding method used by the noise encoding unit 106 of FIG.

  The image synthesis unit 205 synthesizes the image decoded by the image decoding unit 203 and the noise information decoded by the noise decoding unit 204 using the information on the optimum separation point sent from the header analysis unit 202. To restore the original image.

  With the configuration described above, the image decoding apparatus 200 can restore the original image based on the encoded stream output from the image encoding apparatus 100 of FIG.

  In the above description, the case where the original image is separated into the upper bit side image and the lower bit side noise information has been described as an example. However, the original image is divided into a general first image and a second image. The same applies to the case where the original image is separated into two, or the original image is separated into the lossy coded image and the lossless coded image.

  FIG. 3 is a flowchart showing an image encoding method according to the first embodiment of the present invention. In the following, an example in which an input original image is separated into a first image (upper bit side image) and a second image (lower bit side noise information) will be described as an example.

  First, in step 501, the pixel value separation unit 101 inputs an original image (501).

  In step 502, the pixel value separation unit 101 separates the original image input in step 501 into a first image and a second image according to a preset separation point (502).

  Next, in step 503, the image coding cost calculation unit 102 calculates the coding cost of the first image (503). Here, for example, the sum of absolute values of prediction errors (errors of input pixel values and predicted pixel values predicted from surrounding pixels) occurring in each pixel of the first image is calculated, and the calculated prediction error is calculated. The code amount may be estimated based on the sum of absolute values. In addition, you may obtain | require the code amount at the time of actually encoding a 1st image. In particular, the processing can be speeded up by using the predicted value. In this case, the encoding method is the JPEG-LS method, but either the lossless method or the lossy method may be used.

  Next, in step 504, the noise encoding cost calculation unit 103 calculates the encoding cost of the second image (504). Here, for example, the code amount may be estimated based on the appearance probability of a bit-packed symbol. In addition, you may obtain | require the code amount at the time of actually encoding a 2nd image. In particular, the processing can be speeded up by using the predicted value. The encoding method in this case is the LZH method, but any of a general encoding method such as an image encoding method and a dictionary method may be used.

  Next, in step 505, the optimum value adjustment unit 104 determines the optimum value based on the encoding cost of the first image calculated in step 503 and the encoding cost of the second image calculated in step 504. It is determined whether a separation point has been acquired (505). Specifically, it is determined whether or not a separation point has been acquired such that the sum of the encoding cost of the first image and the encoding cost of the second image is the minimum value.

  If the optimum separation point has not been acquired (NO in 505), the optimum value adjustment unit 104 uses the pixel value as the information of the separation point (separation point for adjustment) obtained by changing the separation point used in step 502. This is sent to the separation unit 101. Thereafter, returning to step 502, the pixel value separation unit 101 separates the image in accordance with the adjustment separation point set by the optimum value adjustment unit 104. On the other hand, if the optimum separation point has been acquired (YES in 505), the process proceeds to step 506.

  When the process proceeds to step 506, the image encoding unit 105 encodes the first image (506). Here, the first image separated based on the optimum separation point obtained in step 505 is encoded using a lossless encoding method such as JPEG-LS, lossless JPEG, JPEG2000, and the like. An encoded stream is generated.

  Next, in step 507, the noise encoding unit 106 encodes the second image (507). Here, the second image separated based on the optimum separation point obtained in step 505 is converted into an encoding method generally called a dictionary method based on LZ77 methods such as LZH, ZIP, 7-zip. To generate an encoded stream of the second image.

  In step 508, the stream combining unit 107 combines the encoded stream of the first image and the encoded stream of the second image, and outputs one encoded stream (508). Here, information on the optimum separation point is multiplexed in the header of the output encoded stream.

  Through the processing described above, the image encoding device 100 can efficiently compress losslessly by separating the input original image into the first image and the second image optimally. Also, the image separation method can be made variable for each input original image.

  In the above description, the optimum value adjustment unit 104 sets the optimum separation point by calculating the repetitive coding cost. However, the present invention is not limited to this case. For example, an optimal separation point may be set by preparing a plurality of separation point candidates in advance and calculating the coding cost based on each candidate in parallel. In this case, it is necessary to prepare a plurality of separation point candidates in advance, but since it is parallel processing, it can be processed at high speed.

  FIG. 4 is a flowchart showing the image decoding method according to the first embodiment of the present invention.

  First, in step 601, the data separation unit 201 inputs an encoded stream to be decoded (601).

  In step 602, the data separation unit 201 separates the encoded stream input in step 601 into a header, an encoded stream of the first image, and an encoded stream of the second image (602).

  Next, in step 603, the header analysis unit 202 analyzes the header separated in step 602, thereby acquiring the information on the optimum separation point. The information on the optimum separation point is used for synthesizing the first image and the second image and restoring the original image. Further, if necessary, the first image is used for bit shifting or releasing the bit pack of the second image. If the dynamic range has been adjusted before encoding, the header analysis unit 202 restores this adjustment using information stored in the header.

  Next, in step 604, the image decoding unit 203 decodes the encoded stream of the first image separated in step 602 corresponding to the encoding method used when the first image is encoded. Decoding is performed according to the method (604). If the dynamic range is adjusted before encoding, this adjustment is restored.

  Next, in step 605, the noise decoding unit 204 decodes the encoded stream of the second image separated in step 602 corresponding to the encoding method used when the second image is encoded. Decoding is performed according to the method (605). If a bit pack is set when encoding, the image is decoded after releasing the bit pack.

  In step 606, the image composition unit 205 synthesizes and outputs the first image decoded in step 604 and the second image decoded in step 605 (605). Here, for example, when the image is separated into the first image and the second image at a certain separation point and the pixel value of the first image is right-shifted, the decoded first The first image and the second image are synthesized by calculating the sum of the pixel value of the image shifted to the left and the decoded second image.

  Through the processing described above, the image decoding apparatus 200 can restore the original image based on the encoded stream output from the image encoding apparatus 100 in FIG.

(Second embodiment)
FIG. 5 is a block diagram showing an example of an image encoding device according to the second embodiment of the present invention. The image encoding device 300 includes a pixel value separation / correction unit 301, a lossy encoding unit 302, a lossy decoding unit 303, a lossy encoding cost calculation unit 304, a lossless encoding unit 305, a lossless encoding cost calculation unit 306, an optimum value. An adjustment unit 307 and a stream combination unit 308 are provided.

  The pixel value separation / correction unit 301 separates the input image into a plurality of images. That is, first, the input image is separated into a first image consisting of a set of bit planes on the upper bit side and a second image consisting of a set of bit planes on the lower bit side by a predetermined separation method. Then, the separated first image is sent to the lossy encoding unit 302. In the following description, the separation method is described as a method of separating an image based on information on separation points (which bits are used as bit boundaries).

  Further, the second image is corrected (generated) by taking a difference between the input image and a decoded image of the first image sent from the optimum value adjustment unit 307 described later. Then, the corrected second image is sent to the lossless encoding unit 305.

  In this way, by correcting the second image, the sum of the decoded image of the first image and the decoded image of the second image that are finally combined matches the input image. .

  In addition, the pixel value separation / correction unit 301 first separates an input image into two images based on information on preset separation points. From the second time onward, the input image is separated into two images based on adjustment separation points set by an optimum value adjustment unit 307 described later. Finally, the input image is separated into two images based on an optimum separation point (a separation point to be used) set by an optimum value adjusting unit 307 described later.

  The separation points may be switched in image units, block units, pixel units, or a plurality of these units.

  The lossy encoding unit 302 encodes the first image using a lossy encoding method such as JPEG, JPEG2000, MPEG, H.264 / AVC. The lossy encoding unit 302 may encode by one image unit, or may perform encoding by performing inter-image prediction using a plurality of images. Also, encoding may be performed after increasing / decreasing a certain value by the amount of the separated second image, or the image input to the pixel value separation / correction unit 301 may be encoded as the first image. Good. The lossy encoding unit 302 outputs the generated encoded stream of the first image to the lossy decoding unit 303 and the lossy encoding cost calculation unit 304.

  The lossy encoding cost calculation unit 304 calculates the encoding cost of the first image. The lossy encoding cost calculation unit 304 sends the calculated encoding cost of the first image to the optimum value adjustment unit 307. Also, the encoded stream of the first image is sent to the stream combining unit 308.

  The lossy decoding unit 303 decodes the encoded stream of the first image encoded by the lossy encoding unit 302. However, the first image is once lossy-encoded. Therefore, the first image obtained by decoding here does not completely match the first image before lossy encoding. Therefore, the lossy decoding unit 303 sends the decoded first image to the optimum value adjustment unit 307. As a result, the second image can be corrected and an optimum separation point can be set.

  The lossless encoding unit 305 encodes the second image using an encoding method generally referred to as a dictionary method based on the JPEG-LS method or the LZ77 method such as LZH, ZIP, or 7-zip. The lossless encoding unit 305 outputs the generated encoded stream of the second image to the lossless encoding cost calculation unit 306.

  The lossless encoding cost calculation unit 306 calculates the encoding cost of the second image. The lossless encoding cost calculation unit 306 sends the calculated encoding cost to the optimum value adjustment unit 307. Also, the encoded stream of the second image is sent to the stream combining unit 308.

  The optimum value adjustment unit 307 is based on the encoding cost of the first image calculated by the lossy encoding cost calculation unit 304 and the encoding cost of the second image calculated by the lossless encoding cost calculation unit 306. Thus, an optimum separation point (optimum value) to be used by the pixel value separation / correction unit 301 is set.

  Specifically, the optimum value adjustment unit 307 sends information on the separation point (adjustment separation point) obtained by changing the separation point used by the pixel value separation / correction unit 301 to the pixel value separation / correction unit 301. . The pixel value separation / correction unit 301, the lossy encoding unit 302, the lossy decoding unit 303, the lossy encoding cost calculation unit 304, the lossless encoding unit 305, and the lossless encoding cost calculation unit 306 are information on the separation points for adjustment. Are sequentially operated, and the encoding cost when the separation point for adjustment is used is calculated. The above process is repeated a certain number of times or until the encoding cost optimization is completed. Then, the optimum value adjustment unit 307 is a separation point at which the sum of the coding cost calculated by the lossy coding cost calculation unit 304 and the coding cost calculated by the lossless coding cost calculation unit 306 is minimized. Is set as the optimum separation point. Information about the set optimum separation point is sent to the pixel value separation / correction unit 301.

  The stream combining unit 308 combines the encoded stream of the first image received from the lossy encoding unit 302 and the encoded stream of the second image received from the lossless encoding unit 305, and provides one encoded stream. Output as. The stream combining unit 308 multiplexes information on optimum separation points in the header of the output encoded stream.

  With the configuration described above, the image coding apparatus 300 optimally separates an input image into a first image and a second image, and corrects the second image, thereby efficiently compressing losslessly. be able to. Also, the image separation method can be made variable for each input image.

  FIG. 6 is a flowchart showing an image encoding method according to the second embodiment of the present invention. In the following, the input original image is separated into a first image and a second image, the second image is corrected, the first image is lossy-coded, and the second image is lossless-coded. An example will be described. However, both the first image and the second image may be lossless encoded.

  First, in step 701, the pixel value separation / correction unit 301 inputs an original image (701).

  In step 702, the pixel value separation / correction unit 301 separates the original image input in step 701 into a first image and a second image according to a preset separation point (702). ).

  In step 703, the lossy encoding unit 302 encodes the first image (703). Here, the first image separated in step 702 is encoded using a lossy encoding method such as JPEG, JPEG2000, MPEG, H.264 / AVC, and the encoded stream of the first image is generated. The lossy encoding unit 302 measures the code amount in order to calculate the encoding cost of the first image.

  Next, in step 704, the lossy decoding unit 303 decodes the first image (704). Here, the encoded stream of the first image encoded in step 703 is decoded by a decoding method corresponding to the encoding method in step 703. The decoded first image and code amount are used for encoding cost calculation in step 705 described later.

  Next, in step 705, the lossy encoding cost calculation unit 304 calculates the encoding cost of the first image (705). Here, the encoding cost of the first image is calculated based on the code amount measured in step 703 and the error of the first image that has been decoded and the first image that has not been encoded.

  In step 706, the lossless encoding unit 305 encodes the second image (706). Here, the second image modified after being separated in step 702 is generally based on the dictionary method based on the JPEG-LS, lossless JPEG, JPEG2000, LZ77, such as LZH, ZIP, 7-zip. Encoding is performed using a so-called encoding method. Note that the lossless encoding unit 305 measures the code amount in order to calculate the encoding cost of the second image.

  In step 707, the lossless coding cost calculation unit 306 calculates the coding cost of the second image (707). Here, the encoding cost of the second image is calculated based on the code amount measured in step 706.

  Next, in step 708, the optimum value adjustment unit 307 determines the optimum value based on the encoding cost of the first image calculated in step 705 and the encoding cost of the second image calculated in step 707. It is determined whether a separation point has been acquired (708). Specifically, under the condition that the sum of the decoded image of the first image and the decoded image of the second image to be combined finally matches the original image, the encoding cost of the first image, It is determined whether or not a separation point has been acquired such that the sum of the second image and the coding cost is a minimum value.

  When the optimum separation point has not been acquired (NO in 708), the optimum value adjustment unit 307 uses the pixel value as the information on the separation point (separation point for adjustment) obtained by changing the separation point used in step 702. The data is sent to the separation / correction unit 301. Thereafter, returning to step 702, the pixel value separation / correction unit 301 separates the original image in accordance with the adjustment separation points set by the optimum value adjustment unit 307. On the other hand, if the optimum separation point has been acquired (YES in 708), the process proceeds to step 709.

  Next, in step 709, the stream combining unit 308 combines the encoded stream of the first image and the encoded stream of the second image, and outputs one encoded stream (709). Here, information on the optimum separation point is multiplexed in the header of the output encoded stream.

  Through the processing described above, the image encoding device 300 optimally separates the input original image into the first image and the second image, and corrects the second image to efficiently compress the losslessly. be able to. Also, the image separation method can be made variable for each input original image.

(Third embodiment)
FIG. 7 is a block diagram showing an example of an image encoding device according to the third embodiment of the present invention. The image encoding device 400 includes a pixel value separation unit 401, a first image encoding cost calculation unit 402, a second image encoding cost calculation unit 403, a third image encoding cost calculation unit 404, and an optimum value adjustment. Unit 405, first image encoding unit 406, second image encoding unit 407, third image encoding unit 408, and stream combination unit 409.

  The pixel value separation unit 401 separates the input image into three or more images (here, the first image, the second image, and the third image). For example, each pixel value of the input image is separated into three or more values at bit boundaries, thereby generating a set of three or more bit planes held in the image format. Alternatively, for example, three or more images may be generated by separating each pixel value of the input image using two or more specific pixel values as threshold values.

  The pixel value separation unit 401 initially includes a first image including a low-frequency component, a second image including a high-frequency component (excluding noise), noise based on information on a separation point set in advance. To a third image containing. From the second time onward, the input image is separated into three images based on information on adjustment separation points set by an optimum value adjustment unit 405 described later. Finally, the input image is separated into three images based on the information of the optimum separation point set by the optimum value adjustment unit 405 described later. Note that the separation point may be switched in image units, block units, or pixel units.

  The pixel value separation unit 401 sends the separated first image to the first image encoding cost calculation unit 402. Similarly, the separated second image and third image are sent to the second image encoding cost calculation unit 403 and the third image encoding cost calculation unit 404, respectively.

  The first image encoding cost calculation unit 402 calculates the encoding cost of the first image sent from the pixel value separation unit 401. Since the first image encoding cost calculation unit 402 is the same as the image encoding cost calculation unit 102 of FIG. 1, description thereof is omitted here.

  The second image encoding cost calculation unit 403 calculates the encoding cost of the second image sent from the pixel value separation unit 401. The second image coding cost calculation unit 403 is the same as the noise coding cost calculation unit 103 in FIG.

  The third image encoding cost calculation unit 404 calculates the encoding cost of the third image sent from the pixel value separation unit 401. Since the third image encoding cost calculation unit 404 is the same as the noise encoding cost calculation unit 103 in FIG. 1, the description thereof is omitted here.

  The optimum value adjustment unit 405 encodes the first image calculated by each of the first image encoding cost calculation unit 402, the second image encoding cost calculation unit 403, and the third image encoding cost calculation unit 404. Based on the cost, the encoding cost of the second image, and the encoding cost of the third image, an optimal separation point (optimum value) to be used by the pixel value separation unit 401 is set.

  Specifically, the optimum value adjustment unit 405 sends information on the separation point (separation point for adjustment) obtained by changing the separation point used by the pixel value separation unit 401 to the pixel value separation unit 401. The pixel value separation unit 401, the first image encoding cost calculation unit 402, the second image encoding cost calculation unit 403, and the third image encoding cost calculation unit 404 operate sequentially based on the information on the separation points for adjustment. Then, the encoding cost when the separation point for adjustment is used is calculated. The above process is repeated a certain number of times or until the encoding cost optimization is completed. Then, the optimum value adjustment unit 405 determines a separation point that minimizes the sum of the encoding cost of the first image, the encoding cost of the second image, and the encoding cost of the third image, Set as the optimal separation point. Information about the set optimum separation point is sent to the pixel value separation unit 401.

  The first image encoding unit 406 converts the first image separated by the pixel value separation unit 401 based on the optimum separation point into a lossless encoding method such as JPEG-LS, lossless JPEG, JPEG2000, or JPEG, JPEG2000, MPEG. And encoding using a lossy encoding method such as H.264 / AVC.

  The second image encoding unit 407 generally uses a dictionary method based on the LZ77 method such as LZH, ZIP, and 7-zip for the second image separated by the pixel value separation unit 401 based on the optimum separation point. Encoding is performed using a so-called encoding method.

  The third image encoding unit 408 generally uses a dictionary method based on the LZ77 method such as LZH, ZIP, 7-zip, and the like for the third image separated by the pixel value separation unit 401 based on the optimum separation point. Encoding is performed using a so-called encoding method.

  The stream combining unit 409 includes an encoded stream of the first image generated by the first image encoding unit 406, an encoded stream of the second image generated by the second image encoding unit 407, and a third The encoded stream of the third image generated by the image encoding unit 408 is combined and output as one encoded stream. The stream combining unit 409 multiplexes information on optimum separation points in the header of the output encoded stream.

  With the configuration described above, the image encoding apparatus 400 can efficiently compress losslessly by separating the input image into the first image, the second image, and the third image optimally. . Also, the image separation method can be made variable for each input image.

  Each embodiment of the present invention has been described above. However, each of the above embodiments shows one application example of the present invention, and the technical scope of the present invention is limited to the specific configuration of each of the above embodiments. It is not the purpose. Various modifications can be made without departing from the scope of the present invention.

  Further, by applying the technology relating to the image encoding method and the image decoding method of the present invention to an image recording apparatus and an image transmission apparatus, a medical image diagnostic apparatus capable of recording a large volume of images and transmitting an image with a small bandwidth. , CT, MRI, and other imaging devices can be provided.

100, 300, 400 Image coding apparatus 101, 401 Pixel value separation unit 102 Image coding cost calculation unit 103 Noise coding cost calculation unit 104, 307, 405 Optimal value adjustment unit 105 Image coding unit 106 Noise coding unit 107 308, 409 Stream combination unit 200 Image decoding device 201 Data separation unit 202 Header analysis unit 203 Image decoding unit 204 Noise decoding unit 205 Image synthesis unit 301 Pixel value separation / correction unit 302 Lossy encoding unit 303 Lossy decoding unit 304 Lossy code Cost calculating unit 305 lossless encoding unit 306 lossless encoding cost calculating unit 402 first image encoding cost calculating unit 403 second image encoding cost calculating unit 404 third image encoding cost calculating unit 406 first image encoding unit 407 Second image encoding unit 408 Third image encoding unit

Claims (12)

An image encoding method for encoding an image to be encoded,
A first procedure for separating the encoding target image into at least a first image and a second image according to a pixel value of each pixel of the encoding target image;
A second procedure for calculating the coding cost of the separated first image;
A third procedure for calculating the encoding cost of the separated second image;
The first procedure, the second procedure, and the three procedures are repeated by changing the separation method in the first procedure, and the encoding cost of the first image and the second image calculated for each separation method. A fourth procedure for determining the separation method to be used based on the encoding cost of:
A fifth procedure for encoding the first image separated based on the separation method to be used;
A sixth procedure for encoding the second image separated based on the separation method to be used;
A seventh procedure for generating an encoded stream based on the encoded first image, the encoded second image, and the information of the separation method to be used;
An image encoding method comprising: In the fifth procedure, the first image is encoded by a lossless encoding method using a pixel value prediction method,
The image encoding method according to claim 1, wherein in the sixth procedure, the second image is encoded by a lossless encoding method using a dictionary method.   In the fourth procedure, a separation method that minimizes the sum of the calculated encoding cost of the first image and the encoding cost of the second image is determined as the separation method to be used. The image encoding method according to claim 1, wherein: An image encoding method for encoding an image to be encoded,
A first procedure for separating the encoding target image into a first image and a second image according to a pixel value of each pixel of the encoding target image;
A second procedure for calculating the encoding cost of the first image when the separated first image is encoded by the lossy encoding method;
A third procedure for decoding the separated first image by a decoding scheme corresponding to the lossy encoding scheme;
A fourth procedure for generating the modified second image based on a difference between the encoding target image and the decoded first image;
A fifth procedure for calculating the encoding cost of the second image when the modified second image is encoded by the lossless encoding method;
The first procedure, the second procedure, the third procedure, the fourth procedure, and the fifth procedure are repeated by changing the separation method in the first procedure, and the first procedure is calculated for each separation method. A sixth procedure for determining the separation method to be used based on the encoding cost of the image and the encoding cost of the second image;
A seventh procedure for encoding the first image separated based on the separation method to be used by the lossy coding method;
An eighth procedure for encoding the second image generated based on the difference between the encoding target image and the first image encoded in the seventh procedure by the lossless encoding method;
A ninth procedure for generating an encoded stream based on the encoded first image, the encoded second image, and information on the separation method to be used;
An image encoding method comprising:   In the sixth procedure, a separation method that minimizes the sum of the calculated encoding cost of the first image and the encoding cost of the second image is determined as the separation method to be used. The image encoding method according to claim 4. An image decoding method for decoding an encoded stream,
The encoded stream includes an encoded stream of the first image when a predetermined image is separated into at least a first image and a second image according to a pixel value of each pixel of the image. An encoded stream of the second image, and a header including information on a separation method used in the separation,
The method
Separating the encoded stream into an encoded stream of the first image, an encoded stream of the second image, and the header;
A procedure for obtaining information on the separation method based on the separated header;
A procedure for decoding the encoded data of the separated first image;
A procedure for decoding the encoded data of the separated second image;
A procedure for synthesizing the decoded first image and the decoded second image based on the acquired information on the separation method;
An image decoding method comprising: An image encoding device for encoding an image to be encoded,
A pixel value separation unit that separates the encoding target image into at least a first image and a second image according to the pixel value of each pixel of the encoding target image;
A first encoding cost calculation unit for calculating the encoding cost of the separated first image;
A second encoding cost calculation unit for calculating the encoding cost of the separated second image;
The process of the pixel value separation unit, the first encoding cost calculation unit, and the second encoding cost calculation unit is repeated for each separation method by changing the separation method used by the pixel value separation unit. An optimal value adjusting unit that determines a separation method to be used based on the encoding cost of the first image and the encoding cost of the second image;
A first image encoding unit for encoding the first image separated based on the separation method to be used;
A second image encoding unit for encoding the second image separated based on the separation method to be used;
A stream combining unit that generates encoded data based on the encoded first image, the encoded second image, and information on the separation method to be used;
An image encoding apparatus comprising: The first image encoding unit encodes the first image by a lossless encoding method using a pixel value prediction method,
The image coding apparatus according to claim 7, wherein the second image coding unit codes the second image by a lossless coding method using a dictionary method.   The optimum value adjustment unit determines, as the separation method to be used, the separation method that minimizes the sum of the coding cost of the first image and the coding cost of the second image. The image encoding device according to claim 7. An image encoding device for encoding an image to be encoded,
A pixel value separation unit that separates the encoding target image into a first image and a second image according to a pixel value of each pixel of the encoding target image;
A lossy encoding cost calculation unit for calculating the encoding cost of the first image when the separated first image is encoded by the lossy encoding method;
A lossy decoding unit for decoding the separated first image by a decoding method corresponding to the lossy encoding method;
A pixel value correcting unit that generates the corrected second image based on the difference between the encoding target image and the decoded first image;
A lossless encoding cost calculation unit for calculating the encoding cost of the second image when the modified second image is encoded by a lossless encoding method;
The pixel value separation unit, the lossy encoding cost calculation unit, the lossy decoding unit, the pixel value correction unit, and the lossless encoding cost calculation unit are repeatedly processed by changing the separation method used by the pixel value separation unit. An optimum value adjusting unit for determining a separation method to be used based on the coding cost of the first image and the coding cost of the second image calculated for each separation method;
A lossy encoding unit that encodes the first image separated based on the separation method to be used by the lossy encoding method;
A lossless encoding unit that encodes the second image generated based on the difference between the encoding target image and the first image encoded by the lossy encoding unit by the lossless encoding method;
A stream combining unit that generates an encoded stream based on the encoded first image, the encoded second image, and information on the separation method to be used;
An image encoding apparatus comprising:   The optimum value adjustment unit determines, as the separation method to be used, the separation method that minimizes the sum of the coding cost of the first image and the coding cost of the second image. The image encoding device according to claim 10. An image decoding device for decoding an encoded stream,
The encoded stream includes an encoded stream of the first image when a predetermined image is separated into at least a first image and a second image according to a pixel value of each pixel of the image. An encoded stream of the second image, and a header including information on a separation method used in the separation,
The image decoding device
A data separator that separates the encoded stream into an encoded stream of the first image, an encoded stream of the second image, and the header;
Based on the separated header, a header analysis unit that acquires information on the separation method;
A first image decoding unit for decoding the encoded data of the separated first image;
A second image decoding unit for decoding the encoded data of the separated second image;
An image synthesizing unit that synthesizes the decoded first image and the decoded second image based on the acquired information on the separation method;
An image decoding apparatus comprising:

Images