JP2007088565A - Apparatus and method for processing image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce deterioration in image quality, and to process image data with a high compression ratio. <P>SOLUTION: First and second estimation sections 270-1, 270-2 estimate a pixel value for each pixel by using mutually different estimation methods. An estimation error calculation section 272 calculates respective estimation errors at the first and second estimation sections 270-1, 270-2. A determination section 274 determines whether the pixel of which the pixel value is estimated is the pixel for composing a CG image according to whether respective estimation errors at the first and second estimation sections 270-1, 270-2 are 0. An ROI information creation section 262 creates ROI information so that the pixel for composing the CG image belongs to ROI according to the determined results input from the determination section 274. A JPEG 2000 coding section 266 creates an ROI mask for reflecting the ROI information, and generates a code stream having ROI from the image data so that decoding can be made by a decoder conforming to JPEG2000. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing method for encoding or decoding an image.

  As an image processing device that efficiently transmits or stores image data, the image data is encoded by switching between a lossless encoder and an irreversible encoder according to the image data in order to reduce image quality degradation. It has been known. In this type of image processing apparatus, it is known to perform encoding by switching between the reversible mode and the irreversible mode in units of pixels (see Patent Document 1).

JP 2001-257888 A

  However, in the above conventional example, in order to improve the processing speed, DPCM encoding is adopted for lossy encoding, and a pixel value dictionary that captures RGB of one pixel as one color for lossless encoding. Since it is used, there is a problem that the compression rate cannot be increased.

  SUMMARY An advantage of some aspects of the invention is that it provides an image processing apparatus and an image processing method capable of processing image data at a high compression rate while reducing deterioration in image quality.

  In order to achieve the above object, the first feature of the present invention is that binarization means for binarizing image data for each pixel based on the correlation between the pixel of interest and surrounding pixels, and image data Embedded coding means for reversibly embedding the image and a part of the image data to be embedded by the embedded coding means based on the result of the binarization means binarizing the image data. And an image processing apparatus having control means for controlling. Therefore, the image data can be reversibly or irreversibly embedded based on the correlation between the target pixel and the surrounding pixels, so that the image quality can be reduced with the compression rate based on the correlation between the target pixel and the surrounding pixels. The image data can be compressed while being reduced.

  Preferably, the binarization unit binarizes the image data based on whether or not a pixel having the same pixel value exists around the pixel of interest. Preferably, the binarization means includes at least one prediction means for predicting a pixel value with respect to the target pixel, and a prediction error calculation means for calculating a prediction error of the pixel value predicted by the prediction means. The image data is binarized based on whether at least one of the prediction errors calculated by the prediction error calculation means is zero. For example, since CG (Computer Graphic) images often have the same pixel value in adjacent pixels, the prediction error based on the correlation between the pixel of interest and surrounding pixels is often zero. Therefore, the binarization unit can create, for each pixel, data that classifies pixels constituting a natural image, a scanned image, and the like, and pixels constituting a CG image, depending on the type of image data. The image data can be processed at a high compression rate while reducing the deterioration of the image quality. In addition, a CG image often has a higher compression rate when it is losslessly encoded than when it is lossyly encoded.

  The second feature of the present invention is that a plurality of prediction means for predicting pixel values with respect to the target pixel by different prediction methods, and prediction errors of pixel values predicted by the prediction means are calculated. Based on the prediction error calculation means, the prediction error calculated by the prediction error calculation means, the binarization means for binarizing the image data for each pixel, and the prediction error calculated by the prediction error calculation means is zero. Embedded image encoding means for reversibly and separably embedding any one of the information for specifying the prediction means that predicted the pixel value or the pixel value of the target pixel for each pixel, and the binarization means And an image processing apparatus having control means for controlling the embedded encoding means to discard a part of the data to be embedded based on the binarized result.In other words, the embedded encoding means may reduce the amount of information when encoding the information specifying the prediction means that predicted the pixel value for which the prediction error is 0, compared to when encoding the pixel value of the target pixel. Therefore, the compression rate can be further improved.

  Preferably, the embedded coding means uses the unique number assigned to each of the prediction means or the rank based on the frequency of predicting the pixel value with a prediction error of 0 as information for specifying the prediction means. .

  Preferably, the embedded encoding means embeds data by generating a code stream compliant with JPEG2000, and the binarizing means determines whether the ROI is based on the correlation between the pixel of interest and surrounding pixels. Image data is binarized by creating a mask. That is, when the embedded encoding means generates a code stream compliant with JPEG2000, image data can be binarized for each pixel using an ROI compliant with JPEG2000, and it is necessary to provide a unique binarization means. Absent.

  Preferably, the embedded encoding unit shifts information for specifying the prediction unit to bits higher than the MSB of the pixel value of the pixel of interest to generate a code stream compliant with JPEG2000. Therefore, it is possible to easily separate the pixel value of the target pixel from the information for specifying the prediction means.

  Preferably, the embedded encoding unit generates information on the JPEG2000 in accordance with information specifying the prediction unit as a pixel value.

  Preferably, the image processing apparatus further includes an average value acquisition unit that acquires an average value of pixel values of surrounding pixels, and the embedded encoding unit has a unique number assigned to each of the prediction units or a prediction error. A code stream compliant with JPEG2000 is generated by regarding the value obtained by adding the rank based on the frequency of predicting the pixel value to be 0 and the average value acquired by the average value acquisition unit as the pixel value.

  In addition, preferably, it further includes an average value rounding means for rounding the average value obtained by the average value obtaining means, and the embedded encoding means has a unique number assigned to each of the prediction means or a prediction error. A code stream compliant with JPEG2000 is generated by regarding a value obtained by adding the ranking based on the frequency of predicting a pixel value of 0 and the average value rounded by the average value rounding means as a pixel value.

  Preferably, when the binarized result of the binarizing means is isolated, the control means controls to invert the binarized result for the isolated pixel.

  Preferably, when the binarization result obtained by the binarization unit is isolated in units of 2 × 2 pixel blocks, the control unit binarizes the isolated pixel block. Control to correct. That is, the compression rate can be further improved.

  Preferably, the control unit performs control so that the binarized result is inverted with respect to pixels around pixels having the same binarized result by the binarizing unit. Therefore, since the pixels around the pixel to be encoded reversibly can also be encoded reversibly, the image quality of the pixels around the pixel to be encoded reversibly can be prevented.

  Preferably, the control unit controls the embedded encoding unit so that a pixel value predicted to have a prediction error of 0 is encoded reversibly. That is, it is possible to determine the pixel to be encoded reversibly based on the prediction error.

  The third feature of the present invention is that an embedded decoding means for decoding a code stream in which information or pixel values based on a prediction error is embedded for each pixel, and the embedded decoding means Classification means for classifying the decoded data into information and pixel values based on prediction errors, and based on the classification result of the classification means, image data is reproduced from the decoded data decoded by the embedded decoding means An image processing apparatus having image data reproducing means. Therefore, even if information or pixel values based on prediction errors are mixed for each pixel in a code stream that has been encoded, image data can be reproduced, and image quality can be reduced at a high compression rate while reducing image quality degradation. Data can be processed.

  Preferably, the embedded decoding means conforms to JPEG2000, and the classification means converts the decoded data into information based on a prediction error and a pixel value in accordance with ROI encoded information included in the code stream. Classify into: That is, the decoded data can be classified into information based on the prediction error and the pixel value by a decoder having an ROI decoding unit based on JPEG2000.

  The fourth feature of the present invention is that the image data is binarized for each pixel based on the correlation between the pixel of interest and surrounding pixels, and the image data is reversibly embedded and encoded. There is an image processing method in which a part of image data is discarded based on a binarized result.

  According to the present invention, it is possible to process image data at a high compression rate while reducing deterioration in image quality.

Next, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 shows an outline of an image processing apparatus 1 according to an embodiment of the present invention. The image processing apparatus 1 includes a user interface device (UI device) 10 including a display device and a keyboard, a storage device 12 such as an HDD / CD device, a printing device 14, a communication device 16, and a control device 2. The control device 2 includes a CPU 20 and a memory 22 and controls each unit constituting the image processing device 1.
In other words, the image processing apparatus 1 includes a function as a computer, and processes image data received via the storage medium 120 or the communication device 16 in accordance with, for example, an image encoding condition input via the UI apparatus 10. In addition, an image can be output to the display device of the UI device 10 or the printing device 14.

FIG. 2 is a block diagram illustrating a configuration of the code processing unit 24 included in the control device 2 and configured to encode (compress) and decode (decompress) image data.
As shown in FIG. 2, the code processing unit 24 includes an encoder 26 and a decoder 28. The encoder 26 receives, encodes, for example, image data, and outputs it as a code stream to the decoder 28. The decoder 28 complies with, for example, JPEG2000, acquires the code stream output from the encoder 26, decodes it, and outputs reproduced image data. The encoder 26 may output the code stream to the memory 22 (or the storage device 12) or the like, and the decoder 28 may acquire the code stream from the memory 22 (or the storage device 12) or the like.
The code processing unit 24 may be configured by software (program) executed by the CPU 20 with the encoder 26 and the decoder 28, for example.

Next, details of the encoder 26 will be described.
FIG. 3 is a block diagram showing a configuration of the encoder 26. As shown in FIG. 3, the encoder 26 includes a reversible pixel selection unit 260, a ROI (Region of interest) information creation unit 262, a control unit 264, and a JPEG2000 encoding unit 266.

  The reversible pixel selection unit 260 includes a first prediction unit 270-1, a second prediction unit 270-2, a prediction error calculation unit 272, and a determination unit 274. The first prediction unit 270-1 and the second prediction unit 270-2 accept image data, predict pixel values for each pixel using different prediction methods, and calculate the predicted pixel values for the prediction error calculation unit 272. Output. The method by which the first prediction unit 270-1 and the second prediction unit 270-2 predict pixel values is any one of, for example, previous value prediction, matrix prediction, average prediction, and planar prediction.

  The prediction error calculation unit 272 receives the pixel value predicted by the first prediction unit 270-1, the pixel value predicted by the second prediction unit 270-2, and the image data for each pixel, and the first prediction unit 270-1 predicts the pixel value. The difference (prediction error) between the calculated pixel value and the actual pixel value and the difference (prediction error) between the pixel value predicted by the second prediction unit 270-2 and the actual pixel value are calculated, and the determination unit 274 Respectively.

The determination unit 274 accepts the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2 from the prediction error calculation unit 272, for example, depending on whether each prediction error is 0, the pixel value For example, it is determined whether the predicted pixel is a pixel constituting a CG image, and the determination result is output to the ROI information creating unit 262.
For example, in the case where the prediction error of at least one of the first prediction unit 270-1 and the second prediction unit 270-2 is 0, the determination unit 274 is a pixel in which a pixel value is predicted is a CG image. It is determined that there is, and 1 is output. The determination unit 274 determines that the pixel whose pixel value is predicted is not a pixel constituting the CG image when the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2 are not 0, 0 is output.

  The ROI information creation unit 262 creates ROI information so that the JPEG2000 encoding unit 266 can create the ROI mask so that the pixels constituting the CG image belong to the ROI according to the determination result input from the determination unit 274. And output to the control unit 264 and the JPEG2000 encoding unit 266. Note that the ROI information creation unit 262 may create an ROI mask and output it to the control unit 264 and the JPEG2000 encoding unit 266.

  The control unit 264 controls the JPEG2000 encoding unit 266 according to the ROI information input from the ROI information creation unit 262.

The JPEG2000 encoding unit 266 has an integer type 5/3 lossless filter, and reflects the ROI information input from the ROI information creation unit 262 in accordance with the control of the control unit 264 that acquired the ROI information. Is generated, and a code stream having ROI is generated and output from the input image data so that it can be decoded by a decoder compliant with JPEG2000.
For example, under the control of the control unit 264, the JPEG2000 encoding unit 266 reversibly encodes pixels belonging to the ROI using a 5/3 reversible filter or the like, and reversibly encodes pixels not belonging to the ROI using a 5/3 reversible filter or the like. During the encoding, the encoding is stopped and a code stream is generated. That is, the JPEG2000 encoding unit 266 reversibly encodes pixels belonging to ROI (pixels constituting a CG image) and irreversibly encodes pixels not belonging to ROI (pixels not constituting a CG image). Truncate part of image data.
Also, the JPEG2000 encoding unit 266 may perform encoding by switching between lossless encoding and lossy encoding for each pixel according to the ROI information.

Next, a process in which the encoder 26 encodes image data will be described.
FIG. 4 is a flowchart illustrating a process (S10) in which the encoder 26 encodes image data.
As shown in FIG. 4, in step 100 (S100), the first prediction unit 270-1 and the second prediction unit 270-2 accept image data and predict pixel values for each pixel using different prediction methods.

  In step 102 (S102), the prediction error calculation unit 272 calculates the prediction error of each of the first prediction unit 270-1 and the second prediction unit 270-2.

  In step 104 (S104), the determination unit 274 accepts each prediction error of the first prediction unit 270-1 and the second prediction unit 270-2, and determines the pixel value depending on whether each prediction error is 0 or not. It is determined whether or not the predicted pixel is a pixel constituting the CG image.

  In step 106 (S106), the ROI information creation unit 262 can create the ROI mask according to the determination result input from the determination unit 274 so that the pixels constituting the CG image belong to the ROI. The ROI information is created as follows.

  In step 108 (S108), the JPEG2000 encoding unit 266 generates an ROI mask reflecting the ROI information input from the ROI information generation unit 262, and inputs the ROI mask so that it can be decoded by a decoder compliant with JPEG2000. A code stream having ROI is generated from the image data and output.

  Thus, since the encoder 26 reversibly encodes the pixels belonging to the ROI and generates the code stream in which the pixels not belonging to the ROI are irreversibly encoded, the decoder 28 (FIG. 2) compliant with JPEG2000 It is possible to decode a pixel belonging to the ROI encoded without deterioration and a pixel not belonging to the ROI encoded at a high compression rate.

Next, a modification of the encoder 26 (encoder 26a) will be described.
FIG. 5 is a block diagram showing a configuration of the encoder 26a. As shown in FIG. 5, the encoder 26a includes a reversible pixel selection unit 260a, an ROI information creation unit 262, a control unit 264a, a specific information generation unit 276, a specific information encoding unit 277, a data processing unit 278, and a JPEG2000 encoding unit. 266.
In the encoder 26a, components that are substantially the same as the parts constituting the encoder 26 shown in FIG.

  The reversible pixel selection unit 260a includes a first prediction unit 270-1, a second prediction unit 270-2, a prediction error calculation unit 272a, and a determination unit 274. The prediction error calculation unit 272a accepts the pixel value predicted by the first prediction unit 270-1, the pixel value predicted by the second prediction unit 270-2, and the image data for each pixel, and the first prediction unit 270-1 performs prediction. The difference (prediction error) between the calculated pixel value and the actual pixel value and the difference (prediction error) between the pixel value predicted by the second prediction unit 270-2 and the actual pixel value are calculated, and the specific information generation unit H.276 and the determination unit 274, respectively.

  The control unit 264a controls the specific information generation unit 276, the specific information encoding unit 277, the data processing unit 278, and the JPEG2000 encoding unit 266 according to the ROI information input from the ROI information creation unit 262.

The specific information generation unit 276 receives the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2 from the prediction error calculation unit 272a, for example, the first prediction unit 270-1 having a prediction error of 0 and Information (specific information) specifying the second prediction unit 270-2 or any one of these is generated and output to the specific information encoding unit 277.
For example, the specific information generation unit 276 includes information for specifying the first prediction unit 270-1 and the second prediction unit 270-2 or any one of them, for example, the first prediction unit 270-1 and the second prediction unit. A specific number assigned to each of the 270-2 or a pixel value with a prediction error of 0 is set as a rank based on the predicted frequency, and is output to the specific information encoding unit 277.

  The specific information encoding unit 277 encodes the specific information input from the specific information generation unit 276 and outputs the code to the data processing unit 278.

  The data processing unit 278 generates input image data or data in which the code input from the specific information encoding unit 277 is arranged for each pixel in accordance with the control of the control unit 264a that acquired the ROI information. The data is output to the encoding unit 266. For example, the data processing unit 278 accepts the image data and the code output from the specific information encoding unit 277, the code regarded as the pixel value for the pixel determined to constitute the CG image (the pixel to be losslessly encoded), the CG Data (encoding target data) is processed in accordance with control of the control unit 264a so that pixel values for pixels determined not to constitute an image (pixels to be irreversibly encoded) are mixed according to ROI information. .

For example, the data processing unit 278 accepts image data with a bit width indicating a pixel value of M bits, outputs a pixel value to a pixel determined not to form a CG image, and forms a CG image. For the pixel to be determined, a code regarded as a pixel value may be output by shifting the MSB side by M + G bits. Here, G is the number of guard bits for preventing an overflow of the pixel value, and takes a value such as 1 or 2, for example.
Thus, by shifting the code regarded as the pixel value by M + G bits, it is possible to easily classify the code and the pixel value regarded as the pixel value when decoding.

  In addition, the data processing unit 278 may add the center value of the pixel value (128 if M = 8) to the code regarded as the pixel value. However, it is assumed that the bit width of the code regarded as the pixel value is less than M-1. Further, the sign regarded as the pixel value takes a positive or negative value centered on 0.

  Further, the data processing unit 278 adds an average value of the image data or a value obtained by rounding the average value of the pixel values of the surrounding pixels of the target pixel to the code regarded as the pixel value, and performs irreversible encoding. However, the value may not be changed. However, when adding rounded values, the bit width of the addition result should not be larger than the bit width of the image data. For example, the average value of the pixel values is rounded by a value larger than the bit width of the code regarded as the pixel value. Further, the sign regarded as the pixel value takes a positive or negative value centered on 0.

Next, a process in which the encoder 26a encodes image data will be described.
FIG. 6 is a flowchart illustrating a process (S20) in which the encoder 26a encodes image data.
As shown in FIG. 6, in step 200 (S200), the first prediction unit 270-1 and the second prediction unit 270-2 accept image data and predict pixel values for each pixel using different prediction methods.

  In step 202 (S202), the prediction error calculation unit 272a calculates the prediction error of each of the first prediction unit 270-1 and the second prediction unit 270-2.

  In step 204 (S204), the determination unit 274 accepts the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2, and determines the pixel value according to whether each of the prediction errors is 0 or not. It is determined whether or not the predicted pixel is a pixel constituting the CG image.

  In step 206 (S206), the ROI information creation unit 262 can create the ROI mask according to the determination result input from the determination unit 274 such that the pixels constituting the CG image belong to the ROI. The ROI information is created as follows.

  In step 208 (S208), the specific information generation unit 276 receives the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2 from the prediction error calculation unit 272a, and the first prediction error is 0. Information (specific information) for specifying the prediction unit 270-1 and the second prediction unit 270-2 or any one of them is generated.

  In step 210 (S210), the specific information encoding unit 277 encodes the specific information input from the specific information generation unit 276.

  In step 212 (S212), the data processing unit 278 accepts the image data and the code output from the specific information encoding unit 277, so that the code regarded as the pixel value and the pixel value included in the image data are mixed. Data (encoding target data) is processed under the control of the control unit 264a.

  In step 214 (S214), the JPEG2000 encoding unit 266 creates a ROI mask that reflects the ROI information input from the ROI information creation unit 262, and allows the data processing unit to decode it with a decoder compliant with JPEG2000. A code stream having an ROI is generated from the data input from 278 and output.

Next, a decoder 28a capable of decoding the code stream generated by the encoder 26a and generating reproduced image data will be described.
The decoder 28a is a modification of the decoder 28 and is included in the code processing unit 24 together with the encoder 26a, for example.

FIG. 7 is a block diagram showing the configuration of the decoder 28a. As shown in FIG. 7, the decoder 28a includes a JPEG2000 decoding unit 280, a data classification unit 282, a specific information decoding unit 284, a selection unit 286, a first prediction unit 270-1, a second prediction unit 270-2, and The image data assembly unit 288 is configured.
It should be noted that in the decoder 28a, the same reference numerals are given to substantially the same parts as those constituting the encoder 26a shown in FIG.

The JPEG 2000 decoding unit 280 is configured to be able to decode the code stream generated by the encoder 26a in accordance with JPEG 2000. When the JPEG 2000 decoding unit 280 accepts the code stream, the decoded data and ROI information obtained by decoding the code stream are received. (Or ROI mask) is output to the data classification unit 282.
Note that when the code stream includes an M + G-bit shifted code that is regarded as the pixel value described above, the JPEG2000 decoding unit 280 is configured not to output the ROI information (or the ROI mask).

  The data classification unit 282 receives the decoded data obtained by decoding the code stream from the JPEG2000 decoding unit 280 and the ROI information (or ROI mask), and the code or pixel value (pixel value) regarding the decoded data as a pixel value for each pixel. The code regarded as the pixel value is output to the specific information decoding unit 284, and the pixel value is output to the first prediction unit 270-1 and the second prediction unit 270. -2 and the image data assembly unit 288. That is, the data classification unit 282 classifies and outputs the codes and pixel values regarded as pixel values.

  If the decoded data includes a code obtained by shifting the code regarded as the pixel value described above by M + G bits, the data classification unit 282 sets a bit plane number from 0 to M + G and classifies the decoded data for each pixel. And output. When the code regarded as a pixel value or the most significant bit number of the pixel value is equal to or greater than M + G, the data classification unit 282 determines the code regarded as the pixel value by shifting to the M + G bit LSB side, and decodes the specific information To the conversion unit 284. In addition, the data classification unit 282 changes the value to the maximum value (255 if M = 8) when the most significant bit number of the code or pixel value regarded as the pixel value is greater than or equal to M and less than M + G, In this case, the pixel value is output to the first prediction unit 270-1, the second prediction unit 270-2, and the image data assembly unit 288 without changing the value.

  The specific information decoding unit 284 decodes the code input from the data classification unit 282, reproduces the specific information described above, and outputs it to the selection unit 286.

  The selection unit 286 receives the pixel values predicted by the first prediction unit 270-1 and the second prediction unit 270-2 for each pixel, and determines the first value according to the specific information input from the specific information decoding unit 284. The pixel values predicted by the prediction unit 270-1 and the second prediction unit 270-2 or any one of these are selected for each pixel and output to the image data assembly unit 288.

  The image data assembling unit 288 generates and outputs reproduced image data by assembling the pixel values input from the data classifying unit 282 and the pixel values input from the selecting unit 286 in order of decoded data for each pixel.

Next, a process in which the decoder 28a generates reproduced image data will be described.
FIG. 8 is a flowchart illustrating a process (S30) in which the decoder 28a generates reproduced image data.
As shown in FIG. 8, in step 300 (S300), when the JPEG2000 decoding unit 280 accepts the code stream, the decoded data obtained by decoding the code stream and the ROI information (or ROI mask) are sent to the data classification unit 282. Output.

  In step 302 (S302), the data classification unit 282 classifies the decoded data into either a code regarded as a pixel value or a pixel value (not including a code regarded as a pixel value) for each pixel, and regards it as a pixel value. The code is output to the specific information decoding unit 284, and the pixel value is output to the first prediction unit 270-1, the second prediction unit 270-2, and the image data assembly unit 288.

  In step 304 (S304), the specific information decoding unit 284 decodes the code input from the data classification unit 282, reproduces the specific information described above, and outputs it to the selection unit 286.

  In step 306 (S306), the first prediction unit 270-1 and the second prediction unit 270-2 each predict a pixel value for each pixel and output the pixel value to the selection unit 286.

  In step 308 (S308), the selection unit 286 predicts the first prediction unit 270-1 and / or the second prediction unit 270-2 according to the specific information input from the specific information decoding unit 284. The selected pixel value is selected for each pixel and output to the image data assembling unit 288.

  In step 310 (S310), the image data assembling unit 288 assembles the reproduced image data by assembling the pixel values input from the data classifying unit 282 and the pixel values input from the selecting unit 286 in the order of decoded data. Generate and output.

  In the above embodiment, lossless encoding or lossy encoding of pixel values is performed depending on whether the prediction errors of the first prediction unit 270-1 and the second prediction unit 270-2 are 0 or not. The case of determination has been described as an example, but the present invention is not limited to this, and lossless encoding or lossy encoding of pixel values is performed according to the pixel values of pixels around the pixel of interest (such as 4-connected or 8-connected). You may make it determine.

In addition, when the target pixel is isolated by being surrounded by pixels having different pixel values, the isolated pixel may be erased (change of the pixel value) by determining the target pixel.
In addition, in order to prevent deterioration in image quality of pixels that are irreversibly encoded around the pixel that is to be losslessly encoded by shifting the pixel value, an ROI mask for specifying the pixel that is to be losslessly encoded is generated. In this case, pixels around the pixel to be losslessly encoded may belong to the ROI.

Further, it may be determined that the target pixel is isolated according to the correlation between the pixel values shown in FIGS. For example, when a pixel block 30 composed of four (2 × 2) pixels is surrounded by a pixel 3a to be irreversibly encoded, the pixel 3 to be losslessly encoded disposed in the pixel block 30 or irreversibly encoded. Depending on the arrangement pattern of the pixels 3 a, it may be determined for each pixel block 30 whether or not the target pixel is isolated. For example, no matter how the pixels 3 to be losslessly encoded are arranged in the pixel block 30, it may be determined that the pixels 3 to be losslessly encoded in the pixel block 30 are isolated.
Further, when the pixel block 30 includes three or less pixels 3 to be losslessly encoded, it may be determined that the pixel 3 to be losslessly encoded in the pixel block 30 is isolated. Further, when one pixel 3 to be losslessly encoded is included in the pixel block 30, it may be determined that the pixel 3 to be losslessly encoded is isolated.
Further, as shown in FIG. 9B or 9C, when two pixels 3 to be losslessly encoded are arranged diagonally in the pixel block 30, the pixel 3 to be losslessly encoded is isolated. You may make it determine with having carried out.

1 is a configuration diagram illustrating an overview of an image processing apparatus according to an embodiment of the present invention. It is a block diagram which shows the structure of the code process part which is included in a control apparatus and encodes and decodes image data. It is a block diagram which shows the structure of an encoder. It is a flowchart which illustrates the process (S10) in which an encoder encodes image data. It is a block diagram which shows the structure of the modification of an encoder. It is a flowchart which illustrates the process (S20) in which the modification of an encoder encodes image data. It is a block diagram which shows the structure of the modification of a decoder. It is a flowchart which illustrates the process (S30) which the modified example of a decoder produces | generates reproduction | regeneration image data. It is a pixel arrangement diagram illustrating the correlation of pixel values as a condition for determining that the pixel of interest is isolated, and (A) shows a pixel block surrounded by pixels to be irreversibly encoded. (B) shows a first example in which two pixels to be losslessly encoded diagonally in the pixel block, and (C) shows two pixels to be losslessly encoded diagonally in the pixel block. It is an arrangement figure showing the 2nd example of an arranged state.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus 2 ... Control apparatus 20 ... CPU
DESCRIPTION OF SYMBOLS 22 ... Memory 24 ... Code processing part 26, 26a ... Encoder 260, 260a ... Reversible pixel selection part 262 ... ROI information creation part 264, 264a ... Control part 266 ... JPEG2000 encoding unit 270-1 ... first prediction unit 270-2 ... second prediction unit 272,272a ... prediction error calculation unit 274 ... determination unit 276 ... specific information generation unit 277 ..Specific information encoding unit 278 ... Data processing unit 28, 28a ... Decoder 280 ... JPEG2000 decoding unit 282 ... Data classification unit 284 ... Specific information decoding unit 286 ... Selection unit 288 ... image data assembly unit 3 ... pixel to be losslessly encoded 3a ... pixel to be irreversibly encoded 30 ... pixel block

Claims (17)

  Based on the correlation between the pixel of interest and surrounding pixels, binarization means for binarizing the image data for each pixel, embedded coding means for reversibly embedding the image data, and the binarization means An image processing apparatus comprising: control means for controlling the embedded coding means to discard a part of the image data to be embedded based on a result obtained by binarizing the image data.   The image processing apparatus according to claim 1, wherein the binarization unit binarizes the image data based on whether or not pixels having the same pixel value are around the target pixel.   The binarization means includes one or more prediction means for predicting a pixel value with respect to a pixel of interest, and a prediction error calculation means for calculating a prediction error of a pixel value predicted by the prediction means. The image processing apparatus according to claim 1, wherein the image data is binarized based on whether at least one of the prediction errors calculated by the error calculation means is zero.   A plurality of prediction means for predicting the pixel value of the pixel of interest by different prediction methods, a prediction error calculation means for calculating the prediction error of the pixel value predicted by the prediction means, and the prediction error calculation means Information for identifying the binarization means for binarizing image data for each pixel based on each prediction error, and the prediction means for predicting the pixel value for which the prediction error calculated by the prediction error calculation means is zero; Or embedded coding means for reversibly and separably embedding each pixel value of the target pixel, and the embedded coding based on the result of the binarization means binarizing the image data And an image processing apparatus having control means for controlling the means to discard a part of the data to be embedded.   5. The information for identifying the prediction unit is defined by the embedded encoding unit using a unique number assigned to each of the prediction units or a rank based on a frequency at which a pixel value with a prediction error of 0 is predicted. Image processing apparatus.   The embedded encoding means embeds data by generating a code stream compliant with JPEG2000, and the binarization means creates a ROI mask based on the correlation between the pixel of interest and surrounding pixels. The image processing apparatus according to claim 1, wherein the image data is binarized.   The said embedded encoding means shifts the information which specifies the said prediction means to a bit higher than MSB of the pixel value of a focused pixel, and produces | generates the code stream based on JPEG2000. Image processing device.   The image processing apparatus according to claim 3, wherein the embedded encoding unit generates a code stream compliant with JPEG2000 by regarding information specifying the prediction unit as a pixel value.   It further has an average value acquisition means for acquiring an average value of pixel values of surrounding pixels, and the embedded encoding means obtains a unique number assigned to each of the prediction means, or a pixel value for which the prediction error is zero. The image processing according to any one of claims 3 to 5, wherein a code stream compliant with JPEG2000 is generated by regarding a value obtained by adding a rank based on a predicted frequency and an average value acquired by the average value acquisition unit as a pixel value. apparatus.   The average value acquisition unit further includes an average value rounding unit that rounds the average value acquired by the average value acquisition unit, and the embedded encoding unit assigns a unique number assigned to each of the prediction units, or a pixel value with a prediction error of 0. The image processing apparatus according to claim 9, wherein a value obtained by adding a ranking based on a predicted frequency and an average value rounded by the average value rounding unit is regarded as a pixel value, and a code stream compliant with JPEG2000 is generated.   11. The image according to claim 1, wherein when the binarized result of the binarizing unit is isolated, the control unit performs control so as to invert the binarized result for the isolated pixel. Processing equipment.   The control unit performs control so as to correct the binarized result for the isolated pixel block when the binarized result of the binarizing unit is isolated in units of 2 × 2 pixel blocks. The image processing apparatus according to claim 1.   The image processing according to claim 1, wherein the control unit performs control so as to invert a binarized result with respect to pixels around pixels having the same binarized result by the binarizing unit. apparatus.   The image processing apparatus according to claim 3, wherein the control unit controls the embedded encoding unit so that a pixel value predicted to have a prediction error of 0 is reversibly encoded.   Embedded decoding means for decoding a code stream in which information or pixel values based on prediction errors are embedded for each pixel, and information and pixel values based on prediction errors for decoded data decoded by the embedded decoding means An image processing apparatus comprising: classifying means for classifying the image data; and image data reproducing means for reproducing image data from the decoded data decoded by the embedded decoding means based on the classification result of the classification means.   The embedded decoding means conforms to JPEG2000, and the classification means classifies the decoded data into information based on a prediction error and pixel values according to ROI encoded information included in a code stream. Item 15. The image processing apparatus according to Item 15.   Based on the correlation between the target pixel and the surrounding pixels, the image data is binarized for each pixel, and the image data is reversibly embedded and the image data is partially binned based on the binarized result. Image processing method to be cut off.

Images