JP3985465B2 - Image encoding apparatus, image decoding apparatus, image encoding / decoding apparatus, and methods thereof - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image compression technique, and more particularly to a technique for compressing both binary and multi-valued images.
[0002]
[Prior art]
Since image data generally has an enormous amount of data, it is often compressed to reduce the amount of data when performing communication or storage. There are a number of techniques for encoding techniques for compression. Broadly speaking, there are a reversible method that completely reproduces the input when decoded and a nonreciprocal method that involves some loss. Furthermore, if these two methods are classified by basic algorithms and parameters, they can be subdivided into a number of methods. These compression methods are hereinafter referred to as encoding.
[0003]
Each encoding has its own operating principle, which makes some assumptions on the input image. For example, in the international standard JPEG (Joint Photographic Group) baseline method (hereinafter simply referred to as JPEG), DCT (Discrete Cosine Transform, Discrete Cosine Transform) is used. It is assumed to concentrate in the low frequency region.
[0004]
Such assumptions tend to limit the type of input image. One of the types that are particularly easily limited is a dynamic range of pixel values, for example. Hereinafter, a 1-pixel 1-bit image that is normally expressed as white and black binary is called a binary image, and an image with 1-pixel 2 bits or more is called a multi-value image. At present, multi-valued images often consist of 8 bits for one color of one pixel. For example, there are a gray image of 1 pixel 8 bits and an RGB full color image of 1 pixel 24 bits.
[0005]
It is generally difficult to compress a binary image and a multi-value image in the same manner. For example, in the case of the above-mentioned JPEG, the performance of the binary image is degraded because information is also present in the high frequency region. As an example opposite to this, in the case of the international standard JBIG (Joint Bi-level Image experts Group) specializing in binary images, a multi-valued image is divided into binary planes called bit planes, and processed. This also reduces performance.
[0006]
Now, in general applications, binary images and multi-valued images are used together. It may be mixed in one page or may be mixed page by page in one document. In any case, for encoding such an image, both a binary image and a multi-valued image are used. A mechanism for effective compression is required.
[0007]
In order to solve this problem, encoding for binary images and multi-value images may be used together. However, since the scale of the apparatus increases, it is desirable that encoding is possible so that both images can be handled in the same way. Therefore, Japanese Patent Laid-Open No. 9-109790, which is a prior art that realizes this requirement, will be described as a conventional example. This conventional example is a technique in which encoding is integrated into one by unifying binary encoding and multi-level encoding at the code level.
[0008]
FIG. 15 shows a configuration example of a conventional coding apparatus. Although the terminology has been partially changed to meet the purpose of the description of the present invention, it does not relate to the essence of the conventional example. In the figure, 10 is an image input unit, 61 is a multi-component processing unit, 62 is a method processing unit, 63 is a wavelet transform unit, 64 is a quantization unit, 65 is a horizontal context model unit, 66 is a gray encoding unit, and 40 is An encoding unit, 50 is a code output unit, 110 is input image data, 151 and 152 are processed image data, 153 is coefficient series data, 154 is quantized data, 130 is symbol data, and 140 is code data.
[0009]
Each part of FIG. 15 will be described. The image coding apparatus in FIG. 15 has the following configuration. The image input unit 10 receives an image from the outside and sends it as input image data 110 to the multi-component processing unit 61. The multi-component processing unit 61 performs optional processing such as predetermined color space conversion on the input image data 110 and sends the processed image data 151 to the method processing unit 62. The system processing unit 62 determines the characteristics of the processed image data 151 and transmits the entire image or a partial image to either the wavelet transform unit 63 or the gray encoding unit 66. The wavelet transform unit 63 performs reversible wavelet transform on the processed image data 152 and sends it to the quantization unit 64 as coefficient series data 153. The quantizing unit 64 quantizes the coefficient series data 153 and sends the quantized data 154 to the horizontal context model unit 65 as quantized data 154. The horizontal context model unit 65 models the quantized data 154 based on its importance and sends it as symbol data 130 to the encoding unit 40. On the other hand, the gray encoding unit 66 performs gray encoding on the processed image data 152 and sends the processed image data 152 to the encoding unit 40 as symbol data 130. The encoding unit 40 performs predetermined entropy encoding on the symbol data 130, and sends it to the code output unit 50 as code data 140. The code output unit 50 sends the code data 140 to the outside.
[0010]
The operation of the conventional example based on the above configuration will be described. FIG. 16 is a flowchart showing the operation of the conventional image coding apparatus. The operation of the conventional example will be described below with reference to FIG. In S10, the image input unit 10 inputs an image from the outside. In S51, the multi-component processing unit 61 performs multi-component processing. In S52, the method processing unit 62 performs method processing. In S53, if it is determined as a multi-valued image as a result of S52, the process proceeds to S54, and if not, the process proceeds to S57. In S54, the wavelet transform unit 63 performs wavelet transform. In S55, the quantization unit 64 performs a quantization process. In S56, the horizontal context model unit 65 performs horizontal context model processing. In S57, the gray encoding unit 66 performs gray encoding. In S30, the encoding unit 40 performs predetermined entropy encoding. In S40, the code output unit 50 performs code output processing to the outside.
[0011]
Next, problems of the conventional example will be described. In the conventional example, it is possible to process both binary and multi-valued images, but only the encoding unit 40 that performs entropy encoding is common to both images. In the preceding stage, an independent encoding path is prepared depending on the type. In this respect, it can be said that there is almost no difference from the conventional one. That is, in the conventional example, the encoding of the binary image and the multi-value image is not sufficiently unified, and as a result, it can be said that the scale of the encoding device is increased.
[0012]
In the conventional example, the binary and multi-value paths are independent. If independent paths are provided inside the apparatus in this way, there is a difference in the processing load of each path, so synchronization is finally performed in the part to be integrated finally, in the conventional example, in the part of the encoding unit 40. It becomes difficult to take. In order to solve this, there is no choice but to provide some kind of buffer or perform extremely complicated control.
[0013]
[Problems to be solved by the invention]
As described above, the problems of the conventional example are that the apparatus scale becomes large despite the fact that both binary and multi-valued images can be handled, and the control is complicated.
[0014]
The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an image coding apparatus that can be effectively compressed without distinguishing between binary and multi-valued images with a small and simple configuration.
[0015]
[Means for Solving the Problems]
According to an aspect of the present invention, in an image encoding device, an image input unit that inputs an image, a prediction unit that performs a predetermined prediction process on the image data input by the image input unit, and the prediction unit An encoding unit that encodes a prediction error or a prediction result by a predetermined method; and a code output unit that outputs the encoding result of the encoding unit. The prediction process in the prediction unit is input to the image input unit. The image to be adjusted is adjusted according to whether it is binary or multivalued.
[0016]
According to another aspect of the present invention, in an image decoding apparatus, a code input unit that inputs a code, a decoding unit that decodes a code input by the code input unit by a predetermined method, and the decoding processing result A reverse prediction processing unit that performs predetermined reverse prediction processing to generate an image; and an image output unit that outputs a result of the reverse prediction processing. The reverse prediction processing in the reverse prediction unit is input from the code input unit. The code to be adjusted is adjusted according to whether a binary image is encoded or a multi-value image is encoded.
[0017]
According to still another aspect of the present invention, in the image coding / decoding apparatus, an image input unit that inputs an image, and a prediction unit that performs a predetermined prediction process on the image data input by the image input unit; Encoding means for encoding the prediction error or prediction result of the prediction means by a predetermined method, code output means for outputting the encoding result of the encoding means, code input means for inputting the code, and A decoding unit that decodes the code input by the code input unit by a decoding process that is an inverse process of the encoding unit; and an inverse prediction process that generates an image by performing an inverse prediction process that is an inverse process of the prediction process on the decoding process result And an image output means for outputting the result of the reverse prediction process. The prediction process in the prediction means and the reverse prediction process in the reverse prediction means In which the image to be input to the means to be adjusted depending on whether a binary or multivalue.
[0018]
In the image coding apparatus, the image decoding apparatus, and the image coding / decoding apparatus configured as described above, basically the same predictive coding method is adopted for both the image coding and the image decoding, and then the image is Adjustments are made according to whether binary or multivalued. Note that the prediction code is equally applicable to binary and multi-valued images. In this configuration, the configuration is simplified, and encoding / decoding that matches the characteristics of the image can be performed. In addition, since the coding path is not divided into two, complicated control for synchronization becomes unnecessary.
[0019]
In the configuration as described above, image type designation means for designating whether an input image is binary or multivalued as an image type is provided, and prediction processing in the prediction means and reverse prediction processing in the reverse prediction means May be adjusted according to the image type designated by the image type designation means.
[0020]
The designation of the image type in the image type designation means may be performed in units of one image, one partial image, or one pixel.
[0021]
In addition, the prediction process performed by the prediction unit and the reverse prediction process performed by the reverse prediction unit are performed by packing one pixel at a time when the image type is multi-valued, and packing a plurality of pixels when the image type is binary. May be predicted.
[0022]
In addition, the prediction process performed by the prediction unit and the reverse prediction process performed by the inverse prediction unit calculate a prediction error by subtraction or exclusive OR, and as an example, when the image type is multi-valued, it is subtracted. If the value is binary, the prediction error may be calculated by exclusive OR.
[0023]
In addition, the prediction process performed by the prediction unit and the reverse prediction process performed by the inverse prediction unit are performed when a pixel around the pixel to be encoded is binary and the binary is binary when the image type is multi-valued. Alternatively, the prediction may be performed using pixels located at a position matching the threshold period of the halftoning process.
[0024]
The prediction process performed by the prediction unit and the reverse prediction process performed by the inverse prediction unit are determined by adjusting the number of pixels to be packed and the position of the prediction pixel so as to match the threshold period of the halftoning process. You may do it.
[0025]
The prediction processing performed by the prediction means determines that the prediction matches if the prediction value is within a predetermined range with respect to the pixel value of the pixel to be encoded, or quantizes the prediction error. Alternatively, the inverse prediction process performed by the inverse prediction means by performing both or both of them may be performed by performing an inverse quantization process when necessary corresponding to the prediction process.
[0026]
Further, the quantization of the prediction error performed by the prediction unit and the inverse quantization process of the prediction error performed by the inverse prediction unit are not performed in the normal quantization but in the bit position when the input image is binary. Regardless of this, quantization such as changing 1 to 0 may be used.
[0027]
Further, the encoding performed by the encoding means and the decoding performed by the decoding means may be any of QM coder, LZ encoding, Huffman encoding, Golomb-Rice encoding, block sorting encoding, or a combination thereof. May be.
[0028]
Further, the prediction process performed by the prediction unit and the reverse prediction process performed by the reverse prediction unit may be fixedly designed for either binary or multi-value depending on the application.
[0029]
The image input means includes image type determining means for determining whether the image input is binary or multi-valued as an image type, and the prediction process in the predicting means is determined by the image type determining means. The image type may be adjusted according to the image type, and the image type may be added to the code by the encoding means.
[0030]
The decoding unit may acquire an image type from the code and provide the image type to the inverse prediction unit, and the prediction process in the inverse prediction unit may be adjusted according to the image type.
[0031]
In addition, the image type determination process performed by the image type determination unit may be performed by histogram or frequency analysis, or may be performed by the matching rate between the binary and multi-level prediction processes in the prediction unit.
[0032]
The image type determination result performed by the image type determination unit may be defined as a code on the same code space as the prediction error or the prediction result, or may be defined by a header or a marker as additional information. .
[0033]
The prediction means includes at least two or more pixel value prediction means for predicting the pixel value of the input image, and a prediction error calculation for calculating an error between a prediction value obtained by a predetermined prediction method and an actual pixel value by a predetermined method. And a selection means for selecting one of the pixel value prediction means and the prediction error calculation means, and the selection means selects the prediction means when any of the prediction values matches the actual pixel value. If none of them match, the prediction error may be selected, and the encoding means may encode the selection result in the selection means and the prediction error.
[0034]
The inverse prediction means includes at least two or more pixel value prediction means for predicting the pixel value of the image, a prediction error correction means for calculating a correct pixel value from the prediction error value by a predetermined method, and the pixel value prediction. And a selection unit that selects one of the prediction error calculation units, and the selection unit may perform a selection process based on a selection result or a prediction error obtained by the decoding unit.
[0035]
Further, the selection means is one of a predictor identifier, a prioritized identifier, and a run-length representation of the prioritized identifier, and when the identifier is run-length represented, The encoding means may encode the run length.
[0036]
Further, the prediction means counts at least two or more pixel value prediction means for predicting the pixel value of the input image, and counts the number of consecutive matches between the prediction value of the corresponding pixel value prediction means and the actual pixel value. One or more run counting means, a prediction error calculating means for calculating an error between a predicted value by a predetermined prediction method and an actual pixel value by a predetermined method, and any one of the pixel value predicting means and the prediction error calculating means. Selecting means for selecting, the selecting means selects the longest matching prediction means, when none of them match, the prediction error is selected, the encoding means is a selection result obtained by the selection means, The run length obtained by the run counting means and the prediction error obtained by the prediction error calculation may be encoded.
[0037]
In addition, the inverse prediction means includes at least two or more pixel value prediction means for predicting the pixel value of the image, and at least two that transmit control information for the number of runs input to the corresponding pixel value prediction means. The above-described run development means, a prediction error correction means for calculating a correct pixel value from a prediction error value by a predetermined method, and a selection means for selecting one of the run development means and the prediction error correction means, The decoding unit decodes a selection result, a run length, and a prediction error, and the selection unit performs a selection process based on the selection result. When the run expansion unit is selected, a run length is calculated based on the run length. When the expansion process is performed and the prediction error correction unit is selected, the prediction error correction process may be performed based on the prediction error.
[0038]
It should be noted that the present invention can be realized not only as an apparatus or a system but also as a method, and a part thereof can be realized as a computer program.
[0039]
Each of the above aspects of the invention and other aspects of the invention are set forth in the appended claims and will be described in detail below with reference to examples.
[0040]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described in detail with reference to examples.
[0041]
[Basic principle]
Prior to specific description of the embodiments of the present invention, the basic principle of the present invention will be described. In order to compress a binary image and a multi-valued image without distinction, a small-scale apparatus can be realized by selecting an encoding using only a property common to both images. In this respect, in the case of the conventional example, wavelet coding was used for multi-valued images, and coding was completely different in principle from gray coding for binary images, so the increase in scale could not be suppressed. .
[0042]
The present invention employs predictive coding. Predictive coding here refers to a coding method in which the pixel to be coded next is predicted by referring to the information of the processed image, and the matching information and error information are used as codes.
[0043]
Predictive coding uses a property called image continuity or autocorrelation, but these properties exist without distinction between binary and multivalued, and therefore both can be compressed in common. However, since many processes generally called prediction are designed for multi-valued images, some ideas are required to apply them to binary images. In principle, this is an adjustment for using the autocorrelation of the binary image. Such adjustment can be dealt with by a part of prediction processing, that is, by adjusting parameters. Specific examples will be described in the examples.
[0044]
The present invention realizes compression of both images by adaptively switching between the prediction for binary images and the prediction for multi-valued images prepared in this way. At this time, since both the images other than the prediction processing can be used in common for both images, and the prediction itself can be used by switching the parameters, the apparatus scale can be reduced. In addition, since the coding path is not divided into two, complicated control necessary for synchronization is not necessary.
[0045]
Specific examples of the present invention will be described in Examples. Hereinafter, as an example of the present invention,
(1) General example
(2) Example of determining image type internally
(3) Example applied to predictive coding disclosed in Japanese Patent Laid-Open No. 09-224253 (4) Modification of Example (3)
The following four will be described.
[0046]
[First embodiment]
First, a general example will be described as the first embodiment of the present invention. Hereinafter, the first embodiment of the present invention will be described in detail. 1 and 2 are block diagrams showing an image encoding device and an image decoding device, respectively, in the first embodiment. In the figure, parts similar to those in FIG. In the figure, 20 is an image type designation unit, 30 is a prediction unit, 51 is a code input unit, 41 is a decoding unit, 31 is a reverse prediction unit, 11 is an image output unit, and 120 is image type data.
[0047]
First, the image coding apparatus according to the first embodiment of the present invention will be described with reference to FIG. Each part of FIG. 1 will be described. The image type designation unit 20 inputs a type indicating whether an image input from the outside is binary or multi-valued, and sends it to the prediction unit 30 as image type data 120. The prediction unit 30 performs a prediction process on the input image data 110 in accordance with the image type specified by the image type data 120 and sends the result to the encoding unit 40 as symbol data 130.
[0048]
Next, the image decoding apparatus according to the first embodiment of the present invention will be described with reference to FIG. Each part of FIG. 2 will be described. The code input unit 51 inputs a code from the outside, and sends the code data 140 to the decoding unit 41. The decoding unit 41 performs a decoding process corresponding to the reverse process of the encoding performed by the encoding unit 40 on the code data 140 and sends it to the inverse prediction unit 31 as the symbol data 130. Based on the image type data 120, the inverse prediction unit 31 performs inverse prediction processing corresponding to the inverse processing of prediction performed by the prediction unit 30 on the symbol data 130, and sends the input image data 110 to the image output unit 11. The image output unit 11 outputs the input image data 110 to the outside.
[0049]
Based on the above configuration, the operation of the first embodiment of the present invention will be described. 3 and 4 are flow charts showing the encoding operation and the decoding operation in the first embodiment, respectively. In the figure, the same parts as those in FIG. However, explanations will be added for slightly different parts.
[0050]
First, the encoding process in the first embodiment of the present invention will be described with reference to FIG. In S10, the image input unit 10 and the image type designation unit 20 accept input of an image and an image type from the outside. In S20, the prediction unit 30 performs a prediction process.
[0051]
Next, the decoding process in the first embodiment of the present invention will be described with reference to FIG. In S41, the code input unit 41 and the image type designation unit 20 accept the input of the code and the image type from the outside. In S31, the decoding unit 41 performs a decoding process. In S21, the inverse prediction unit 31 performs an inverse prediction process. In S11, the image output unit 11 outputs an image.
[0052]
In the above operation, the encoding process and the decoding process in the encoding unit 40 and the decoding unit 41 are inverse processes. Similarly, the prediction process and the reverse prediction process in the prediction unit 30 and the reverse prediction unit 31 are also reverse processes. Specific examples of these will be described later in detail.
[0053]
The image type designation in the image type designation unit 20 may be performed for each image, or may be performed for each partial image or pixel unit.
[0054]
Next, the prediction process in the prediction unit 30 will be described. When a multi-valued image is designated by the image type designation unit 20, this prediction is performed based on the peripheral pixel value of a pixel to be encoded (hereinafter referred to as a target pixel) or a calculation result thereof. These details are omitted because they are known techniques such as JPEG lossless compression.
[0055]
When a binary image is designated by the image type designation unit 20, this prediction performs adjustment for a binary image. In order to explain this, first, the difference in prediction between a binary image and a multi-valued image will be described. Prediction coding for a natural image such as a photograph rarely matches the actual pixel value with the predicted value, and therefore, in general, this difference, so-called prediction error, is coded. If the image is multi-valued, the prediction error distribution is a monotonous distribution with the vertex at 0, so there is an effect of reducing the amount of information, but if it is binary, the prediction will not match or not, so such an effect I can not expect.
[0056]
In addition, the multi-value prediction as described above is based on the autocorrelation of an image in principle. However, in the case of an image generated by a so-called halftoning (also referred to as screen processing) based on a binary image, particularly a multi-value image, the threshold for binarization differs for each pixel, so that it is the same as such a multi-value image. The autocorrelation of is not necessarily established.
[0057]
Thus, prediction of a binary image and a multi-value image cannot be performed in the same way. Therefore, in the present invention, in order to compress a binary image by predictive coding, the following adjustment is performed for prediction. That is,
1) Predict multiple pixels together,
2) Calculate the prediction error by exclusive OR,
3) Reflect the halftoning cycle.
It is three. For example, 1) is an adjustment in which eight 1-bit pixels are grouped together (hereinafter referred to as packing) and handled as if they were one 8-bit pixel. In addition, 2) is an adjustment using exclusive OR instead of subtraction generally used in multi-valued images. With these two, it is possible to expect a reduction in the amount of information due to prediction errors. Further, 3) is an adjustment for performing a prediction process in accordance with an assumed halftoning process. For example, in a case where halftoning processing is performed with a 12 × 12 systematic dither and packing is performed in units of 8 bits in 1), the pixel value of the 8-bit pixel of interest is 2 more than the pixel value of the right-left pixel. The correlation with the pixel value of the leftmost pixel becomes high. This is because the dither is 12 pixel cycles, and the packing is 8 pixel cycles. Therefore, the least common multiple of these 24 pixel cycles, that is, the pixel value after packing is highly correlated with every other pixel value. This is the cause. Using this, for example, if the previous pixel value is adjusted to be the predicted value, the accuracy of the prediction is increased. 3) indicates a prediction that takes into account the halftoning cycle and the packing cycle of 1).
[0058]
These adjustments are preferably performed simultaneously, but only one of them may be selectively performed as long as there is a restriction such as processing load.
[0059]
In addition, when the input image data is limited to either one, the fixed prediction process may be performed. Even in this case, since most of the design and implementation of the present embodiment configured for the case where the input image cannot be limited can be used, the cost can be reduced as a whole. For example, when a binary image system and a multi-value image system exist separately, this embodiment can be applied to each of the embodiments. In this case, all parts other than the prediction process can be used in common, so there are many cost reduction effects such as shortening the design man-hours and sharing the mounting core compared to the case where separate coding is adopted. You can expect.
[0060]
In addition, lossy encoding can be realized by allowing an error in these prediction processes, but such a configuration may be adopted. For example, a process for determining that the prediction matches if the predicted value is within a predetermined range, a process for adding some quantization to the prediction error, and the like are given as specific examples. At this time, since the prediction error value obtained by exclusive OR for a binary image has no meaning in the magnitude relationship, the bit position closer to the LSB is not the normal quantization that is easily ignored, but the bit position. Regardless of this, quantization such as changing 1 to 0 is effective.
[0061]
The symbol data 130 is composed of a prediction error or the number of consecutive prediction matches. This specific example will be described in detail in a later embodiment.
[0062]
The encoding performed by the encoding unit 40 is not particularly limited as long as it is a general entropy coder. Examples include QM coder, LZ coding, Huffman coding, Golomb-Rice coding, and the like. However, since the present invention aims at a small-scale configuration, a small-scale entropy coder such as Huffman is particularly preferable.
[0063]
As described above, according to the first embodiment, with regard to a binary image, it is possible to perform a highly accurate prediction by simply making a simple adjustment to the prediction for a multi-value image. An image encoding process and an image decoding process that can efficiently process a binary image and a multi-valued image can be performed.
[0064]
[Second Embodiment]
As a second embodiment of the present invention, an example will be described in which the image type is not given externally as in the first embodiment but is determined internally as in the conventional example.
[0065]
The second embodiment of the present invention will be specifically described below. 5 and 6 are block diagrams showing an image encoding device and an image decoding device, respectively, according to the second embodiment of the present invention. In the figure, the same parts as those in FIG. 15, FIG. 1 and FIG. 21 is an image type determination unit, 42 is an encoding unit, and 43 is a decoding unit.
[0066]
First, the image coding apparatus according to the second embodiment of the present invention will be described with reference to FIG. Each part of FIG. 5 will be described. The image type determination unit 21 performs a predetermined image type determination process on the input image data 110 and sends it to the prediction unit 30 and the encoding unit 42 as image type data 120. The encoding unit 42 performs predetermined entropy encoding on the symbol data 130 and the image type data 120 and sends the encoded data 140 to the code output unit 50.
[0067]
Next, an image decoding apparatus according to the second embodiment of the present invention will be described with reference to FIG. Each part of FIG. 6 will be described. The decoding unit 43 performs a decoding process corresponding to the reverse process of the encoding process performed by the encoding unit 42 on the code data 140, and sends it to the inverse prediction unit 31 as the symbol data 130 and the image type data 120.
[0068]
Since the operation of the second embodiment of the present invention can be easily inferred from the operation of the first embodiment of the present invention, description thereof will be omitted.
[0069]
Among the operations described above, the image type determination process in the image type determination unit 21 determines whether the whole or part of the input image data 110 is a binary multi-value. In the conventional example, this corresponds to a determination process in the method processing unit 62. The operation principle of the determination process is not described in detail because there are many known techniques, but it can be determined by, for example, a histogram or frequency analysis. Further, in the present embodiment, it is not important to distinguish between binary and multivalue, and the purpose is to know which prediction is likely to match. For example, the prediction for binary and multivalue is partially applied. Therefore, it may be determined based on the matching rate.
[0070]
The encoding unit 42 is different from the encoding unit 40 in the first embodiment of the present invention in that the image type data 120 is encoded in addition to the symbol data 130. The relationship between the decoding unit 43 and the decoding unit 41 in the first embodiment of the present invention is also the same.
[0071]
Next, a method for discriminating between the symbol data 130 and the image type data 120 in the code data 140 will be described. What is simple is that codes corresponding to the symbol data 130 and the image type data 120 have a one-to-one correspondence in the same code space. In this way, it is possible to determine which information is from each code.
[0072]
When the image type data 120 is small with respect to the symbol data 130 such as for each image or for each partial image, the above-described method causes a decrease in encoding efficiency. Therefore, in such a case, it is preferable to embed in the code data 140 as additional information in the form of a header or a marker. FIG. 7 is a conceptual diagram of an example of such a header. As a variation, the information included in the header may be transmitted as additional information data that is completely different from the code data 140.
[0073]
As described above, according to the second embodiment, an image encoding device and an image decoding device that perform encoding by switching between binary and multi-value without specifying an image type from outside are realized on a small scale. can do.
[0074]
[Third embodiment]
As a third embodiment of the present invention, an example in which the present invention is applied to the predictive coding disclosed in Japanese Patent Laid-Open No. 09-224253 will be described.
[0075]
The third embodiment of the present invention will be specifically described below. 8 and 9 are block diagrams showing an image encoding device and an image decoding device, respectively, according to the third embodiment of the present invention. In the figure, the same parts as those in FIGS. 15, 1, 2, 5, and 6 are denoted by the same reference numerals, and description thereof is omitted. In the figure, 32 is a first prediction unit, 33 is a second prediction unit, 36 is a prediction error calculation unit, 37 is a selection unit, 38 is a prediction error correction unit, 111 and 112 are prediction value data, and 115 is prediction error value data. 116, 117 are control data.
[0076]
First, an image coding apparatus according to a third embodiment of the present invention will be described with reference to FIG. Each part of FIG. 8 will be described. Each of the first prediction unit 32 and the second prediction unit 33 performs a predetermined prediction process based on the image type data 120 for the input image data 110, and sends the input data as prediction value data 111 and 112 to the selection unit 37. The prediction error calculation unit 36 performs a predetermined prediction process on the input image data 110 based on the image type data 120, calculates a difference between the prediction value and the target pixel value by a predetermined method, and generates prediction error value data. 115 is sent to the selection unit 37. The selection unit 37 compares the input image data 110 with the prediction value data 111 and 112, and if the predictions match, the identifier of the predictor that matches, and if the predictions do not match, the prediction error value data 115 as the symbol data. 130 is sent to the encoding unit 40.
[0077]
Next, an image coding apparatus according to a third embodiment of the present invention will be described with reference to FIG. Each part of FIG. 9 will be described. The selection unit 37 selects any one of the first prediction unit 32, the second prediction unit 33, and the prediction error correction unit 38 based on the symbol data 130, and sets the control data 116 and 117 and the prediction error value data 115 according to each of them. Send it out. The first prediction unit 32 and the second prediction unit 33 are the same as those in the above-described image encoding device, but operate only when the control data 116 is received, and send the prediction result to the image output unit 11 as input image data 110. The point is different. The prediction error correction unit 38 performs the inverse process of the prediction error calculation performed by the prediction error calculation unit 36 and sends the input image data 110 to the image output unit 11.
[0078]
Since the operation is apparent from other embodiments, the description thereof is omitted.
[0079]
In the above operation, the first prediction unit 32, the second prediction unit 33, and the prediction error correction unit 38 have the input image data 110 as input and output, but the input is delayed by one pixel from the output. That is, by inputting the pixel value decoded immediately before, prediction processing and prediction error correction processing for the next pixel are performed.
[0080]
Next, the prediction process performed in the 1st prediction part 32 and the 2nd prediction part 33 is demonstrated. In Japanese Patent Laid-Open No. 09-224253, prediction based on the peripheral pixel value itself is given as an example, but when a binary image is targeted, the adjustment as described in the first embodiment of the present invention is performed. Here, assuming that 16 × 2 dither processing has been performed, packing is performed in units of 8 pixels, and every other predictor is used. FIG. 10 shows a part of an example of a predictor for a multi-level image disclosed in Japanese Patent Laid-Open No. 09-224253, and FIG. 11 shows an example of a predictor adjusted for a binary image. This is an example, and adjustment is made according to the halftoning cycle as described in the first embodiment of the present invention.
[0081]
The prediction error calculation unit 36 and the prediction error correction processing unit 38 are also as described in the first embodiment of the present invention. In this embodiment, when a multi-value image is specified, subtraction and binary image are specified. If this is the case, the prediction error is calculated by exclusive OR.
[0082]
The symbol data 130 is composed of the identifiers and prediction error values of the matched predictors. However, as disclosed in Japanese Patent Application Laid-Open No. 09-224253, the identifiers may be prioritized or expressed in run length. Details of these will be omitted from the gist of the present invention because they are not.
[0083]
As described above, according to the third embodiment, since a plurality of predictors are provided, an image coding apparatus and an image that can be compressed without distinction between binary and multivalue particularly for an artificial image such as CG. A decoding device can be realized.
[0084]
[Fourth embodiment]
As a fourth embodiment of the present invention, a modification of the third embodiment of the present invention will be described. Here, in the third embodiment of the present invention, a configuration is adopted in which the prediction unit is selected based on the number of continuous matches of the prediction unit, not the mismatch of the prediction unit.
[0085]
Hereinafter, the fourth embodiment will be described in detail. 12 and 13 are block diagrams of an image encoding device and an image decoding device, respectively, according to the fourth embodiment of the present invention. In the figure, the same parts as those in FIGS. 15, 1, 2, 5, 5, 6, 8 and 9 are denoted by the same reference numerals, and the description thereof is omitted. Reference numerals 34 and 35 are run counting units, 71 and 72 are run developing units, and 113 and 114 are run data.
[0086]
First, an image coding apparatus according to a fourth embodiment of the present invention will be described with reference to FIG. Each part of FIG. 12 will be described. The run counting units 34 and 35 compare the image input data 110 with the predicted value data 111 and 112, respectively, count the number of consecutive matches, and send the results to the selection unit 37 as run data 113 and 114, respectively. If the larger one of the run data 113 and 114 is not 0, the selection unit 37 sends the identifier and run of the corresponding prediction unit, and if it is 0, the prediction error data 115 is sent to the encoding unit 40 as the symbol data 130.
[0087]
Next, an image decoding apparatus according to the fourth embodiment of the present invention will be described with reference to FIG. Each part of FIG. 13 will be described. The selection unit 37 selects one of the run development units 71 and 72 and the prediction error correction unit 38 based on the symbol data 130, and sends the run data 113 and 114 and the prediction error value data 115 according to each of them. The run development units 71 and 72 send the control data 116 and 117 to the first prediction unit 32 and the second prediction unit 33 as many times as indicated by the run data 113 and 114, respectively.
[0088]
Since the operation is apparent from other embodiments, the description thereof is omitted.
[0089]
Since the prediction unit can be selected after counting the number of consecutive matches in the above operation, the compression rate can often be made higher than in the third embodiment of the present invention.
[0090]
Therefore, in order to confirm the effect of the present embodiment, a simulation of the present embodiment was performed on a computer. FIG. 14 shows the experimental results. As can be seen from the figure, JPEG and JBIG select image types, whereas in this embodiment, both can be efficiently compressed.
As described above, according to the fourth embodiment of the present invention, the image encoding process for compressing binary and multi-valued images and the amplification process are realized more efficiently than the third embodiment. it can.
[0091]
【The invention's effect】
As is apparent from the above description, according to the present invention, high-efficiency compression processing is realized with a small-scale configuration, particularly in an image encoding device and an image decoding device that compress both binary and multilevel images. be able to.
[Brief description of the drawings]
FIG. 1 is a configuration diagram illustrating a first embodiment of an image encoding device according to the present invention;
FIG. 2 is a block diagram showing a first embodiment of the image decoding apparatus of the present invention.
FIG. 3 is a flowchart showing an example of the operation in the first embodiment of the image encoding device of the present invention;
FIG. 4 is a flowchart showing an example of operation in the first embodiment of the image encoding device of the present invention;
FIG. 5 is a block diagram showing a second embodiment of the image coding apparatus according to the present invention.
FIG. 6 is a block diagram showing a second embodiment of the image decoding apparatus of the present invention.
FIG. 7 is an explanatory diagram of an example of code data in the second embodiment of the image encoding device and the image decoding device of the present invention.
FIG. 8 is a block diagram showing a third embodiment of the image coding apparatus according to the present invention.
FIG. 9 is a block diagram showing a third embodiment of the image decoding apparatus of the present invention.
FIG. 10 is an explanatory diagram of a prediction process in the third embodiment of the image encoding device and the image decoding device according to the present invention.
FIG. 11 is an explanatory diagram of a prediction process in the third embodiment of the image encoding device and the image decoding device according to the present invention.
FIG. 12 is a block diagram showing a fourth embodiment of the image coding apparatus according to the present invention.
FIG. 13 is a block diagram showing a fourth embodiment of the image decoding apparatus of the present invention.
FIG. 14 is an explanatory diagram showing an example of experimental results according to the fourth embodiment of the image encoding device and the image decoding device of the present invention;
FIG. 15 is a block diagram showing a conventional coding selection apparatus.
FIG. 16 is a flowchart showing an example of the operation of the conventional encoding selection apparatus.
[Explanation of symbols]
10 Image input section
11 Image output unit
20 Image type designation part
21 Image type determination unit
30 Predictor
31 Inverse prediction unit
32 First prediction unit
33 Second prediction unit
34 Run counter
35 Run counter
36 Prediction error calculator
37 Selector
38 Prediction error correction unit
40 Encoding unit
41 Decryption unit
42 Encoding unit
43 Decryption unit
50 Code output section
51 Code input section
61 Multi-component processing unit
62 Method processor
63 Wavelet transform unit
64 Quantizer
65 Horizontal context model part
66 Gray coding part
71 Run development department
72 Run Development Department
110 Input image data
111 Predicted value data
112 Predicted value data
113 run data
114 run data
115 Prediction error value data
116 Control data
117 Control data
120 Image type data
130 Symbol data
140 Code data
151 Processed image data
152 Processed image data
153 Coefficient series data
154 Quantized data

Claims (15)

Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means If the image data of the data 2 value, in a second prediction process aspect, a prediction means for performing the predetermined prediction process on the image data of the binary,
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Possess a code output means for outputting the encoded result of the encoding means,
The prediction process performed by the prediction means is a process of predicting by packing one pixel at a time when the image type is multi-valued, and packing a plurality of pixels when the image type is binary,
When the image type is multi-valued, the prediction error is calculated by subtraction. When the image type is binary, the prediction error is calculated by exclusive OR,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LSB side when the input image is multi-valued, and uses a bit instead of normal quantization when the input image is binary. An image encoding apparatus characterized in that the quantization is such that 1 is changed to 0 regardless of the position of the image. Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means When the data is binary image data, in a second prediction processing mode, prediction means for performing the predetermined prediction processing on the binary image data;
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Code output means for outputting the encoding result of the encoding means,
The prediction process performed by the prediction means is a process of predicting by packing one pixel at a time when the image type is multi-valued, and packing a plurality of pixels when the image type is binary,
The prediction process performed by the prediction means calculates a prediction error by subtraction or exclusive OR,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LSB side when the input image is multi-valued, and uses a bit instead of normal quantization when the input image is binary. An image encoding apparatus characterized in that the quantization is such that 1 is changed to 0 regardless of the position of the image. Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means When the data is binary image data, in a second prediction processing mode, prediction means for performing the predetermined prediction processing on the binary image data;
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Code output means for outputting the encoding result of the encoding means,
When the image type is multi-valued, the prediction processing performed by the prediction means is performed at a position in which the peripheral pixels of the pixel to be encoded are matched with the threshold period of the halftoning process when the image type is binary. It is a process to predict using a certain pixel,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LSB side when the input image is multi-valued, and uses a bit instead of normal quantization when the input image is binary. An image encoding apparatus characterized in that the quantization is such that 1 is changed to 0 regardless of the position of the image. In the image decoding apparatus which decodes the code | cord | chord encoded by the image encoding apparatus in any one of Claims 1-3,
A code input means for inputting the code,
Decoding means for decoding the code input by the code input means by a decoding process corresponding to a reverse process of the encoding means;
When the code input by the code input means is obtained by encoding multi-valued image data, an inverse prediction process corresponding to a predetermined prediction process is performed on the decoding process result output by the decoding means. When the multi-valued image data is generated by performing the first inverse prediction processing mode corresponding to the prediction processing mode, and the code input by the code input unit is encoded binary image data, For the decoding processing result output by the decoding means, performing the reverse prediction processing in the second reverse prediction processing mode corresponding to the second prediction processing mode in which the correlation of the binary image data is high, and inverse predictive processing unit for generating the images of the binary,
An image decoding apparatus comprising: an image output unit that outputs a result of the inverse prediction process. Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means If the image data of the data 2 value, in a second prediction process aspect, a prediction means for performing the predetermined prediction process on the image data of the binary,
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Code output means for outputting the encoding result of the encoding means ,
The prediction process performed by the prediction means is a process of predicting by packing one pixel at a time when the image type is multi-valued, and packing a plurality of pixels when the image type is binary,
When the image type is multi-valued, the prediction error is calculated by subtraction. When the image type is binary, the prediction error is calculated by exclusive OR,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LSB side when the input image is multi-valued, and uses a bit instead of normal quantization when the input image is binary. Quantization that changes 1 to 0 regardless of the position of
further,
Code input means for inputting the code;
Decoding means for decoding the code input by the code input means by a decoding process corresponding to a reverse process of the encoding means;
When the code input by the code input means is obtained by encoding multivalued image data, the inverse prediction process corresponding to the predetermined prediction process is performed on the decoding process result output by the decoding means. When the multi-valued image data is generated by performing the first inverse prediction processing mode corresponding to the one prediction processing mode, and the code input by the code input unit is encoded binary image data for decoding result output by the decoding means, the second inverse predictive processing mode corresponding to the first prediction processing mode, perform the inverse predictive processing, to generate the images of the binary Reverse prediction processing means;
Reverse prediction processing means for generating image data by performing reverse prediction processing corresponding to reverse processing of the prediction processing for the decoding processing result;
An image encoding / decoding apparatus comprising: an image output unit that outputs a result of the inverse prediction process. Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means When the data is binary image data, in a second prediction processing mode, prediction means for performing the predetermined prediction processing on the binary image data;
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Code output means for outputting the encoding result of the encoding means,
The prediction process performed by the prediction means is a process of predicting by packing one pixel at a time when the image type is multi-valued, and packing a plurality of pixels when the image type is binary,
The prediction process performed by the prediction means calculates a prediction error by subtraction or exclusive OR,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LS B side when the input image is multi-valued, and is not normal quantization when the input image is binary. Quantization that changes 1 to 0 regardless of the bit position,
further,
Code input means for inputting the code;
Decoding means for decoding the code input by the code input means by a decoding process corresponding to a reverse process of the encoding means ;
When the code input by the code input means is obtained by encoding multivalued image data, the inverse prediction process corresponding to the predetermined prediction process is performed on the decoding process result output by the decoding means. When the multi-valued image data is generated by performing the first inverse prediction processing mode corresponding to the one prediction processing mode, and the code input by the code input unit is encoded binary image data Inversely, the decoding process result output by the decoding means is subjected to the inverse prediction process in the second inverse prediction process mode corresponding to the first prediction process mode to generate a binary image. Prediction processing means;
Reverse prediction processing means for generating image data by performing reverse prediction processing corresponding to reverse processing of the prediction processing for the decoding processing result;
An image encoding / decoding apparatus comprising: an image output unit that outputs a result of the inverse prediction process. Image input means for inputting image data;
When the image data input by the image input means is multivalued image data, a predetermined prediction process is performed on the multivalued image data in a first prediction processing mode, and the image input by the image input means When the data is binary image data, in a second prediction processing mode, prediction means for performing the predetermined prediction processing on the binary image data;
Entropy encoding means for encoding the prediction error quantization result and the prediction result of the prediction means by a predetermined method;
Code output means for outputting the encoding result of the encoding means,
When the image type is multi-valued, the prediction processing performed by the prediction means is performed at a position in which the peripheral pixels of the pixel to be encoded are matched with the threshold period of the halftoning process when the image type is binary. It is a process to predict using a certain pixel,
The prediction error quantization process performed by the prediction means ignores a predetermined bit on the LSB side when the input image is multi-valued, and uses a bit instead of normal quantization when the input image is binary. Quantization that changes 1 to 0 regardless of the position of
further,
Code input means for inputting the code;
Decoding means for decoding the code input by the code input means by a decoding process corresponding to a reverse process of the encoding means;
When the code input by the code input means is obtained by encoding multivalued image data, the inverse prediction process corresponding to the predetermined prediction process is performed on the decoding process result output by the decoding means. When the multi-valued image data is generated by performing the first inverse prediction processing mode corresponding to the one prediction processing mode, and the code input by the code input unit is encoded binary image data Inversely, the decoding process result output by the decoding means is subjected to the inverse prediction process in the second inverse prediction process mode corresponding to the first prediction process mode to generate a binary image. Prediction processing means;
Reverse prediction processing means for generating image data by performing reverse prediction processing corresponding to reverse processing of the prediction processing for the decoding processing result;
An image encoding / decoding apparatus comprising: an image output unit that outputs a result of the inverse prediction process. An image type specifying unit that specifies whether an input image is binary or multi-valued as an image type, and the prediction process in the prediction unit is a prediction according to the image type specified by the image type specifying unit the image coding apparatus according to any one of claims 1-3, characterized in that it is executed in the process aspect. The prediction processing performed by the prediction unit, the image encoding apparatus according to claim 1 or 2, characterized that predicted that the period of the threshold number of pixels and c Futoningu process of the packing fits. In the prediction process performed by the prediction means, if the predicted value is within a predetermined range with respect to the pixel value of the pixel to be encoded , it is determined that the prediction matches, and otherwise, the prediction error is quantized. the image coding apparatus according to any one of claims 1 to 3, wherein the Turkey turn into. The prediction means includes at least two or more pixel value prediction means for predicting pixel values of the input image;
A prediction error calculating means for calculating an error between a predicted value by a predetermined prediction method and an actual pixel value by a predetermined method;
Selection means for selecting either the pixel value prediction means or the prediction error calculation means, and the selection means selects the prediction means when any of the prediction values matches the actual pixel value, 4. The image encoding apparatus according to claim 1 , wherein when the two do not coincide with each other, the prediction error is selected, and the encoding unit encodes the selection result of the selection unit and the prediction error. The selection means is one of a predictor identifier, a prioritized identifier, and a run-length representation of the prioritized identifier, and when the identifier is run-length representation, 12. The image encoding apparatus according to claim 11 , wherein the encoding means encodes the run length. The prediction means includes at least two or more pixel value prediction means for predicting pixel values of the input image;
At least two or more run counting means for counting the number of consecutive matches between the corresponding prediction value of the pixel value prediction means and the actual pixel value;
A prediction error calculating means for calculating an error between a predicted value by a predetermined prediction method and an actual pixel value by a predetermined method;
A selection unit that selects one of the pixel value prediction unit and the prediction error calculation unit, the selection unit selects the prediction unit that has the longest match, and if neither match, selects the prediction error, said encoding means selection result obtained by said selecting means, either of the preceding claims, characterized in that the encoded run length obtained by the run counting means, and the prediction error obtained by the prediction error calculation An image encoding apparatus according to claim 1. The image decoding device for decoding a code output from the image encoding device according to claim 11 , wherein the inverse prediction means includes at least two or more pixel value prediction means for predicting pixel values of the image,
Prediction error correction means for calculating a correct pixel value from the prediction error value by a predetermined method;
Selection means for selecting either the pixel value prediction means or the prediction error calculation means, and the selection means performs a selection process based on a selection result or a prediction error obtained by the decoding unit. The image decoding apparatus according to claim 2 . An image decoding apparatus for decoding a code output from the image encoding apparatus according to claim 13 , wherein the inverse prediction means includes at least two or more pixel value prediction means for predicting pixel values of the image,
At least two or more run development means for sending control information for the number of runs input to the corresponding pixel value prediction means;
Prediction error correction means for calculating a correct pixel value from the prediction error value by a predetermined method;
Selection means for selecting one of the run expansion means and the prediction error correction means, the decoding unit decodes a selection result, a run length, and a prediction error, and the selection means is based on the selection result. When selecting the run expansion means, the run length expansion process is performed based on the run length. When the prediction error correction means is selected, the prediction error correction process is performed based on the prediction error. the image decoding apparatus according to claim 2, wherein the performing.

Images