JP2015091057A - Error tolerance processing method, and prediction coding/decoding device using the same and color coordinate conversion device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform processing appropriately even if an error is included in at least one of two pieces of data, in a series of processes where the difference of the two pieces of data is taken, and the results are subjected to surplus contraction before a predetermined data conversion process is performed.SOLUTION: A value area reduction process is performed to input data S1. The degree of reduction depends on the degree of an error that can be included in an input signal. Subsequently, the value area after a difference process S2 is determined. Thereafter, surplus contraction is performed conventionally S3. Consequently, representation can be made with the same amount of information as that of each piece of data before taking the difference. Thereafter, a predetermined process, e.g., compression coding, including quantization is performed. The data thus obtained is stored and transmitted S5. Subsequently, the processes reverse from the processes of the past is progressed in the reverse order, thus restoring the original data.

Description

  The present invention relates to an error resilience processing method and a predictive encoding / decoding device and a color coordinate conversion device that employ the method, and in particular, after taking a difference between two data and performing a remainder reduction on the result, The present invention relates to an error resilience processing method for performing a series of processes for performing a predetermined data conversion process, and a predictive encoding / decoding device and a color coordinate conversion device employing the method.

  In predictive coding performed on image information, differential codes are widely used. Specifically, for example, assuming that the predicted pixel value is P and the processing target pixel value is X, the compression rate can be improved by encoding Y = X−P which is the difference. Similarly, the difference code is used in the color coordinate conversion process. Specifically, when each pixel is expressed by RGB information, the compression rate is improved by converting Y = G, U = RG, and V = BG. However, when P, X, R, G, and B are 8-bit information, for example, the differences Y, U, and V are 9-bit information, and the amount of information increases.

  As a countermeasure for the above problem, modulo reduction (hereinafter also referred to as “residue reduction”) processing is widely used when reversible compression is employed as a compression technique. This means that although 9 bits are required to express the difference value as it is, for example, for each predicted pixel value P, the range of the difference always falls within the range of 256, If the value of the difference is appropriately shifted, the fact that it can always be expressed as 8 bits is used. As shown in FIG. 6, the general process is performed by associating values of −255 to −129 with 1 to 127 and values of 128 to 255 with −128 to −1. As a result, the value can be reduced to a value of -128 to 127, that is, a value that can be expressed by 8 bits.

  As a specific example, as shown in FIG. 7, for example, the value range of XP is −P to 255-P, and 9 bits are required for expression. In this case, the value of 128 to 255-P is used. Is associated with -128 to -1-P and values of -P to -129 are associated with 256-P to 127, so that the value range is -128 to 127 and can be expressed in 8 bits.

  The above-described technique is disclosed in, for example, Patent Document 1.

JP 2006-261835 A

  As described above, the remainder reduction process can be employed for a process in which an error is not included in a difference code, for example, a predictive encoding process involving lossless compression at a later stage as described above. In other words, it cannot be adopted for processing that may add an error, such as lossy compression. For example, when the predicted pixel value P is 255 and the target pixel value is 0, the difference value is −255, and as a result of the reduction, +1 is encoded. Here, an error is added to the difference value, and +0 Then, the pixel value after decoding becomes 255, and the image quality is greatly deteriorated. Accordingly, the remainder reduction process is not used for a process that may add such an error. However, even in a process in which an error may be included in a differential code, such as predictive coding that employs irreversible compression, an improvement in compression efficiency and transmission efficiency can be expected if a residual reduction process can be employed.

  The present invention has been made for the above-mentioned circumstances, and the object of the present invention is to take a difference between two data, and after applying a remainder reduction to the result, perform a series of predetermined data conversion processes. In the processing, even when at least one of the two data includes an error, an error resilience processing method capable of effectively restoring the two data when the reverse processing of the series of processing is performed, and its An object of the present invention is to provide a predictive encoding / decoding device and a color coordinate conversion device employing the method.

  In order to achieve the above object, an error resilience processing method according to the present invention is an error resilience processing method in a data processing apparatus, and for each data according to an error that can be included in two input data. The gist is to perform a range reduction process, take a difference between the two data, perform a conversion process based on the remainder reduction for the difference data, and perform a predetermined data conversion process on the obtained result. Here, the predetermined data conversion process is typically a compression encoding process.

  On the other hand, the range reduction process can be performed by deleting data relating to a value having a predetermined width corresponding to the error at both ends of the range.

  Further, it is preferable that the range reduction processing is performed by dividing a predetermined range of the range according to the error into a predetermined number and thinning them out from the range evenly.

  In the range reduction process, it is more preferable that the range is linearly reduced by a predetermined width corresponding to the error.

  Further, an inverse conversion process of the predetermined data conversion process is performed on the data after the predetermined data conversion process obtained by the above process, and the remainder reduction is performed on the obtained data. The gist is to recover the two data by performing an inverse conversion process of the conversion process based on and performing an inverse process of the difference process on the obtained data.

  In order to achieve the above object, the predictive coding apparatus of the present invention is changed in value by a range reduction unit that performs a range reduction process on the input pixel value to be processed and a range reduction process of the range reduction unit. A subtraction unit that subtracts a predicted value from the obtained pixel value and outputs a difference value; a residue reduction unit that outputs the difference value by performing conversion processing based on a residue reduction; and a reduction by the residue reduction unit A quantization unit that performs quantization processing on subsequent data, a variable length coding unit that performs variable length coding on the data quantized by the quantization unit, and a quantization unit that performs quantization The inverse quantization unit that outputs the processed data by performing the inverse process of the quantization, and the range shift process that is the reverse of the remainder reduction unit for the data output from the inverse quantization unit And a reverse remainder reduction unit for recovering the difference value, and the reverse remainder And the difference value from the contraction portion, and summarized in that comprises a prediction unit, an outputting said predicted value based on the past of the processing pixel value itself stored.

  In order to achieve the above object, the predictive decoding apparatus of the present invention receives the data output from the variable length encoding unit in the predictive encoding apparatus and reverses the processing in the variable length encoding unit. A variable-length decoding unit that performs decoding and outputs, and inverse quantization that outputs the data output from the variable-length decoding unit by performing processing reverse to the quantization in the quantization unit A reverse residual contracting unit that recovers the difference value by performing a range shift process on the data output from the inverse quantizing unit and reverse to the residual contracting unit, and the inverse residual contracting unit Output from the difference value and a past processing target pixel value stored by itself, a prediction unit that outputs the prediction value, an addition unit that adds the difference value and the prediction value, and an output from the addition unit In the range reduction unit, The management is summarized as further comprising a range extension to restore the original pixel value by performing the inverse of the expansion process, the.

  In order to achieve the above object, the color coordinate conversion apparatus according to the present invention performs a range reduction process on a plurality of input color pixel values according to an error assumed in the process after color coordinate conversion. A range conversion unit that performs a color coordinate conversion including a difference process on a plurality of color pixel values output from the range reduction unit, and outputs a plurality of other color pixel values; and The gist of the invention is that the coordinate conversion unit includes a residue reduction unit that performs a conversion process based on a residue reduction on the color pixel value obtained by the difference process and outputs the result.

  In order to achieve the above object, the color coordinate conversion apparatus according to the present invention performs a range shift opposite to that of the remainder reduction section with respect to data output from the remainder reduction section in the color coordinate conversion apparatus described above. A reverse residual reduction unit that performs processing and recovers the other plurality of color pixel values, and a coordinate conversion opposite to the coordinate conversion in the coordinate conversion unit is performed on the data output from the reverse residual reduction unit. A coordinate reverse conversion unit that recovers a plurality of color pixel values before processing by the coordinate conversion unit, and the range reduction unit for the plurality of color pixel values output from the coordinate reverse conversion unit. The gist of the present invention is to include a range expansion unit that performs an expansion process opposite to the process and recovers a plurality of original color pixel values.

  According to the present invention, when a difference between two data is taken, and after the result is subjected to remainder reduction, in a series of processes for performing a predetermined data conversion process, at least one of the two data includes an error However, when the reverse process of the series of processes is performed, the two data can be effectively restored.

It is a flowchart which shows the process sequence of one Embodiment of the error tolerance improvement processing method of this invention. FIG. 6 is a diagram for explaining a residual reduction process after performing range reduction in an embodiment of the present invention. This is to explain a specific example of how to reduce the range. It is a figure which shows the prediction encoding / decoding apparatus which employ | adopted the error resilience processing method of this invention, The figure (a) is a prediction encoding apparatus, The figure (b) is a block diagram of a prediction decoding apparatus. . It is a figure which shows the color coordinate conversion apparatus which employ | adopted the error tolerance improvement processing method of this invention, the figure (a) is a normal conversion apparatus, and the figure (b) is a block diagram of an inverse conversion apparatus. It is a figure for demonstrating remainder reduction. It is a figure for demonstrating remainder reduction.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a flowchart showing a processing procedure of an embodiment of the error resilience processing method of the present invention. FIG. 2 is a diagram for explaining the remainder reduction process after the range reduction is performed in the embodiment of the present invention. FIG. 3 is a diagram for explaining a specific example of how to reduce the range. Note that the processing shown in FIG. 1 can be realized as software running on a processor or as a hardware circuit.

  A processing procedure of an embodiment of the error resilience processing method of the present invention will be described along FIG. 1 with reference to FIGS. 2 and 3 as appropriate. First, a range reduction process is performed on the input data related to the difference process at the next stage (step S1). For example, a signal related to difference processing is set as a processing target pixel value x and a prediction pixel value p in predictive encoding, each of which is 8-bit data, and the range of the processing target pixel value x is 0 to 255 as shown in FIG. To do. In the range reduction process in step S1, the range is set to m to 255-m. That is, it is reduced by 2 m. As a specific method of this reduction, as shown in FIG. 3A, those having values of 0 to m-1 and 255 to m + 1 to 255 may be simply deleted, or FIG. As shown in FIG. 5, the thinning may be performed evenly at a predetermined interval. More preferably, the original value range x = 0 to 255 may be subjected to a process of xx ((255-2m) / 255) + m to be uniformly compressed to m to 255-m. Further, a predetermined conversion table may be prepared, and a region may be converted and reduced by determining a new value corresponding thereto.

  Next, the range after the difference process is obtained (step S2). As described above, when the predicted pixel value p is known, the range after the differential processing is mp to 255-mp as shown in FIG. However, if it remains as it is, it cannot be expressed in 8 bits, and becomes 9-bit information. Therefore, the conventional residual reduction process is performed (step S3). According to this, as shown in FIG. 2, the value range is in the range of −128 to 127 as in the conventional case, and can be expressed by 8 bits. However, in the present invention, since the range reduction processing is performed as described above, as shown in FIG. 2, as a result, the value of −p is sandwiched between the center and 2 m (= M. The value of the range of “error margin” will not appear. In this way, the value of −p is sandwiched in the center so that a value in the range of 2 m is not taken, so that even if an error is added to the difference value, encoding can be performed properly (they are inappropriately skipped by quantization). A proper value can be restored by subsequent decoding. Therefore, m is set to a value that takes into account errors that may occur in a specific device or the like.

  Note that although the predictive encoding process has been described above, the same applies to the color coordinate conversion process. That is, in the color coordinate conversion, for example, R, G, B data is converted into Y, U, V, but includes a difference process such as U = RG or V = BG, and a remainder. Reduction processing can be performed.

  Returning to FIG. 1, next, in step S4, a predetermined process is performed. The predetermined processing includes compression encoding including quantization. That is, irreversible compression encoding processing is allowed. Then, the obtained data is stored in a storage means such as a memory or transmitted via a communication line (step S5). For example, for subsequent processing on the decoding device or receiving device side, the reverse processing of the processing from step S1 to S4 is performed in reverse order (steps S6 to S9). As a result, the original data can be restored.

Next, an apparatus to which the above-described error resilience processing method of the present invention is applied will be described.
FIG. 4 is a diagram showing a predictive encoding / decoding device that employs the error resilience processing method of the present invention. FIG. 4A shows a predictive encoding device, and FIG. 4B shows a predictive decoding device. It is a block diagram.

  First, referring to FIG. 4A, the predictive coding apparatus shown in FIG. 4 includes a range reducing unit 11, a subtracting unit 12, a remainder contracting unit 13, a quantizing unit 14, a variable length coding unit 15, and an inverse. A quantization unit 16, an inverse remainder reduction unit 17, an addition unit 18, and a prediction unit 19 are provided.

  Therefore, the range reduction unit 11 performs the range reduction process as described above on the input processing target pixel value x. In this predictive coding apparatus, the error margin M (= 2m) as the amount to be reduced at this time is a maximum δ when the quantization coefficient in the quantizing unit 14 is δ. Therefore, it is preferable to set M = 2δ.

  The subtracter 12 subtracts the predicted pixel value p output from the prediction unit 19 from the value changed by the range reduction unit 11 and outputs the result as a difference value d. The residue reduction unit 13 performs the conversion process based on the residue reduction as described above on the difference value d and outputs the result. The quantization unit 14 performs a quantization process on the data output from the remainder reduction unit 13, and the obtained result is variable length encoded by the variable length encoding unit 15 and illustrated as encoded data y. Not stored in memory or transmitted over a communication line. The information stored in the memory or transmitted via the communication line includes information on the margin of error in the processing in the range reduction unit 11. However, if the degree of error margin is determined according to the quantization coefficient, this information need not be sent.

  On the other hand, the inverse quantization unit 16 performs an inverse process of the quantization on the quantized data output from the quantization unit 14 and outputs the result. The inverse remainder contraction unit 17 performs a range shift process on the data output from the inverse quantization unit 16, opposite to the residue reduction unit 13, and recovers the difference value d. The adding unit 18 adds the recovered difference value d and the predicted value p, and supplies them to the prediction unit 19 as past pixel values. The prediction unit 19 stores a series of past pixel values x, and generates a prediction value p from those data and the recovered difference value d.

  Next, referring to FIG. 4B, the decoding apparatus shown in FIG. 4 includes a variable length decoding unit 20, an inverse quantization unit 16, an inverse remainder reduction unit 17, an addition unit 18, a prediction unit 19, And a range extension unit 21.

  Therefore, the variable length decoding unit 20 receives the encoded data, performs decoding that is the reverse process of the processing in the variable length encoding unit 15, and outputs the result. The processes in the next-stage inverse quantization unit 16, inverse remainder reduction unit 17, addition unit 18, and prediction unit 19 are the same as those in the encoding device shown in FIG. Then, the range expansion unit 21 performs an expansion process opposite to the process in the range reduction unit 11 on the data output from the addition unit 18 to recover the original pixel value x.

  As described above, in the predictive encoding / decoding device shown in FIG. 4, even if an error based on quantization is included in the predicted value p, it jumps to an incorrect value through the remainder reduction. There is nothing. Therefore, even if encoding and decoding are performed by such an apparatus, the residual contraction does not adversely affect the restoration of the original data.

  Next, FIG. 5 is a diagram showing a color coordinate conversion apparatus employing the error resilience processing method of the present invention, where FIG. 5A is a block diagram of a normal conversion apparatus, and FIG. 5B is a block diagram of an inverse conversion apparatus. It is. Here, a device for converting the color pixel values R, G, and B into Y, U, and V is shown.

  First, referring to FIG. 4A, the positive conversion apparatus with color coordinate conversion shown in FIG. 4 includes a range reduction unit 31, a coordinate conversion unit 32, and a remainder reduction unit 33.

  Therefore, the range reduction unit 31 performs the range reduction process as described above on the input color pixel value. An error margin M (= 2 m) as an amount to be reduced at this time is generated when the maximum error of each component of processing assumed after color coordinate conversion is δ in a positive conversion apparatus including this color coordinate conversion. The maximum error that can be estimated is 2δ (there is a possibility that both values of the difference target include errors), so it is preferable to set M = 4δ. As the error that may occur, one caused by compression distortion or one caused by noise in the transmission path is assumed.

  The coordinate conversion unit 32 calculates Y = G, U = RG, V = BG based on the input color pixel values R, G, B, thereby converting the converted pixel values Y, U, V. Ask for. The remainder reduction unit 33 performs the conversion process based on the remainder reduction as described above and outputs the color pixel values U and V obtained by the difference process. Thereafter, the data is compressed and encoded and stored in a memory or transmitted via a communication line. The information stored in the memory or transmitted via the communication line includes information on the margin of error in the processing in the range reduction unit 31.

  Next, referring to FIG. 5 (b), the inverse transformation apparatus with color coordinate inverse transformation shown in FIG. 5 includes an inverse remainder reduction unit 41, a coordinate inverse transformation unit 42, and a range expansion unit 43. .

  Therefore, the inverse remainder reduction unit 41 performs a range shift process opposite to that of the residue reduction unit 33 to recover the color pixel values Y, U, and V. The coordinate inverse conversion unit 42 recovers the color pixel values R, G, and B by performing a conversion process opposite to the conversion process in the coordinate conversion unit 32. Then, the range expansion unit 43 performs an expansion process opposite to the process in the range reduction unit 31 on the color pixel values R, G, and B output from the coordinate inverse conversion unit 42 to obtain the original color pixel values. Recover.

  As described above, in the color coordinate conversion apparatus shown in FIG. 5, even if an error due to compression or noise is included in the color pixel value, a large value is skipped through a residual reduction. There is nothing. Therefore, with such a device, residual contraction does not adversely affect the restoration of the original data.

  An error resilience processing method of the present invention, and a predictive encoding / decoding device and a color coordinate conversion device that employ the method are applied to a processing method and device that employs a modulo reduction process after difference processing. Applicable.

11, 31 Range reduction unit 12 Subtraction unit 32 Coordinate conversion unit 13, 33 Remainder reduction unit 14 Quantization unit 15 Variable length encoding unit 20 Variable length decoding unit 16 Inverse quantization unit 17, 41 Inverse remainder reduction unit 18 Adder 42 Coordinate inverse converter 19 Predictor 21, 43 Range extension unit

Claims (10)

An error tolerance processing method in a data processing apparatus,
Depending on the error that can be included in the two input data, perform a range reduction process on each data,
Taking the difference between the two data,
A conversion process based on the remainder reduction is performed on the difference data,
An error resilience processing method, wherein predetermined data conversion processing is performed on the obtained result.   2. The error resilience processing method according to claim 1, wherein the range reduction processing is performed by deleting data relating to a value having a predetermined width corresponding to the error at both ends of the range.   2. The error tolerance reduction process according to claim 1, wherein the range reduction process is performed by dividing a predetermined range of the range according to the error into a predetermined number and thinning them out from the range evenly. Method.   2. The error tolerance processing method according to claim 1, wherein the range reduction process linearly reduces the range by a predetermined width corresponding to the error.   5. The error resilience processing method according to claim 1, wherein the predetermined data conversion process is a compression encoding process. Performing reverse conversion processing of the predetermined data conversion processing on the data after the predetermined data conversion processing obtained by the processing according to claim 1;
The obtained data is subjected to an inverse conversion process of the conversion process based on the remainder reduction,
An error resilience processing method characterized by recovering the two data by performing reverse processing of the difference processing on the obtained data. A range reduction unit that performs a range reduction process on the input pixel value to be processed;
A subtraction unit that subtracts the predicted value from the pixel value whose value has been changed by the range reduction process of the range reduction unit and outputs a difference value;
A remainder contracting unit that performs a conversion process based on a remainder contraction and outputs the difference value;
A quantization unit that performs a quantization process on the data after reduction by the remainder reduction unit;
A variable length coding unit that performs variable length coding on the data quantized by the quantization unit;
An inverse quantization unit that outputs the data quantized by the quantization unit by performing an inverse process of the quantization;
For the data output from the inverse quantization unit, an inverse remainder reduction unit that recovers the difference value by performing a range shift process opposite to the residue reduction unit;
A prediction unit that outputs the prediction value based on the difference value from the inverse remainder reduction unit and a past processing target pixel value stored by itself;
A predictive coding apparatus comprising: 8. A variable-length decoding unit that receives data output from the variable-length encoding unit in the predictive coding apparatus according to claim 7, performs decoding that is a reverse process of the process in the variable-length encoding unit, and outputs the decoded data When,
An inverse quantization unit that outputs the data output from the variable length decoding unit by performing a process reverse to the quantization in the quantization unit;
For the data output from the inverse quantization unit, an inverse remainder reduction unit that recovers the difference value by performing a range shift process opposite to the residue reduction unit;
A prediction unit that outputs the prediction value based on the difference value from the inverse remainder reduction unit and a past processing target pixel value stored by itself;
An adder for adding the difference value and the predicted value;
A range expansion unit that performs an expansion process opposite to the process in the range reduction unit on the data output from the addition unit to recover the original pixel value;
A predictive decoding apparatus comprising: A range reduction unit that performs a range reduction process on a plurality of input color pixel values according to an error assumed in the process after color coordinate conversion;
A coordinate conversion unit that performs color coordinate conversion including difference processing on a plurality of color pixel values output from the range reduction unit, and outputs other color pixel values;
In the coordinate conversion unit, a residue reduction unit that performs a conversion process based on a residue reduction and outputs the color pixel value obtained by the difference process; and
A color coordinate conversion device comprising: 10. The data output from the residual reduction unit in the color coordinate conversion apparatus according to claim 9 is subjected to a range shift process opposite to that of the residual reduction unit to recover the other plurality of color pixel values. An inverse remainder contraction part;
Coordinates for recovering a plurality of color pixel values before processing by the coordinate conversion unit by performing coordinate conversion opposite to the coordinate conversion in the coordinate conversion unit for the data output from the inverse remainder reduction unit An inverse transform unit;
A range expansion unit that performs an extension process opposite to the process in the range reduction unit for the plurality of color pixel values output from the coordinate inverse transform unit to recover the original plurality of color pixel values;
A color coordinate conversion device comprising:

Images