JPH04320161A - Coder and method for dpcm and decoder and method for dpcm - Google Patents

Abstract

PURPOSE:To prevent occurrence of a quantization error causing a slope overload in the DPCM method. CONSTITUTION:A digital picture data is quantized and coded by a DPCM coding circuit 10 by using a digital picture data as a 1st fixed value. A quantization error discrimination circuit 32 discriminates whether or not a quantization error is in existence, and a DPCM coding circuit 20 generates a correction data when any quantization error is in existence. A picture in which deterioration in the picture is suppressed is obtained by adding the correction data to a digital picture data subject to DPCM coding.

Description

[Detailed description of the invention]

0001

[Industrial Field of Application] This invention is based on fixed-length prior prediction differential coding (DPCM=Differential Pulse Coding).
The present invention relates to a DPCM decoding device and method, and a DPCM decoding device and method.

0002

2. Description of the Related Art DPCM is a relatively simple data compression algorithm, and is used to compress image data, audio data, and the like. As shown in Equation 1, DPCM uses actual input data St and data obtained by prediction *
Difference data Et between S(t=1) (usually created using previous input data) is quantized and encoded using a predetermined fixed value as an upper limit.

0003

[Formula 1] Et = St - * S (t-1) ...Formula 1

0004

In DPCM, the difference data Et is quantized using a predetermined fixed value as the upper limit, so the difference data Et may exceed the predetermined fixed value A as shown in FIG. In this case, the slope
A quantization error called overload occurs.

DPCM is suitably used for compressing image data. By giving the quantization level in the DPCM encoding circuit a nonlinear characteristic that matches human visual characteristics, it becomes possible to express the difference data Et with a small number of bits.

However, in areas where the luminance changes sharply, such as at the edge of an image, that is, where the correlation between adjacent pixels is small, the difference data Et exceeds a predetermined fixed value A, resulting in the above-mentioned slope overload. A phenomenon known as this occurs, resulting in deterioration of image quality. Overall S
Even in an image with a good N ratio, when this slope overload occurs, the deterioration in image quality becomes noticeable due to human visual characteristics.

[0007] Although it is conceivable to adopt a variable length encoding method to prevent such image deterioration, compatibility with image data compressed using a fixed length method cannot be maintained.

An object of the present invention is to maintain compatibility with data compressed by the conventional fixed length method and to prevent slope overload due to quantization errors.

[0009]

[Means for Solving the Problem] FIG. 2 shows a D according to the present invention.
Input data of the PCM encoding device (indicated by a solid curve) Data encoded by the first DPCM encoding circuit (
(shown as a bar graph). A DPCM encoding device according to the present invention will be explained with reference to FIG.

That is, the DPCM encoding device according to the present invention converts the difference data D(t) between the previous first prediction value data P(t-1) and the current input data In(t) into a predetermined first prediction value data. Quantize with a fixed value A as the upper limit, create primary encoded data by encoding the quantized data, and dequantize the quantized data,
The first D is added to the previous first predicted value data P(t-1) to create the current first predicted value data P(t).
PCM encoding circuit, the previous first predicted value data P(t-
1) means for determining whether the difference data D(t) between the current input data In(t) and the current input data In(t) is larger than a predetermined first fixed value A, and the previous first predicted value data P(t-
1) When it is determined that the difference data D(t) between the current input data In(t) and the current input data In(t) is larger than the above-mentioned predetermined first fixed value A, the current input data In(t) and the above-mentioned first means for calculating the difference from the current first predicted value data P(t) created by the DPCM encoding circuit No. 1 as the current quantization error data QE(t); means for creating position data representing a position on an input data string of The quantized data is encoded to create secondary encoded data, and the quantized data is dequantized and added to the previous second predicted value data to create the current second predicted value data. 2
The present invention is characterized in that it includes a second DPCM encoding circuit that creates predicted value data.

[0011] The first DPCM encoding circuit and the second D
The PCM encoding circuit and the DPCM encoding circuit can also be realized by one DPCM encoding circuit. In this case, first,
The PCM encoding circuit creates primary encoded data from the input data, stores the quantization error data calculated by the quantization error data calculation means in the creation of the primary encoded data in the memory, and then stores the quantization error data calculated by the quantization error data calculation means in the memory. Secondary encoded data is created by reading out quantization error data from the DPCM encoding circuit and applying it to the DPCM encoding circuit.

The DPCM encoding method according to the present invention is also shown in FIG.
To explain with reference to , the difference data D(t) between the previous first predicted value data P(t-1) and the current input data In(t) is set to a predetermined first fixed value A as an upper limit. By quantizing and encoding the quantized data, primary encoded data is created, and at the same time, the quantized data is dequantized and the previous first predicted value data P(t-
1) Added to the current first predicted value data P(t)
is created, and the difference data D(t) between the previous first predicted value data and the current input data In(t) is the predetermined first predicted value data.
is larger than the fixed value A, and the previous first
predicted value data P(t-1) and current input data In
(t) and the current input data In(t) and the current first predicted value data P(t) when it is determined that the difference data D(t) is larger than the predetermined first fixed value A.
The difference between the current quantization error data QE(t) is calculated, position data representing the position of the above encoding error data QE(t) on the input data string is created, and the difference is calculated as the current quantization error data QE(t). Quantize the difference data between the current quantization error data QE(t) and the current quantization error data QE(t) using a predetermined second fixed value as the upper limit, and create secondary encoded data by encoding the quantized data. The method is characterized in that the quantized data is dequantized and added to the previous second predicted value data to create the current second predicted value data.

The DPCM decoding device according to the present invention includes a first DPCM decoding circuit that decodes primary encoded data obtained by DPCM encoding input data,
DP is the quantization error that occurs when creating the next encoded data.
A quantization error occurs according to a second DPCM decoding circuit that decodes secondary encoded data obtained by CM encoding, and position data representing the position on the input data string where the quantization error occurs. In position,
A correction means is provided for restoring original data by correcting data obtained by the first DPCM decoding circuit using corresponding data obtained by the second DPCM decoding circuit.

[0014] The first DPCM decoding circuit and the second DP
The CM decoding circuit and the DPCM decoding circuit can also be realized by one DPCM decoding circuit. In this case, first, the above DPCM
The primary encoded data is decoded by the decoding circuit to obtain primary data, the primary data is then stored in a memory, and the secondary encoded data is then decoded by the DPCM decoding circuit to obtain correction data, By supplying this correction data to the correction means, the correction means is caused to perform the correction processing.

The DPCM decoding method according to the present invention creates primary data by decoding primary encoded data obtained by DPCM encoding input data;
Correction data is created by decoding the secondary encoded data obtained by DPCM encoding the quantization error that occurs in the creation of the above primary encoded data, and According to the position data representing the position, the original data is restored by correcting the primary data using the correction data at the position where the quantization error has occurred.

[0016]

[Operation] According to the DPCM encoding device and method according to the present invention, the difference data D(t) between the previous first predicted value data P(t-1) and the current input data In(t) is obtained.
is quantized using the first fixed value A as an upper limit and encoded, thereby creating primary encoded data. When the difference data D(t) is larger than the first fixed value A, the difference between the current input data In(t) and the current first predicted value data P(t) is the quantization error data QE(t).
), and this quantization error data QE(t)
Similarly, secondary encoded data is created by DPCM encoding.

Furthermore, position data representing the position on the data string where the quantization error has occurred is created.

According to the DPCM decoding apparatus and method according to the present invention, primary data is created by decoding the primary encoded data, and correction data is created by decoding the secondary encoded data. Then, the original data is restored by correcting the primary data using the correction data at the position where the quantization error has occurred, according to the position data representing the position where the quantization error has occurred.

[0019]

[Effect of the invention] As described above, according to this invention,
Quantization errors that occur in fixed-length DPCM encoding are also DPCM encoded, and data is corrected using this quantization error in DPCM decoding, making it possible to prevent slope overload from occurring.

In particular, when the present invention is applied to image data, images can be correctly reproduced even in areas where brightness changes sharply, and high image quality can be maintained.

Furthermore, since circuit parts other than those related to quantization error processing can be used as they are for conventional DPCM encoding and decoding, compatibility with data compressed by conventional DPCM encoding can also be maintained.

[0022]

Embodiments Hereinafter, an embodiment in which the present invention is applied to image data compression will be described in detail.

FIG. 1 is a block diagram showing an embodiment of a DPCM encoding device according to the present invention, and FIG.
It is a graph showing the relationship between input image data of the M encoding device and encoded data.

The DPCM encoding device includes two first DPs.
CM encoding circuit 10 and second DPCM encoding circuit 2
Contains 0. The first DPCM encoding circuit 10 encodes input image data using the DPCM method. The second DPCM encoding circuit 20 further converts the quantization error generated in the DPCM encoding circuit 10 into DPCM
This is a circuit that encodes and creates correction data.

A clock signal (not shown) is applied to the DPCM encoding device shown in FIG.
Except for the encoding circuit 20, operations and data transfer in each block are performed in synchronization with this clock signal.

Input image data is applied to a subtracter 11 of a first DPCM encoding circuit 10 . Image data is provided to the DPCM encoder in a fixed order (generally horizontal and vertical scan order) with respect to position on the screen.

The DPCM encoding circuit 10 includes a latch circuit 1.
This latch circuit 15 contains predicted value data P(t-1) at the previous sampling time (t=1).
is retained. The subtracter 11 represents the difference between the image data In(t) at the current sampling time t input to the DPCM encoding circuit 10 and the previous predicted value data P(t-1) held in the latch circuit 15. First difference data D(t) is calculated. This first difference data D(t) is given to the quantization circuit 12.

The quantization circuit 12 is a circuit that quantizes input first difference data D(t) with a first fixed value A as an upper limit, and the quantized first difference data D(t) is The signal is applied to an encoder 16 and encoded. This encoded data becomes primary encoded output data of the first DPCM encoding circuit 10.

The first signal quantized by the quantization circuit 12
The difference data D(t) is also given to the dequantization circuit 13 and dequantized. The adder 14 adds the previous predicted value data P(t-1) and the output dequantized data of the dequantization circuit 13 to obtain the current predicted value data P.
(t) is calculated and given to the latch circuit 15 and held.

[0030] It is determined whether a quantization error has occurred in the encoding operation in the first DPCM encoding circuit 10, and if there is a quantization error, the position of the image data that has caused the quantization error is identified. In order to
is provided.

The first difference data D(t) of the first DPCM encoding circuit 10 is provided to a quantization error determination circuit 32. The quantization error determination circuit 32 stores a first fixed value A indicating the upper limit of quantization in the quantization circuit 12, and the quantization error determination circuit 32 determines the input first difference data D(t). It is determined whether or not is larger than the first fixed value A, that is, whether a quantization error has occurred.

On the other hand, the input image data In(t) is held in the latch circuit 31 for one clock period, and then provided to the quantization error determination circuit 32. At this point, the first D
In the PCM encoding circuit 10, the current predicted value data P(t)
is created, this predicted value data P(t) is also input to the quantization error determination circuit 32. The quantization error determination circuit 32 determines that the first difference data D(t) is larger than the first fixed value A (that is, a quantization error has occurred).
When it is determined that the current image data In(t)
Quantization error data QE(t) is calculated by subtracting the current predicted value data P(t) from the current prediction value data P(t), and this quantization error data QE(t) is provided to the switch circuit 32. If the first difference data D(t) is greater than or equal to the first fixed value A, a control signal is given from the quantization error determination circuit 32 to the switch circuit 33. By closing the switch of the switch circuit 33, the quantization error data QE(t) is provided to the second DPCM encoding circuit 20.

When a quantization error occurs, the bit map data memory is sent from the quantization error determination circuit 32 to specify the position on the image data string of the input image data that caused the quantization error. 34 is given one bit of data. The bit map data memory stores 1-bit data indicating whether there is a quantization error depending on the position (address) of the image data. For example, data 1 is for image data that has caused a quantization error, and data 0 is for image data that has no quantization error.
are stored in the form of a map.

The second DPCM encoding circuit 20
It has the same configuration as the PCM encoding circuit 10, and the second
The difference data between the predicted value data and the current quantization error data is quantized and encoded and output as secondary encoded data.

The DPCM encoding circuit 20 includes a latch circuit 2.
5, and this latch circuit 25 holds second predicted value data created based on the previous quantization error data. When the current quantization error data QE(t) is input to the DPCM encoding circuit 20, the subtracter 21
Then, second difference data representing the difference between this quantization error data and the previous predicted value data is calculated. This second
The difference data is given to the quantization circuit 22.

The quantization circuit 22 is a circuit that quantizes the input second difference data with a second fixed value as the upper limit, and the quantized second difference data is given to the encoder 26 and encoded. be converted into The encoded data is given to the memory 27 and stored as secondary encoded data.

The quantized second difference data is also provided to the inverse quantization circuit 23. The second difference data is dequantized by the dequantization circuit 23 and provided to the adder 24 .

The adder 24 adds the previous predicted value data and the output dequantized data from the inverse quantization circuit 23 to calculate the current predicted value data, which is provided to the latch circuit 25 and held.

Each time the quantization error measuring circuit 32 determines that there is a quantization error, the circuit 32 sends a second clock pulse.
is applied to the DPCM encoding circuit 20 of the second DPCM
Encoding circuit 20 performs the operations described above in response to this clock pulse.

The selector switch 40 is always set to the first DPCM.
Output data of the encoding circuit 10 is selected. Therefore, each time primary encoded data is output from the first DPCM encoding circuit 10, this primary encoded data is output to the outside via the changeover switch 40. This primary encoded data may be stored in a memory or transmitted to a reproduction side device via a transmission path.

First DP for one screen worth of image data
When the encoding process by the CM encoding circuit 10 is completed, the bit map is completed in the bit map data memory 34, and the secondary encoded data is stored in the memory 27. When sending out one screen worth of primary encoded data is completed, the selector switch 40 is switched to select the output of the Huffman encoding circuit 35. The bit map data stored in the bit map data memory 34 is Huffman encoded by a Huffman encoding circuit 35, thereby compressing the data and output via a changeover switch. Since bit map data is considered to include many data 0s, considerable data compression is expected by this Huffman encoding.

After this, the changeover switch 40 switches to the second DPC.
It is switched to select the M encoding circuit 20. Second
The secondary encoded data stored in the memory 27 of the DPCM encoding circuit 20 is outputted through the changeover switch 40.

In the above, it is also possible to realize this function with the first DPCM encoding circuit 10 without providing the second DPCM encoding circuit 20. In this case, the quantization error data calculated by the quantization error determination circuit 32 are temporarily stored in the memory in the order in which they appear. After the encoding process for one screen worth of image data in the first DPCM encoding circuit 10 is completed, the quantization error data stored in the memory is transferred to the first DPCM encoding circuit 10 according to the order.
is applied to the DPCM encoding circuit 10 of. 1st DPC
The secondary encoded data created by the M encoding circuit 10 may be sequentially output via the changeover switch 40 each time it is created, or it may be stored once in a memory and then output.

In the above, since the first DPCM encoding circuit is the same as the conventional DPCM encoding circuit, the DPCM encoding apparatus shown in FIG.
PCM encoded data can also be created.

FIG. 3 is a block diagram showing an embodiment of a DPCM decoding apparatus according to the present invention.

This DPCM decoding apparatus includes a DPCM decoding circuit 50, which is commonly used for decoding primary encoded data and secondary encoded data.

First, primary encoded data is given to the DPCM decoding device and input to the DPCM decoding circuit 50.

The latch circuit 53 included in the DPCM decoding circuit 50 holds the previous first predicted value data. The current primary encoded data input to the DPCM decoding circuit 50 is given to the decoded data table 51, where it is dequantized and decoded and output as the current first difference data. This first difference data is sent to the adder 52.
the previous first output from the latch circuit 53.
The current image data is restored by adding it to the predicted value data. The restored image data is given to the memory 61 and temporarily stored.

When the decoding process of one screen's worth of primary encoded data is completed, Huffman encoded bit map data is input to the DPCM decoding device. This bit
The map data is decoded by Huffman decoding circuit 71 and stored in memory 72.

Subsequently, the secondary encoded data is sequentially applied to the DPCM decoding circuit 50 of the DPCM decoding device, and is decoded in the same manner as the primary encoded data to become quantization error data.

The location where the quantization error occurs is located in the memory 72.
is represented by bit map data stored in . The CPU 73 uses an adder 62 to add the image data stored in the memory 61 and the decoded quantization error data corresponding to that position only for positions where quantization errors occur, and stores the addition result in the memory 63. control so that it is stored at the corresponding location (address). This addition process corrects the quantization error that occurred in the first DPCM encoding. For image data without quantization errors, the image data in the memory 61 is directly transferred to the corresponding location (address) in the memory 63 and stored therein. In this way, almost completely restored image data is finally stored in the memory 63.

In the DPCM decoding circuit 50, the decoded data table 51 can be used in common for primary encoded data and secondary encoded data. A decoded data table for the next encoded data and a decoded data table for the secondary encoded data may be separately prepared and switched between them. Furthermore, separate DPCM decoding circuits may be provided for primary encoded data and secondary encoded data.

In the above, since the DPCM decoding circuit 50 can also be applied to data encoded by conventional DPCM encoding, this DPCM decoding device must maintain compatibility with conventional DPCM encoded data. Can be done.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a DPCM encoding device.

FIG. 2 is a graph showing data input to a DPCM encoding device and predicted value data.

FIG. 3 is a block diagram of a DPCM decoding device.

FIG. 4 is a graph for explaining that slope overload occurs.

[Explanation of symbols]

10, 20 DPCM encoding circuit 32 Quantization error judgment circuit 50 DPCM decoding circuit 73 CPU

Claims (8)

[Claims] [Claim 1] The difference data between the previous first predicted value data and the current input data is quantized using a predetermined first fixed value as the upper limit, and the quantized data is encoded to obtain a primary code. A first DP that creates quantized data, dequantizes the quantized data, and adds it to the previous first predicted value data to create the current first predicted value data.
CM encoding circuit; means for determining whether the difference data between the previous first predicted value data and the current input data is larger than a predetermined first fixed value; The difference data with the input data is the predetermined first data.
When it is determined that the current input data is larger than the fixed value, the difference between the current input data and the current first predicted value data created by the first DPCM encoding circuit is calculated as the current quantization error data. means for generating position data representing the position of the quantization error data on the input data string; and means for generating position data representing the position of the quantization error data on the input data string; Quantize with a fixed value as the upper limit, and create secondary encoded data by encoding the quantized data, inversely quantizing the quantized data, and adding it to the previous second predicted value data. The second DPC creates the current second predicted value data.
A DPCM encoding device equipped with an M encoding circuit. Claim 2: The first DPCM encoding circuit and the second DPCM encoding circuit.
The DPCM encoding circuit is realized by one DPCM encoding circuit, and first, the DPCM encoding circuit creates primary encoded data from input data, and in creating this primary encoded data, the quantization error is eliminated. The quantization error data calculated by the data calculation means is stored in the memory, and then the quantization error data is read out from the memory and the D
2. The DPCM encoding apparatus according to claim 1, further comprising means for controlling the creation of secondary encoded data by supplying the second encoded data to a PCM encoding circuit. 3. The DPCM encoding apparatus according to claim 1, further comprising means for encoding the position data created by the position data creation means. 4. Output primary encoded data of the first DPCM encoding circuit, position data created by the position data creation means, and output secondary encoded data of the second DPCM encoding circuit. 2. The DPCM encoding apparatus according to claim 1, further comprising a switching means for switching and outputting. [Claim 5] The difference data between the previous first predicted value data and the current input data is quantized using a predetermined first fixed value as an upper limit, and the quantized data is encoded to obtain a primary code. At the same time, the quantized data is dequantized and added to the previous first predicted value data to create the current first predicted value data, and the previous first predicted value data is created.
It is determined whether the difference data between the predicted value data and the current input data is larger than a predetermined first fixed value, and the difference data between the previous first predicted value data and the current input data is the predetermined value. When it is determined that the current input image data is larger than the first fixed value, the difference between the current input image data and the current first predicted value data created by the first DPCM encoding circuit is calculated as the current quantization error. create position data that represents the position of the quantization error data on the input data string, and calculate the difference data between the previous second predicted value data and the current quantization error data as a predetermined second predicted value data. Quantize with a fixed value as the upper limit, and create secondary encoded data by encoding the quantized data, inversely quantizing the quantized data, and adding it to the previous second predicted value data. DPCM to create the second predicted value data for this time.
Encoding method. [Claim 6] A first DP that decodes primary encoded data obtained by DPCM encoding input data.
a CM decoding circuit, a second DPCM decoding circuit that decodes secondary encoded data obtained by DPCM encoding a quantization error occurring in the creation of the primary encoded data, and an input in which the quantization error occurs. According to the position data representing the position on the data string, the data obtained by the first DPCM decoding circuit is converted to the data obtained by the second DPCM decoding circuit at the position where the quantization error occurs, using the corresponding data obtained by the second DPCM decoding circuit. A DPCM decoding device comprising a correction means for restoring original data by correction. 7. The first DPCM decoding circuit and the second DPCM decoding circuit are realized by one DPCM decoding circuit, and first, the DPCM decoding circuit decodes the primary encoded data to obtain the primary data. After that, the primary data is stored in the memory, and then the secondary encoded data is stored in the DP.
7. The DPCM decoding apparatus according to claim 6, further comprising means for decoding by a CM decoding circuit to obtain correction data, and controlling the correction means to perform the correction processing by supplying the correction data to the correction means. . 8. Create primary data by decoding primary encoded data obtained by DPCM encoding input data, and DPCM encode the quantization error that occurs in creating the primary encoded data. Correction data is created by decoding the secondary encoded data obtained by decoding, and according to the position data representing the position on the input data string where the quantization error occurs, at the position where the quantization error occurs, A DPCM decoding method that restores original data by correcting the primary data using the correction data.