JPH04183171A - Image coding and decoding method - Google Patents

Abstract

PURPOSE:To make it possible to carry out coding and decoding using two-dimensional discrete cosine transformation by adding, to picture elements that could not be processed for preparation of a nXn picture element block, a block of picture elements that has been coded once to form and code a nXn picture element block, and, for decoding of picture element, using only the picture elements that could not be processed for preparation of block as picture elements. CONSTITUTION:First, coding of a 8X8 picture element block is carried out sequentially in the main scanning direction, and when the block coding in the main scanning direction is completed, setting of a 8X8 picture element block including the remaining 8X4 picture elements is carried out, followed by its coding. When the block preparation in the subscanning direction is completed, since 4X8 picture elements remain in the subscanning direction, the preceding picture elements are added to the remaining 4X8 picture elements to form a 8X8 picture element block, and processed for coding sequentially in the main scanning direction. In a way similar to the coding, when picture elements are decoded sequentially in the main scanning direction, since the last 8X4 picture elements remain undecoded, preceded decoding regenerated 8X4 picture elements are added to the last 8X4 picture elements left undecoded to decode one block of 8X8 picture elements, and of one block of these decoded picture elements, 4X8 picture elements left unregenerated are regenerated as a picture image. With this, coding/decoding using two-dimensional discrete cosine transformation can be performed.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an image encoding method and an image decoding method for compressing image data, which are applied to image filing devices, image transmission devices, and the like.

2. Description of the Related Art Conventionally, transform encoding using orthogonal transforms such as cosine transform and Hadamard transform has been widely known as a data compression technology for digital image data having gradation. This transform encoding is a method in which an image is divided into small blocks of n×n pixels, orthogonal transform is performed for each block to obtain a transform coefficient matrix, and a quantization level is determined for each component of the matrix. This is shown, for example, in Japanese Unexamined Patent Publication No. 109662/1983.

Problems to be Solved by the Invention However, when encoding is performed using two-dimensional discrete cosine transform (DCT), processing is performed for each n×n pixel block, so the target image is direction) n
It is limited to multiples of .

This point will be explained in detail. FIG. 4 shows a block diagram of the encoding method using DCT. First, the input image is converted into DCT transform unit l in block units of 8×8 pixels.
Perform CT to find DCT coefficients. Here, 8×8 2
In the dimensional DCT transformation, the original pixel is X1□, the transformation coefficient is yu,
Then, it is defined by the following formula.

However, u, v are 0, 1. ~,7 is an integer, and C(
w) is C(w)=2−”” when W=O, w=1.
2. , 7, C(w)=1.

Next, the quantization unit 3 linearly quantizes the CT coefficients using the quantization matrix in the quantization matrix storage unit 2 in which different quantization step sizes are set for each coefficient. Find the coefficient. At this time, the quantization matrix finely quantizes the low-order coefficients in accordance with human visual characteristics, as shown in FIG. 5, for example. In order to control the amount of code or decoded image quality, a scaling factor storage unit 4 is specified from the outside.
Quantization is performed using the value obtained by multiplying the quantization matrix by the scaling factor in the multiplier 5 as the step size. For quantization coefficients, the Huffman encoding unit uses Huffman encoding, which reduces the overall amount of code by assigning short codes to coefficients that appear frequently and long codes to coefficients that appear less frequently. 6, and code data is sent out.

Furthermore, during decoding, the reverse processing of encoding may be performed, and the image can be reproduced by processing in the order of Huffman decoding, inverse quantization, and inverse DCT.

However, when encoding in this way, the number of pixels in the main scanning direction and sub-scanning direction of the image is n=8, as shown in FIG.
If it is not a multiple of , there will be surplus pixels as shown by diagonal lines at the edges of the image, and the image cannot be made into an 8×8 block.

Incidentally, the above-mentioned publication does not take such cases into consideration at all.

Means for solving the problem Image encoding divides an image into blocks of n×n pixels, performs two-dimensional discrete cosine transform on each block, and quantizes and encodes the obtained coefficients. In this method, when the number of pixels in the main scanning direction and the sub-scanning direction is not a multiple of n, the pixels of the block that has been encoded once are added to the remaining pixels that cannot be divided into blocks to create a block of n × n pixels. was formed and encoded.

A corresponding decoding method is an image decoding method in which the code is dequantized for each n×n pixel block and decoded by performing an inverse two-dimensional discrete cosine transform. When the number of pixels in the direction is not a multiple of n,
Decode one block more than the number of blocks that can be taken in the main scanning direction, compensate for the undecoded pixels in the main scanning direction with the pixels of the block that has been decoded one block more, and decode l blocks more than the number of blocks that can be taken in the sub-scanning direction. turned into
The undecoded pixels in the sub-scanning direction are supplemented with the pixels of the decoded block that is one block larger.

Since the image size in the main and sub-scanning directions is not a multiple of n,
For pixels that could not be made into xn pixel blocks, pixels from the previously encoded block are added to form an n×n pixel block and encoded. During decoding, only the pixels that could not be made into blocks are By making it an image,
Even when the image size in the main and sub-scanning directions is not a multiple of n, encoding and decoding using two-dimensional discrete cosine transformation are possible.

Embodiment An embodiment of the present invention will be described with reference to FIGS. 1 to 3. Components that are the same as those shown in FIGS. 4 to 6 are indicated using the same reference numerals.

First, the concept of the system of this embodiment will be explained with reference to FIG. As shown in Figure 6, if n=8 is not a multiple of 8 and there are surplus pixels, after encoding the number of pixels that can be divided into 8x8 pixel blocks, the 8x8 pixels are For pixels that could not be converted into blocks, as shown by diagonal lines in Figure 2, the pixels of the block that has been encoded once are added to form an 8 x 8 pixel block, and then encoded. be. When decoding such a block, only the pixels that could not be made into a block (i.e., the pixels shown with diagonal lines in FIG. 6) are reproduced as an image.
This enables encoding using DCT even when the image size in the main and sub-scanning directions is not a multiple of 8.

Such processing is performed according to the flowcharts shown in FIGS. 1 and 3.

Figure 1 shows the encoding process. First, 8x8 pixel blocks are encoded sequentially in the main scanning direction, and when the blocks in the main scanning direction are completed, 8x4 pixels remain in the main scanning direction. Therefore, the remaining 8×4 pixels are formed into an 8×8 pixel block and then encoded. Such scanning is sequentially performed for each main scanning line block. When the block in the sub-scanning direction is completed, there will be 4 x 8 pixels left in the sub-scanning direction, so add the preceding pixels in the sub-scanning direction that have already been encoded to the remaining 4 x 8 pixels to create 8 x 8 pixels. Pixel blocks are formed and sequentially encoded in the main scanning direction. In this case as well, when a block in the main scanning direction is completed, there will be 8 x 4 pixels left in the main scanning direction, so the preceding encoded pixels are added to these 8 x 4 pixels to form an 8 x 8 pixel block. , performs its encoding.

FIG. 3 is a flowchart showing the process of decoding an image encoded in this manner. First, decode the code and 8×
Regenerate an 8 pixel block. As with encoding, when decoding is performed sequentially in the main scanning direction, the last 8x4 pixels are not decoded, so the 8x4 pixels for one block are added to the previous 8x4 pixels that have been decoded and reproduced. ×8 pixels are decoded, and only the unregenerated 4×8 pixels are reproduced as an image.

In this manner, each main scanning line block is processed in the sub scanning direction. When the block in the sub-scanning direction is finished,
Since 4 x 8 pixels are not reproduced in the sub-scanning direction, one block of 8 x 8 pixels is decoded by adding the 4 lines of decoded and reproduced images in advance in the sub-scanning direction, and among them, Only the unreproduced 4×8 pixels are reproduced as an image. When a block in the main scanning direction is completed, 2 x 2 pixels at the bottom right remain, so add the decoded and reproduced pixels that precede it in the main and sub scanning directions to decode one block of 8 x 8 pixels. , among which only the unregenerated 2×2 pixels are reproduced as an image.

Effects of the Invention In the present invention, as described above, since the image size in the main and sub-scanning directions is not a multiple of n, for pixels that cannot be made into an n×n pixel block, pixels from a block that has been encoded once are added. By forming n×n pixel blocks and encoding them, and decoding, only the pixels that could not be made into blocks are used as images, even if the image size in the main and sub-scanning directions is not a multiple of n. However, it is possible to perform encoding and decoding using two-dimensional discrete cosine transform.

[Brief explanation of the drawing]

1 to 3 show an embodiment of the present invention,
Figure 1 is a flowchart showing the encoding process, Figure 2 is an explanatory diagram conceptually showing the method of this embodiment, Figure 3 is a flowchart showing the decoding process, and Figure 4 shows the conventional DCT encoding method. A block diagram, FIG. 5 is an explanatory diagram of a quantization matrix, and FIG. 6 is an explanatory diagram showing a case where it cannot be divided into blocks. ]71 Figure main scanning mark 4 King-scanning

Claims (1)

[Claims] 1. An image code that divides an image into blocks of n×n pixels, performs two-dimensional discrete cosine transformation on each block, and quantizes and encodes the obtained coefficients. In the method, when the number of pixels in the main scanning direction and the sub-scanning direction is not a multiple of n, for the remaining pixels that cannot be divided into blocks,
An image encoding method characterized in that pixels of a block that has been encoded once are added to form a block of n×n pixels, and encoding is performed. 2. In an image decoding method in which the code is dequantized for each block of n×n pixels and decoded by performing inverse two-dimensional discrete cosine transformation, the number of pixels in the main scanning direction and the sub-scanning direction to be reproduced is n. If the number is not a multiple of , one block more than the number of blocks that can be taken in the main scanning direction is decoded, and the undecoded pixels in the main scanning direction are supplemented with the pixels of the decoded block that is one block more than the number of blocks that can be taken in the sub-scanning direction. An image decoding method characterized in that one additional block is decoded, and undecoded pixels in the sub-scanning direction are supplemented with pixels of the one block more decoded.