JP2001101427A - Image compressing method, image compressing device, image expanding device and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce a file size after compression without damaging the quality of an image being a source image. SOLUTION: An image file 1 is separated into a layer 3 expressing the area and shape of an image, a layer 4 expressing the line and painting-out part of the image and a layer 4 replacing the line and the painting-out part with the same color as that of a proximate part, then an optimal compressing method is applied to each of layers.

Description

DETAILED DESCRIPTION OF THE INVENTION

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression method, an image compression apparatus, an image decompression apparatus for compressing an image having a relatively limited number of colors, such as a manga or an animation.
It relates to a recording medium.

2. Description of the Related Art In recent years, comics, animation images, sounds, and the like, which have been distributed on paper media such as magazines and books, have been distributed through networks such as the Internet. Moving image data such as manga image data and animation is composed of a very large number of images, and if you try to distribute the information of each image as it is, a very large data capacity will be required . Therefore, an image compression method and an image compression device for compressing such image data so as not to reduce the amount of information as much as possible, an expansion device for expanding the image thus compressed, and recording the image thus compressed Recording media have been proposed. FIG. 6 is a block diagram showing a system for distributing cartoon or animation contents. According to the figure, an original image scanning step of taking in an original image of a manga or animation by a scanner or the like, a retouching step of correcting the taken-in original image, and a coloring step of coloring the retouched original image as necessary are performed. After that, in order to distribute the data as electronic data, a complete package file including a collection of cartoon or animation image files is generated through an authoring process, which is an editing operation, and a formatter process, which converts the file into a desired file format. This complete packet file is recorded on a recording medium such as a CD-ROM or stored in a server for distribution through a network such as the Internet, as necessary. The server creates a file for distribution using a file builder and distributes it to each terminal using distribution software so that these complete packet files can be distributed to the user under optimal conditions according to the user's playback environment and communication environment. Will be delivered. On the user side, through a network such as the Internet, a desired cartoon or animation image is reproduced by a reproduction application in a personal computer, plug-in software added to an Internet browser, a dedicated multimedia player, or the like. In addition, sound such as effect effects of the image, dialogue, BGM, etc. may be added in accordance with such an image. In such a system, as an image compression method that can be put to practical use even in a low transfer rate environment on the user side such as a general public line, a method of conventionally performing compression in a bitmap format such as JPEG or GIF, a vector graphics method, or the like. Various image compression methods have been proposed. However, such a system has sufficiently satisfactory quality and compression performance in the image quality and file size of the distributed content, and
There was no compression scheme optimized for these contents. In particular, when the image size of one episode of a manga or animation is compressed using the above-described compression method with respect to the file size of the content to be distributed, the file size becomes enormous, and although the distribution is taken into consideration. Did not. Further, the file size required on the server side for storing these files also becomes enormous as the number of content titles increases, and it is difficult to cope with them.
For example, in the case of an image compression method based on JPEG, this is a method including a conversion loss, and much information is lost at the time of compression. Also, GIF which is lossless coding
Since PNG and PNG use a palletized color table, they greatly depend on the consistency of the color configuration used in the image. In other words, in the distribution of comics and the like, each frame is displayed on the same screen one after another. Therefore, when compressing and encoding an image in units of frames, if a plurality of frames are simultaneously displayed on one page, the color used in each frame is changed. Is limited. In the case of an image compression method using vector graphics, if used properly, the file size can be reduced to some extent and an image can be rendered with a wide range of resolutions. However, if the artwork was not previously drawn in vector format, conversion to vector format would have to be done manually by a skilled artist or technician, which is very inefficient. Also, it is difficult to accurately render complex and subtle details of the artwork using vector artwork without making the necessary significant changes from time to time. Although there are tools to automate these processes, they have not been of sufficient quality without additional manual modifications. JPE mentioned above
A compression algorithm using DCT (Discrete Cosine Transform), such as the G method, divides an image into small blocks, removes redundant information or information that is too fine and unnecessary, and performs a process for each of the small blocks described above. , For converting image information into a color space different from the original color space. Also,
Since compression algorithms based on DCT are primarily aimed at optimally compressing natural images such as photographs, these compression converters require subtle differences over various blocks when trying to achieve high compression ratios. About the part
They are truncating with some choice. Furthermore, even within each block, small parts are blurred,
Since information may be completely lost during the compression process, it can be said that these compression methods are lossy compression methods. On the other hand, there is a compression method called run-length coding, which supports various color depths while retaining detailed information of the original image.
NCY information and control information can also be described. However, in the compression method based on run-length coding, the compression ratio greatly depends on the consistency of the colors included in the original image, so that the file size increases as the original image becomes more complicated. Also, S like GIF
The compression method based on the string key has a problem similar to that of the run-length coding, but can compress a complicated image to some extent. However, this compression method requires more CPU power.

As described above, in the conventional distribution system, a conventional JPEG, GIF, or the like is used as an image compression method that can be practically used even under a low transfer rate environment on the user side such as a general public line. Various image compression methods such as a compression method using a bitmap method and vector graphics have been proposed.
There has been no image compression method that has sufficiently satisfactory quality and compression performance and is optimized for these contents. In particular, assuming that the file to be distributed is the size of one episode of a manga or animation, the distribution of content compressed using these conventional image compression methods is very practical from the viewpoint of file size. It was not something. Also, the storage file size required on the server side has become enormous as the number of titles increases.

In order to solve the above-mentioned problems, an image file 1 is divided into a transparent layer 3 representing an area and a shape of an image, a mask layer 4 representing a line and a filled portion of the image, and And an image compression method characterized in that the filled part is separated into a shape layer 5 in which the color of the part adjacent thereto is replaced by the same color as the shape layer 5, and an optimal compression method is applied to each of the layers. is there. Further, the image file 1 is replaced with a transmissive layer 3 representing an area and a shape of an image, a mask layer 4 representing a line and a filled portion of the image, and replacing the line and the filled portion with the same color as a color of a portion adjacent thereto. Layer separation section 2 for separating the transmission layer 3 into a shape layer 5, a first compression encoder 6 for compressing the transmission layer 3 based on a compression algorithm optimized for the transmission layer 3, and optimization for the mask layer 4. A second compression encoder 7 for compressing the mask layer 4 based on the compression algorithm
And a third compression encoder 8 for compressing the shape layer 5 based on a compression algorithm optimized for the shape layer 5
The present invention provides an image compression apparatus having a compression file synthesizing unit 9 for synthesizing the compressed files output from the first compression encoder 6 to the third compression encoder 8. Further, the compressed image file 10 is converted into a transparent layer 3 representing the area and shape of the image, a mask layer 4 representing the lines and painted portions of the image, and the lines and painted portions having the same color as that of the portion adjacent thereto. A file separating unit 12 for separating into a shape layer 5 replaced by a color, a first decompression decoder 13 for performing decompression based on a compression algorithm optimized for the transmission layer 3, and an optimization for the mask layer 4. A second decompression decoder 14 for decompression based on the compression algorithm, a third decompression decoder 15 for decompression based on the compression algorithm optimized for the shape layer 5, and the first decompression decoders 13 to An image decompression device having an image synthesizing unit 19 for synthesizing an image output from the third decompression decoder 15 is provided. Further, the image file 1 is obtained by converting the image file 1 into a transmissive layer 3 representing an area and a shape of an image, a mask layer 4 representing a line and a filled portion of the image, and the line and the filled portion having the same color as the color of a portion adjacent thereto. A recording medium characterized by recording a compressed file (10) separated into a replaced shape layer (5) and compressed by applying an optimal compression method to each layer.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an image compression method according to the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing an embodiment of an image compression apparatus according to the present invention. According to FIG. 1, an image 1 serving as an original image is separated into three layers by a layer separation unit 2. The three hierarchies include a TRANSPARENCY layer 3 indicating an area of a frame of an image serving as an original image, a MASK layer 4 indicating a line-painted area in the image serving as an original image and a (solid-painted) area painted with the same color as the line-work, and the above-described line-art and fill-in area. This is the SHAPE layer 5 in which the shaded area is replaced with a color close to this area. Then, the image file is compressed using a high-efficiency encoding algorithm so that the optimum compression efficiency is obtained for each of these layers. Specifically, TRANSPARENCY
The layer 3 is compressed using a compression encoder 6 optimized based on the characteristics of the data of this layer, and the MASK layer 4 is compressed using a compression encoder 7 optimized based on the characteristics of the data of this layer.
Then, the SHAPE layer 5 is compressed by the compression encoder 8 optimized based on the characteristics of the data of this layer. In this way, the files compressed for each layer are packaged by the file formatter 9 to become one compressed package file 10. Next, a method of decompressing a compressed file to obtain an image serving as an original image will be described with reference to FIG. First, the compressed file 10 is subjected to TRANSPARENC by the file analysis / separation means 12.
The data is analyzed and separated into Y layer data, MASK layer data, and SHAPE layer data. And TRANSPARE
The data of the NCY layer is expanded by the expansion decoder 13 based on the characteristics of the data and based on the optimized compression algorithm. Similarly, the data of the MASK layer is decompressed by the decompression decoder 14 based on the characteristics of the data and based on an optimized compression algorithm. Furthermore, SHAPE
The layers are also decompressed by the decompression decoder 15 based on the characteristics of this data and based on an optimized compression algorithm. By such processing, the TRANSPARENCY layer 3, M
The ASK layer 4 and the SHAPE layer 5 are respectively decompressed and recombined by the recombining means 19 to obtain the image 1 serving as the original image. By the way, in cartoons and animations, pictures are often drawn in such a manner that the image fills a certain area with a fixed color or a fixed pattern. Therefore,
Focusing on such characteristics, an image compression method according to the present invention will be described with reference to FIG. 3 taking a manga as an example. Usually, comics are drawn in frame units, and each frame is surrounded by a frame line. The shape of this frame line varies. The TRANSPARENCY layer describes information such as the size and shape of the frame line and the layout position.
First, an image 1 serving as an original image is scanned and converted into data by a scanner or the like (31). Then, the outermost contour of each comic frame is extracted using the data of the image 1 serving as the original image (32). Here, since the image 1 serving as the original image is classified for each frame, frame extraction is performed based on the contour obtained in the above-described process (33). Then, frame images are cut out from the extracted frames in the order in which the reader reads the frames (34). At this time, the frame image may be cut out in an arbitrary order by assigning a number to each frame in the order in which the reader proceeds to read the frame without defining the cutout order, and the frame images may be rearranged in order in a later process. . Next, the frame number, the position of the frame, and the shape of the frame of the frame image forming each frame are detected (35), and TRANSPARE is performed based on the detected information.
An NCY layer file is created (36). This TRAN
The SPARENCY layer file 37 is compressed by a predetermined compression method (44). Here, TRANSPARENC
The reason that the Y layer is necessary is that the shape of the comic frame is not always rectangular. Next, the original image 1 is binarized by a threshold value. In the case of comics, this is done by extracting only a portion corresponding to a black line drawing or a solid black painting drawn with a pen or the like. The line and solid portion are set to 1 and all other portions are set to 0, and the two-tone processing is performed by bringing the threshold value of the luminance of the boundary to be binarized close to the luminance (brightness) of the line and the solid portion. This makes it possible to separate the line and solid-painted portions from other portions. In this way, the binarization is performed (38), a MASK layer file 40 is created (39), and the MASK layer file is compressed by a predetermined compression method (45). Then SH
An APE layer file 43 is created. If the compression algorithm optimized based on the characteristics of the data of the SHAPE layer is, for example, run-length coding that is performed in the horizontal direction with respect to the screen, the SHAPE layer file 43 can be created as follows. As shown in FIG. 4, the base frame image cut out by the frame image cutout 34 and the frame image of the MASK layer are simultaneously linearly scanned in pixel units from the top. The portions a and a 'in FIG. 4 correspond to the scan line a of the original frame image.
Has a shape as shown in FIG. 5A, and the scan line a 'of the MASK layer has a shape as shown in FIG. 5B. Here, the scan line a of the original frame image and the MAS
The scan line a ′ of the K layer is examined one dot at a time from the left. First, since the first dot is black in the MASK layer scan line, it can be seen that this is the MASK area. The MASK area in the base frame image is S
When converted to a HAPE layer file, it is converted to some color data. This color data is performed by replacing the color that appears immediately before or immediately after the MASK area, thereby lengthening the area of the same color in the SHAPE layer, which is advantageous during run-length coding compression.
In this embodiment, since the area is the MASK area from the first dot, the color data can be extracted by looking at the scan line a of the original frame image. That is, as shown in FIG. 5B, the color immediately after is the B color, so that the MASK area (A color) in the scan line a of the original frame image is all B colors.
Replaced by color. Next, looking at the scan line a of the original frame image, after passing through the B color area, the MASK
Area. Immediately after this MASK area, there is an area of C color. Therefore, this MASK area may be replaced with color data of B color or C color. Similarly, the next MASK area is replaced with B color or C color. Further, the next MASK area is a B color both immediately before and immediately after this area, so that it is replaced with the B color as it is. Then, such processing is repeated, and the SHAP as shown in FIG.
A scan line of the E layer can be obtained. If this is run-length coded, it is possible to greatly improve the compression ratio, as is apparent from comparison with the scan line a of the original frame image and the MASK layer scan line a '. In addition, two color regions may be adjacent to each other without the MASK layer interposed therebetween.
No processing is performed until the ASK area is reached. In addition, as described above, when the B color is immediately before the MASK area and the C color is immediately after the MASK area, which color should be replaced may be advantageous during compression. For example, in the case of the GZIP compression method in which it is more advantageous to include the same value as much as possible, the length of the color area may be adjusted within the MASK area so as to obtain the previously existing RUN-LENGTH. The length of the immediately preceding B color area is 10 dots, the length of the MASK area is 7 dots, the length of the C color area is 13 dots, MASK
The area is 2 dots or more and the length of the B color is already 1
If there are 2 dots and the length of the C color is 20 dots,
The color is 12 dots (10 dots from the base + the first two dots in the MASK area), and the C color is 20 dots (the remaining 5 dots in the MASK area + 13 dots from the base + the next MASK area).
(2 dots in the area). Thus S
A HAPE layer file 43 is created (42), and this SHA
The PE layer file 43 is compressed by a predetermined compression method (46). Subsequently, the next frame image is cut out (44), and the same processing is performed for all frames. In the present embodiment, the description has been made using the horizontal scan lines, but the scanning direction is not limited to this. And the compressed TRANSPAREN
CY layer file 37, MASK layer file 40, SHA
Each of the PE layer files 43 is compressed by a predetermined compression method and then synthesized for each frame (47),
The image file 48 is compressed on a frame basis. The image file 48 compressed in units of frames is a text file corresponding to a comic speech bubble, an after-recording audio file in which a text portion is voiced, an audio file such as sound effects and music, and a frame display effect file for applying an effect to a frame. Along with other files 49, etc., to a distribution file formatter 50 for distributing these data. Then, the distribution file 51 is created. In the above-described embodiment, the image compression method in the case where the original image is a color image is described. However, even in the case of a monochrome image, the entire SHAPE layer is set to white and the compression of the SHAPE layer is omitted. Thus, the present invention can be dealt with.

As described in detail above, according to the image compression method of the present invention, an image file is divided into a layer representing an area and a shape of an image, a layer representing a line and a filled portion of the image, By separating the filled part into layers where the colors of the adjacent parts have been replaced with the same color and applying the optimal compression method to each layer, the compressed file can be saved without impairing the quality of the original image. The effect is that the size can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an image compression apparatus according to the present invention.

FIG. 2 is a block diagram showing an embodiment of an image decompression device according to the present invention.

FIG. 3 is a block diagram showing an embodiment of an image compression method according to the present invention.

FIG. 4 is a diagram showing an image serving as an original image and an image of a MASK layer according to the present invention.

FIG. 5 is an enlarged view showing a scanned portion of an image serving as an original image, an image of a MASK layer, and an image of a SHAPE layer according to the present invention.

FIG. 6 is a block diagram illustrating a comic data distribution system.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Original image 2 Layer separation part 3 TRANSPARENCY layer 4 MASK layer 5 SHAPE layer 6,7,8 Compression encoder 9 File formatter 10 Compressed file 12 File analysis and separation part 13,14,15 Decompression part 19 Recomposition part

Claims (4)

[Claims] An image file is replaced with a transmissive layer representing a region and a shape of an image, a mask layer representing a line and a filled portion of the image, and replacing the line and the filled portion with the same color as a color of a portion adjacent thereto. An image compression method, wherein the compression method is applied to each of the layers. 2. The image file is replaced with a transmissive layer representing a region and a shape of the image, a mask layer representing a line and a filled portion of the image, and replacing the line and the filled portion with the same color as a color of a portion adjacent thereto. A layer separation unit that separates the transmission layer into a plurality of shape layers, a first compression encoder that compresses the transmission layer based on a compression algorithm optimized for the transmission layer, and a compression algorithm that is optimized for the mask layer. A second compression encoder that compresses the mask layer by compression, a third compression encoder that compresses the shape layer based on a compression algorithm optimized for the shape layer, a first compression encoder to a third compression encoder. And a compressed file synthesizing means for synthesizing a compressed file output from the compression encoder. 3. A compressed image file is converted into a transmissive layer representing an area and a shape of an image, a mask layer representing a line and a filled portion of the image, and the line and the filled portion having the same color as that of a portion adjacent thereto. File separation means for separating into a shape layer replaced by a color, a first decompression decoder for performing expansion based on a compression algorithm optimized for the transmission layer, and a compression algorithm optimized for the mask layer A second decompression decoder that performs decompression according to a compression algorithm optimized for the shape layer, a third decompression decoder that performs decompression based on a compression algorithm optimized for the shape layer, and outputs from the first to third decompression decoders. An image decompression device having an image synthesizing unit for synthesizing an image. 4. The image file is replaced with a transmissive layer representing a region and a shape of the image, a mask layer representing a line and a filled portion of the image, and replacing the line and the filled portion with the same color as a color of a portion adjacent thereto. And recording a compressed file compressed by applying an optimal compression method to each layer.