KR20090124177A - Device for processing image of large size into scaled image tiles and the method therefor - Google Patents

Abstract

PURPOSE: A device for processing a large size image into scaled image tiles and a method therefor are provided to reduce the usage of a memory. CONSTITUTION: An image interpreter(111) writes out an index for a partial access. By using image conversion information necessary for conversion process, a preprocessor(113) generates the tile arrangement information for each scale. Finally, a small unit converter(115) creates a self-image tile by downscaling the transferred tiles after receiving the tiles for original scale from the image analyzer. An output processor(117) transfers collected information to a storage medium after collecting the tile arrangement information inputted in the preprocessor.

Description

DEVICE FOR PROCESSING IMAGE OF LARGE SIZE INTO SCALED IMAGE TILES AND THE METHOD THEREFOR}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly, to a mass image processing apparatus and a method for splitting and converting a large-scale and high-resolution image of mega (M) to g (G) pixel class into multiple scales.

With the development of electronic technologies such as cameras and scanners, the size of image data handled by computers and networks is increasing rapidly. However, storage and transmission technologies that handle and transmit large amounts of video data have not changed much in the last decades.

In general, the size of an image to be transmitted is limited by the network quality of the network and the processing capacity of the server and the client because the quality of service (QoS) must be guaranteed to the user. However, the processing capacity of the network and the client does not increase with the demand and growth of media data. Accordingly, data that is accessible in the network is mainly made up of low capacity and low quality data.

Recent advances in electronic technology have made it possible to generate large-capacity images in units of gigapixels, which are at least 3 gigabytes (even if each byte is allocated to each of R, G, and B channels). 3GB) of image processing space. Even if the storage capacity is reduced to 1/12 using a compression method such as JPEG, the memory usage is at least 3 gigabytes since decoding is required again at the time of using these images.

However, in most cases, a display device capable of displaying a large amount of images expresses much fewer pixels than a gigapixel image, so it must be scaled once again, which incurs a great deal of computation and time loss. Has greatly hurt its responsiveness.

Today, the trend is shifting to 64-bit computing, but up to 4 gigabytes (3 gigabytes in Windows because of its architecture) can only handle this data on 32-bit computers that are still widely used. In other words, because the microprocessor has a limit on managing the address of memory, an image of 1 gigapixel is the size that the OS can handle. In a 64-bit computing environment, the amount of memory that can be handled is greatly increased, but the actual physical memory footprint is typically between 1 and 2 gigabytes, and even rapidly increasing capacity will not exceed 10 gigabytes in a few years. In such a situation, in the future, the processing and display of gigapixel images will still be expensive, long time, and low service availability.

In order to solve this problem, the Tagged Image File Format (TIFF) standard defines tiling techniques for fast viewing of large images. Most of them do not properly support implementation and lack the definition of scaling. Esau's utility is very low. Because only gigapixel data or an e-book is viewed, the original data is rescaled in real time to match the display resolution of the final terminal, rather than in full-scale. In this process, the full-scale image is always displayed. Because of the need for high-capacity data at all times, the current tiled TIFF (Tiled-TIFF), which lacks the necessary scaling, remains at an unnecessary standard.

1 is a view for explaining a conventional image processing and conversion method, the computer image processing algorithm is to decompose the original image (1) into a plurality of pieces in the memory in a plurality of tile pieces (2) Save the converted tile fragments (split images) to your hard disk.

In addition, by downsampling the original image 1 into a reduced image 3, the converted image is decomposed into a plurality of pieces, converted into a plurality of tile pieces 4, and then the converted tile pieces are converted into a hard disk. Store in

In addition, the downscaled image is converted to a further reduced image by downsampling, the converted image is decomposed into a plurality of pieces, converted into a plurality of tile pieces, and the converted tile pieces are stored in a hard disk. The above method is repeated according to the scale to be reduced.

That is, conventionally, in order to smoothly display a large amount of images, the original image is scaled and tiled in advance to perform multi resolution, multi layer, and multi tiling. However, in the tiling and scaling transformation, which is a necessary preprocessing before displaying an image, the existing method repeatedly performs the scaling and tiling of the original image and the tiling of the scaled image again.

As a result, scaling and tiling operations in memory using a large amount of the original image results in a huge amount of memory space and I / O traffic, resulting in significantly lower processing efficiency. Moreover, when scaling, a large amount of data is used in the memory, and it is almost impossible to process image data beyond a gigapixel.

As described above, the conventional method as shown in FIG. 1 is considerably insufficient in the technical aspect to process a large and super-large image such as a gigapixel technology that has recently been in the spotlight.

An object of the present invention is to provide a large-capacity image processing apparatus and method for maximizing the processing power of the image by not loading the entire original image in the memory in the process of segmental tiling and scaling of the image.

SUMMARY OF THE INVENTION An object of the present invention is to process image by tile instead of the entire image by loading and reducing image tiles having a larger scale than itself in the process of segmented tiling and scaling of an image, and then configuring them. It is to provide a large-capacity image processing apparatus and method thereof that can maximize the.

Another object of the present invention is not to form a plurality of images in sequence when the original image is reduced and converted to an image having various scales, but to form random data simultaneously by forming a plurality of tiles simultaneously corresponding to each other. It is an object of the present invention to provide a large-capacity image processing apparatus and method thereof capable of maximizing processing performance by avoiding redundant read / write.

Technical means of the present invention for achieving the above object is an image analyzer for analyzing the header (Header) of the original image file to create an index for the partial access to process the request for access to the partial information and the size information of the original image; A preprocessor for generating image conversion information necessary for a conversion operation based on the original image-related information input from the image analyzer and generating tile arrangement information for each scale using the generated image conversion information; In response to a tile generation request of an arbitrary scale input from the preprocessor, a higher-level tile corresponding to a tile of an arbitrary scale is called to construct itself by referring to tile arrangement information for each scale, and finally, an original A sub-unit converter which receives the scale tiles from the image analyzer and reduces them to a predetermined ratio to generate their own image tiles; And an output processor which collects the image tile generated by the subunit converter and the tile arrangement information for each scale input from the preprocessor and delivers the tile information to the storage medium.

In detail, the subunit converter may generate its own image tile by retrieving at least two or more tiles from a higher scale and down sampling at a predetermined reduction ratio when generating a tile having an arbitrary scale.

The image interpreter analyzes a header of the original image data, analyzes an image attribute necessary for conversion, format information, compression information, and a data storage method of the original image, and has an index for partial access.

The image conversion information generated by the preprocessor includes a tiling reference size for the size of the tile to be divided, a maximum scale for dividing the original image using the tiling reference size, and a minimum scale and reduction ratio to be reduced to the maximum. Characterized in that it comprises at least one or more.

Each image tile included in the arbitrary scale is characterized in that the image generation operation is performed in parallel with each other.

The output processor adds image-related index information to a Taylor, which is a rear end of the image, when the index information is collected on the generated image tile.

In addition, the large-capacity image processing apparatus is mounted on a web server, in accordance with the user terminal access and image transfer request bundles the generated image tiles into a predetermined number of image tiles and transmits them to the user terminal through a wired or wireless network. do.

The technical method of the present invention for achieving the above object comprises: a first step of creating an index for each size of the original image and the partial region of the original image to access the original image for each partial region; A second step of setting image conversion information including a minimum scale based on the original image related information; A third step of sequentially reducing the current scale according to the set reduction ratio until the minimum scale is set, and generating each scaled scale and tile arrangement information for each scale; And a fourth step of generating an image tile by reducing the image tiles of a higher scale to a predetermined ratio by calling image tiles of a higher scale to construct an image tile of the tile according to the tile arrangement information for each scale. do.

Specifically, the image conversion information of the second step, the tiling reference size for the size of the tile to be divided, the maximum scale to distinguish the original image using the tiling reference size, and the minimum scale and reduction to maximize It characterized in that it comprises at least one or more of the ratio.

The third step may further include disposing a plurality of scales for each tile corresponding to each other.

The maximum scale is characterized in that it comprises a virtual region other than the original image to match the magnification when the original image is tiled to a tiling reference size, wherein each scale is obtained by reducing the maximum scale by a set ratio. It is done.

The tile arrangement information for each scale may include map information on image tiles corresponding to each scale.

The image tile generated in the fourth step is further included in the step of collecting the index information associated with the image stored in the storage medium, when generating the image tile in the fourth step, at least Importing two or more tiles and down-sampling at a predetermined reduction ratio may be generated.

When the index information is collected in the generated tile or image, image related index information is added to a Taylor which is a rear end of the image.

Each tile of the lower scale is characterized in that the image conversion and generation is performed independently of each other.

In the fourth step, if the tiles of the higher scale are not configured, it is determined whether the next higher scale tiles corresponding to the specific tile of the higher scale are configured according to the tile arrangement information for each scale, and the corresponding image tiles are called. Characterized in that.

The generated image tiles are transmitted by being bundled into a predetermined number of image tiles through a wired / wireless network according to a request of a user terminal, and linked to a specific position of an image when transmitted in units of image tiles through the wired / wireless network. An additional file such as a video, an image, an animation, a sound, or a 3D may also be transmitted and displayed, and the additional file may be reduced and enlarged at a scale ratio equivalent to that of the image.

As described above, the present invention has a merit that a stable performance can be obtained because the amount of memory can be reduced regardless of the increase in size of the original image.

In addition, by using a recursive method of calling tiles of higher scale to construct one's own tiles, tiles for each scale are processed interdependently so that multi-threaded and multi-process operations can be performed when converting an image. There is an advantage in achieving breakthrough high efficiency and high performance conversion efficiency.

Also, unlike the existing header structure, the Taylor structure can continuously add information to the Taylor of the converted image data so that various information or programs can be added to a single file without changing the data, thereby partially updating. And management, which can be extended to various applications.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

2 is a view showing an image conversion apparatus according to an embodiment of the present invention.

The image conversion apparatus 110 includes an image analyzer 111, a preprocessor 113, a subunit converter 115, and an output processor 117. The image analyzer 111, the preprocessor 113, the subunit converter 115, and the output processor 117 are functionally separated from the image converter 110, and do not mean that they are physically separated from each other.

The image interpreter 111 analyzes the header of the original image data stored in the storage medium, analyzes the size, attributes, and metadata of the original image for conversion, and index information for partial access to process the request for accessing the partial image. It is configured to write.

The preprocessor 113 generates and manages image conversion information and tile arrangement information for each scale necessary for the conversion operation based on the original information received from the image analyzer 111.

That is, the preprocessor 113 specifies a reference size for tiling the original image, compares the tiling reference size with the size of the original image, and specifies the maximum size of the image, and the minimum scale to reduce the original image to the maximum. In this case, the reduction ratio is specified. The maximum size is equal to or larger than the actual size of the original image. The reduced size information is obtained by reducing the maximum size to a set reduction ratio based on the designated image conversion information as described above, and repeats this until it becomes the minimum scale size. Accordingly, scaled scale size information and tile arrangement information for each scale are generated.

Sub-scale conversion is performed by using the generated scale and tile arrangement information for each scale. The sub-scale conversion is a solution of a higher scale to gradually construct an image or tile of an initial scale received from the preprocessor 113 by itself. This is a recursive process that calls our image tiles.

The subunit converter 115 converts the tiles of the upper scale corresponding to the subscale converter 115 to form an image of a minimum scale or a specific image tile according to the scale and tile arrangement information for each scale input from the preprocessor 113. If the requested image tile is a scale of the original size, the corresponding image tile is received from the image analyzer 111 and downsampled at a predetermined reduction ratio to generate a predetermined tile in each sub-unit converter. To the output processor 117.

Here, the image interpreter 111 receives the scale and image tile arrangement information from the preprocessor 113 in advance, takes the image tile at the original scale required by the subunit converter 115, and delivers the image tile.

The output processor 117 collects the tile inputted from the sub-unit converter 115 and the scale information and the position-related index information of the tile transferred from the preprocessor 113 and transmits the collected information to the storage medium 130.

The image conversion apparatus 110 configured as described above is a high-speed and minimum computer memory and resources without limiting the size of the image by using a recursive conversion method in converting a large amount of high resolution images into a segmentation process and a multi-scale form. Allow for efficient conversion. To this end, the image conversion apparatus 110 reads the necessary data before the operation, prepares the conversion, divides the image into recursive small work units, and stores and converts the necessary data.

The image conversion operation according to FIG. 2 will be described in more detail using the flowcharts shown in FIGS. 3, 4, and 6 as follows.

As a preparatory work before converting the image as shown in Figure 3, the image interpreter 111 analyzes the header of the original image stored in the storage medium (S11), the attributes and metadata of the analyzed original image and the size of the original image, etc. The information is collected and stored (S12).

Subsequently, the image interpreter 111 creates an index for partial access in order to process the partial data access request (S13). The index is an index using header information provided according to a format or a compression method of an original image.

Next, the preprocessor 113 generates and manages image conversion information and tile arrangement information for each scale necessary for the conversion operation based on the original information received from the image analyzer 111.

That is, as shown in FIG. 4, the preprocessor designates a reference size for tiling the original image (S21), and designates a maximum size of the required image by comparing the size of the specified tiling reference size with the original image (S22). The maximum size is equal to or larger than the actual size of the original image.

For example, as shown in FIGS. 5A and 5B, when the original image is virtually divided according to a determined tiling reference size, there is no exact case except for an ideal case. That is, it includes virtual regions other than the original image, such as hatched portions. When the image tiles are divided as shown in FIG. 5A, T4, T8, T12, and T13 to T16 may correspond to the image tiles including the virtual area. When the image tiles are divided as shown in FIG. 5B, T1 to T4, T5, T8, and T9. , T12 and T13 to T16 will be the image tiles including the virtual area. Depending on how the original image is tiled, the tiles that contain the virtual region will vary.

Subsequently, a minimum scale to reduce the image to the maximum is specified, and a reduction ratio to reduce the image is also specified (S23). Although the tiling reference size and the minimum scale and reduction ratio may be set every time for the image conversion operation, it is most preferable to automatically set appropriate values set by default according to the size of the original image.

On the basis of the designated image conversion information as described above, the image of the maximum size is reduced at a set reduction ratio (S24) to obtain reduced scale information (S26), and this is repeated until the minimum scale size is obtained.

That is, the current image is reduced at a predetermined ratio (S24), and it is determined whether the reduced image is smaller than the set minimum scale size (S25). If the reduced scale size is smaller than the predetermined minimum scale size, the scale reduction is stopped, and the scales reduced in multiple stages arrange tile position information corresponding to each other to complete initialization (S27).

Table 1 shows an example of tile arrangement information initialized as described above. That is, the scaled tile arrangement information is designated with the scale information and indexes of tiles corresponding to each other when the image is reduced. That is, the tile arrangement information for each scale includes tree arrangement information in a tree form corresponding to each scale as an image conversion map.

For example, assuming the reduction ratio and the minimum scale is _^1/16 1/4 ⁿ (n is a natural number), reduced scale of the original image will be a _^1/4-scale and _^1/16-scale in addition to the original scale. That is, the original image according to the fondness the minimum scale is _^1/16 may be divided into 64 tiles, each tile has, for example, be indexed to 16-16, 16-1, 16-2, .... Accordingly tile 16-1 ~ 16-4 of the original image are corresponding to a tile of the 4-1 ^/ _{4-scale, ^1/4} of the tiles 4-1 through 4-4 in the scale are _^1/16-scale image of a Will correspond to (1). The remaining tiles will correspond to each other in the same manner as above.

scale Tile index original 16-1 16-2 16-3 16-4 16-5 16-6 16-7 16-8 16-9 16-10 16-11 16-12 16-13 16-14 16-15 16-16 _^1/4 4-1 4-2 4-3 4-4 ^/ ₁₆ One

Tiling refers to dividing one image into a plurality of image tiles having the same size, and tiles refer to individual images in a state in which one image is divided into a plurality of images. The pixel size of the image tiles for each scale is the same, and downscaling the four image tiles on the upper scale to 1/4 produces image tiles of the same size.

Subsequently, the sub-unit converter 115 performs the conversion of the sub-work units by using the tile arrangement information for each scale as shown in Table 1 prepared in the preprocessor 113. As shown in FIG. 6, the sub-unit converter ( 115 receives the scale step to be processed for each small work unit created by the preprocessor 113 and the tile arrangement information for each scale of the image area, and performs image calling and conversion according to the tile arrangement information for each scale. .

First, preprocessor 113 instructs subunit converter 115 to generate an image of the minimum scale size. Accordingly, the sub-unit converter 115 needs higher scale image tiles to construct the minimum scale according to the tile arrangement information for each scale as shown in Table 1 transmitted from the preprocessor 113 (S31). If the image tiles do not exist, a list of tiles necessary to obtain them is prepared (S32), and another subunit converter 115 requests this (S33). This process is continued until the maximum scale is reached, and the scale information and tile information are continuously collected accordingly (S34), and when the image tile stage at the maximum scale is reached, the corresponding region is copied through the image interpreter 111. (S35).

The image interpreter 111 detects an index, which is location-related information of the original image corresponding to the image tile designated in the preprocessor 113, and reads the corresponding image region data from the storage medium and provides it to the subunit converter 115. .

The sub-unit converter 115 constructs itself by collecting the transferred original image tiles, and takes the collected tile information S34 to complete the continuous output operation of the sub-unit converter 115 through repetitive down sampling. In other words, each subunit converter completes itself continuously.

For example, in the image tiles in the upper scale _^(1/4-scale) to form an image (1) of the subunit converter 115 is the minimum scale _{^(1/16-scale)} according to the scale-specific tile layout information in Table 1 (4 -1 to 4-4) is called (S31).

However, each of the image tiles in the upper scale _{^(1/4-scale),} so before converting the image are not already configured state. Accordingly, subunits of converter 115 is the next higher scale (original image) to form a single image tiles (4-1) of the back upper scale _^(1/4-scale) according to the scale generated by tile arrangement information (S32) The tiles 16-1 to 16-4 are called (S33), and the scale and tile information thereof are collected (S34).

Since the image exists at the highest scale (original image), the sub-unit converter 115 receives the tiles 16-1 to 16-4 of the corresponding image area from the image interpreter 111 and configures itself. (S35), thus completing a single tile (4-1) of the upper scale _^(1/4-scale) by downsampling the given specific ratio (S36). Subsequently, the sub-unit converter 115 sequentially transfers the remaining tiles 16-5 to 16-8, 16-9 to 16-12, and 16-13 to 16-16 of the highest scale through the image interpreter 111. After receiving it and reducing it to a specific ratio, the remaining tiles 4-2, 4-3 and 4-4 of the upper scale are completed in sequence.

Although the above-described subunit conversion operations related to higher scale may be performed sequentially, it is also possible to proceed in parallel at the same time using a plurality of processors.

Then, the converter subunit 115 if this reduction in a specific ratio in accordance with the tiles (4-1~4-4) are both composed of the upper scale _^(1/4-scale) of one of the least significant scale _{^(1/16-scale)} The image 1 is completed.

In addition, the output processor 117 collects the scale information and the position-related index information for each image tile completed by the subunit converter 115 and the corresponding tile transmitted from the preprocessor 113, and delivers the image information to the storage medium 130. It is to be stored (S37).

When the above steps S31 to S37 are repeated to complete the images of all scales, the image conversion operation is terminated.

The structure of the Taylor index of the image file converted for each scale and stored in the storage medium is shown in FIG. 7.

In the image conversion according to the present invention, the index is not in the format normally added to the header, but in the Taylor, which is the rear end of the data for each scale, and has a format continuously added when necessary. The present invention has a structure in which index information (T1) and metadata (T3, T4) are added to a recorded file, and index and metadata can be easily changed by placing index information at the end of the file. The change has an excellent advantage.

The image conversion file has a plurality of chunks in which at least one image tile data is stored, and the chunks are divided by scale. The location of each chunk is indicated by the chunk locator T2 in the table. The Taylor size T3 indicating the size of the image file is recorded immediately before the Taylor ID T4, and finally, the Taylor ID which is the image file name is recorded. The identification of the image is completed by reading the Taylor size and the Taylor ID located at the end of the file, and then confirming that the Taylor size and the Taylor ID are the same.

8 illustrates copying an arbitrary image area from an original image stored in a storage medium. The image interpreter 111 reads an image area at a specific position from the original image and transmits the image area to the sub-unit converter 115. That is, the image interpreter 111 transfers the specific image tile data predetermined in the preprocessor 113 to the subunit converter 115 through the index information of the original.

9 is assumed to be _^1/64-scale (Scale) of the final reduction state of, for convenience converting a conceptual view showing a conversion operation in a recursive manner according to the present invention, the original image. For convenience conversion scale is most ideally assumed to be one of the original scale, for example, _^1/4 of the original scale of the original _^1/16-scale, the original _^1/64-scale, and each scale are the same image.

The scale and reduction ratio are assumed for convenience of description and are not intended to limit the reduction ratio of the image or the scale of the tile.

First, the minimum scale _{^(1/64-scale)} is called for the the image tiles of the upper scale _{^(1/16-scale)} (4-1 ~ 4-4) to configure their own image (1). However, since the initial state is the tiles (4-1 through 4-4) in the _^1/16-scale that has not been configured to convert an image, each tile of the _^1/16-scale (4-1 through 4-4) It will call back the image tiles can configure your own (16-1 ~ 16-16) to the upper scale _^(1.4 scale).

Likewise, initially the image tiles in the _^1/4-scale for converting the image (16-1 - 16-16), because it is still not configured state, _^1/4 of each image tile in the scale (16-1 ~ 16- 16) calls the tile to configure itself to the original image tiles (64-1 ~ 64-64) of the highest scale.

스케일로 축소함에 따라 ¹/₄스케일의 하나의 타일(16-1)을 완성하게 된다. 이어, ¹/₄스케일은 원본 이미지로부터 다른 4개의 타일(64-5~64-8)을 각각 가져와서

스케일로 축소함에 따라 ¹/₄스케일의 다른 하나의 타일(16-2)을 완성한다. 이와 같은 과정을 반복하여 ¹/₄스케일은 한 그룹의 타일들(16-1~16-4)을 모두 완성하게 된다.Here, the _^1/4-scale came each take four tiles (64-1 ~ 64-4), from the source image defined in advance

It is complete the single tile of a _^1/4-scale (16-1), as reduced to a scale. Then, _^4.1 scales come each take four tiles (64-5 ~ 64-8) different from the original image

The _^1/4 completion of the other of the plates (16-2) of the scale, as reduced to a scale. Repeat the above process _^1/4-scale is completed all of the tiles in one group (16-1 ~ 16-4).

스케일로 축소함에 따라 ¹/₁₆스케일의 하나의 타일(4-1)을 완성하게 된다. Accordingly, _^1/16 scale because all the image tiles of the upper scale to organize their image tiles (4-1) (16-1 ~ 16-4) completed, the four plates (16-1 to 16 -4) take each one

As the reduction in the scale _^1/16 is completed a single tile (4-1) scale.

However, _^1/64-scale three tiles of the upper scale to configure their own image (1) (4-2 ~ 4-4) because it is not yet completed the remaining three plates (4-2, 1-4 Wait until -4) is complete.

스케일로 축소하여 ¹/₄스케일의 또다른 타일을 완성하는 과정을 반복하여 수행하게 되며, 이에 따라 ¹/₁₆스케일 단계도 자신의 상위 스케일의 4개의 타일이 완성되면 이를 이용하여 자신의 타일을 완성해 나간다.On the other hand, the _^1/4-scale is brought to another four tiles from the original image respectively, the top-level scale

The reduction in the scale _^1/4 is performed to repeat the process to complete the other tile of the scale, so that _^1/16-scale step is also finished their tiles by using this, if the four tiles of their higher scale completed Do it.

로 축소하여 자신의 마지막 이미지 타일(16-16)을 구성하고, ¹/₁₆스케일은 상위 스케일(¹/₄스케일)의 마지막 4개의 타일(16-13~16-16)을 불러와서

로 축소하여 자신의 마지막 이미지 타일(4-4)을 구성하게 된다.Thus through the same process _^4/1 scale steps are recalled each of the last tiles (64-61 ~ 64-64) of the original top-level scale

Reduction by configuring their final image tiles (16-16) and a _^1/16-scale is to be recalled and the last four tiles (16-13 - 16-16) of the upper scale ^_(1/4-scale)

Zoom out to form your last image tile 4-4.

로 축소하여 자신의 이미지(1)를 구성하게 된다.Four of the ¹ / if of the four tiles of the ₁₆ scale (4-1 through 4-4) are both completed, the _^1/64-scale high-scale _{^(1/16-scale)} to configure their own image (1) in the Call up each of the tiles 4-1 to 4-4

The image 1 is then reduced to.

As a result, _{^{1/4-scale, 1/16-scale}} and _^1/64-scale image of the original image is to be completed at substantially the same time by one. It has each process recursive and not redundant. At each scale, for example, ¹ / s in ₁₆ scales each tile (4-1, 4-2, 4-3 and 4-4) means that the process can be processed separately from,, multiprocessors, each subunit, multithreaded Parallel processing can be applied.

That is, the image of the minimum scale calls the tiles of the higher scale to construct itself. Load all the images and downsample them to their scale. However, tiles of higher scale are needed to form one tile of the called image. To do this, the higher scale tile calls the next higher scale tile. In this way, when the next higher scale tile is called and there is no next scale to be configured, the tile scales down accordingly by calling tiles of a specific image portion directly from the original image data instead of the next step.

This can be divided into the exclusive (exclusive) a particular process region (e.g., _^1/16 4-1 tiles, tiles 4-2, 4-3 and 4-4 tiles tiles of scale) for each scale as shown in the figure, and In other words, the divided processing region may be divided into another exclusive region at a higher scale and recursively dependent on a specific region image at another scale.

Therefore, the image conversion operation according to the present invention follows a data processing structure completely independent of each other for each tile in each scale step. That is, a particular tile of any scale will exactly match tiles of that higher scale. Due to this feature, the image conversion process according to the present invention can be operated independently of each unit of work in a multi-threaded, multi-process environment, and has the advantages of high efficiency and high performance compared to the conventional method of the sequential conversion method.

In addition, the maximum memory usage at every point in time is limited to the difference of the next scale unit in each scale unit operation and is kept constant during the entire conversion process. If each scale unit is twice the difference, only four units of memory are needed. In the conventional method, since the downsampling operation is applied to the full scale data, the difference is evident as the scale of the image increases as compared to the minimum memory required by adding the next image to the original image scale. In addition, I / O (Input / Output) operations are also reduced to less than half because there is no need to reload and reprocess the previously stored work units.

Also, because it is processed recursively, there is no limit on the conversion capacity compared to the conventional method. For example, the maximum convertible image capacity is limited to 1 gigapixel when processed in a 32-bit CPU according to the conventional method. However, the present invention is not limited to a conversion capacity of 100 gigapixel or 10 gigapixel according to the present invention.

FIG. 10 is a conceptual diagram illustrating multiple image area copying according to another embodiment of the present invention, and uses a virtual canvas to treat an image composed of multiple pages, such as an image attached to a plurality of photographs or an e-book, as one image.

The virtual canvas 101 transmits data to the image interpreters 111-1 and 111-2 instead of the original image, and may have a plurality of image interpreters 111-1 and 111-2 inside the virtual canvas. The plurality of image interpreters 111-1 and 111-2 have position and scale information, respectively, and receive and collect a part of the positions of the image interpreters 111-1 and 111-2 corresponding to the requested area. It passes to the analyzer 111. The image interpreters 111-1 and 111-2 in the virtual canvas receive data by referring to the original and deliver the data to the virtual canvas.

This is used when multiple photos or rendered document pages are bundled into a single image, each with an image parser that indexes each photo or rendered document location and has a full image parser that manages it again. Form.

11 is a diagram illustrating a system for high resolution image processing and a service thereof according to the present invention, and includes a web server 100, a wired / wireless network 200, and a user terminal 300.

The web server 100 includes an image conversion apparatus 110, a memory 120, a storage medium 130, an HTTP server program 140, an image unit browser 150, a script server program 160, and the like as shown in FIG. It comprises a communication means 170, the HTTP server program 140 and the script server program 160 is a configuration of a typical web server 100, the image conversion device 110 and the image unit browser 150 ) Is the configuration according to the present invention.

The image conversion apparatus 110 divides and scales an image source according to the present invention, and in the process of segmenting tiling and scaling, the image conversion apparatus 110 loads and reduces tiles having a larger scale than itself, and configures the storage medium 130. The image unit browser 150 is configured to call the image stored in advance in the storage medium 130 and transmit the image to the user terminal 300 in response to a user's image request.

The user terminal 300 includes an input unit 310, a display unit 320, an internet browser 330, an image unit browser 350, an image processor 360, and a communication unit 370. The user terminal 300 is a personal computer such as a desktop, a laptop, a handheld, or a mobile communication terminal such as a PDA and a mobile phone, and operates a predetermined wire / wireless network by driving an internet browser 330 or a virtual machine browser 340. 200 may be connected to the web server 100.

The user terminal 300 is connected to the web server 100 through the Internet browser 330 or VM browser 340 and the communication means 370, and to the display means 320 through the input means 310, such as a mouse. When the image to be displayed is selected, the image unit browser 350 requests the web server 100 for image data corresponding to the scale corresponding to the display area of the display means 320, and the image transmitted from the web server 100. The data is output to the display means 320 by the image processor 360.

The image unit browser 350 installed in the user terminal 300 by the web server 100 may be directly plugged into the Internet browser 330 or plugged in to the Internet browser 330 such as Flash Player and Silver Light. It has a feature that can be used in conjunction with various VM (Virtual Machine) browser 340 or Ajax, and has a feature that can be plugged in directly to a variety of VM browsers used in PDAs, mobile phones, and the like. The image unit browser 150 of the web server 100 may be automatically downloaded to the user terminal 300 when the user terminal 300 requests an image tile by accessing the web server 100.

As such, the tiles, that is, the image data, made through the image conversion process of the present invention are stored in the storage medium 130 of the server such as the same drawing, and the stored image data is transmitted to the user terminal 300 through the wired / wireless network 200. It is transmitted through the image unit browser 150 in units of image tiles requested by the user rather than the entire file.

12 is a view illustrating a real-time transmission method of the image unit according to the present invention, how the tiled and scaled image stored in the storage medium 130 of the web server 100 to the user in the image tile unit in real time A conceptual diagram showing if it is sent.

The location information of each image tile is included in the index T1 of the image file as shown in FIG. 7, and the necessary image tile can be transmitted using the index information. That is, as shown in FIG. 11, the image data transmitted to the user terminal 300 transmits only image tiles belonging to a portion to be displayed on the display means 320 of the user terminal 300.

Here, the image unit browser 350 of the web server 100 is responsible for the communication between the index and the user terminal 300. As described above with reference to FIG. 11, the image unit browser 350 may have its own separate browser, plug-in to an Internet browser, or plug-in to a VM browser, used by a user, or when a user requests an image tile. It is automatically imported and used.

For example, when a user inputs a reduction command of an image currently displayed on the display means 320 through an input means 310 such as a mouse, the image unit browser 350 requests the corresponding image tile from the web server 100. do. Accordingly, the web server 100 transmits the image tile of the corresponding position in the layer having a smaller scale than the layer being displayed on the display means 320 of the current user terminal 300 to the user terminal 300. Of course, when the user inputs an image magnification command, the web server 100 displays the image tile at the corresponding position in the layer having a larger scale than the layer being displayed on the display means 320 of the current user terminal 300. To 300. When the web server 100 transmits an arbitrary scale layer to the user terminal 300, the web server 100 does not transmit all the tiles of the layer, but belongs to a portion displayed in consideration of the scale of the display means 320 of the user terminal 300. Only image tiles will be sent.

That is, as the user reduces or enlarges the displayed image, the web server 100 changes the image corresponding to N in layer 0 and transmits the changed image. In addition, when the user moves the screen, the web server 100 transmits the corresponding tile located next to the current layer being displayed.

In addition, by using the x, y pixel position information at a specific position between tiles, the individual multimedia file (Object file) such as sound, video, animation, image, hyperlink, etc. is called and transmitted to the user terminal 300 at the corresponding position. It may be. That is, when the user calls the tile of the corresponding position, the object file (OF) is called together at the position and displayed on the predetermined x and y coordinates. In this case, the object file (OF) is reduced and enlarged at the same rate for each layer, and information necessary for this can be used in addition to Taylor.

Preferred embodiments of the present invention are disclosed for purposes of illustration, and those skilled in the art will be able to make various modifications, changes, and additions within the spirit of the present invention. Therefore, such modifications, changes and additions should be determined not only by the claims below, but also by equivalents to those claims.

1 is a diagram illustrating a conventional image processing and conversion method.

2 is a view showing an image conversion apparatus of a web server according to an embodiment of the present invention.

3, 4 and 6 are flowcharts illustrating an image conversion process according to an embodiment of the present invention.

5A and 5B illustrate a tiling diagram of an original image according to the present invention.

7 is a diagram illustrating a structure of a Taylor index of a file converted for each scale according to the present invention.

8 is a conceptual diagram showing that the image interpreter retrieves the image from the original as a result of the sub-unit work according to the present invention.

9 is a conceptual diagram illustrating a recursive image conversion operation according to the present invention.

10 is a conceptual diagram illustrating multiple image area copy according to another embodiment of the present invention.

11 is a diagram illustrating a system for high resolution image processing and a service thereof according to the present invention.

12 is a view illustrating a real-time transmission method of an image unit according to the present invention.

* Explanation of symbols for the main parts of the drawings

100: web server 110: image conversion device

111: image interpreter 113: pre-processor

115: subunit converter 117: output processor

120: memory 130: storage medium

140: HTTP server program 150: image unit browser

200: wired and wireless network 300: user terminal

310: input means 320: display means

330: Internet browser 350: Image unit browser

Claims (21)

An image analyzer for analyzing an header of the original image file and creating an index for partial access to process the size information of the original image and a request for access to the partial image; A preprocessor for generating image conversion information necessary for a conversion operation based on the original image-related information input from the image analyzer and generating tile arrangement information for each scale using the generated image conversion information; In response to a tile generation request of an arbitrary scale input from the preprocessor, a higher-level tile corresponding to a tile of an arbitrary scale is called to construct itself by referring to tile arrangement information for each scale, and finally, an original A sub-unit converter which receives the scale tiles from the image analyzer and reduces them to a predetermined ratio to generate their own image tiles; And And an output processor for collecting image tile produced by the sub-unit converter and tile arrangement information for each scale input from the preprocessor and transferring the image tile to the storage medium. The method according to claim 1, When the sub-unit converter generates a tile of any scale, at least two or more tiles from the upper scale and down-sampling at a predetermined reduction ratio to generate their own image tiles. The method according to claim 1, And the image interpreter analyzes a header of the original image data, and analyzes an image property necessary for conversion, format information, compression information, and data storage method of the original image, and has an index for partial access. The method according to claim 1, The image conversion information generated by the preprocessor includes a tiling reference size for the size of the tile to be divided, a maximum scale for dividing the original image using the tiling reference size, and a minimum scale and reduction ratio to be reduced to the maximum. A mass image processing apparatus comprising at least one or more. The method according to claim 1, Each image tile included in the arbitrary scale is dependent on each other in parallel to a large-capacity image processing apparatus, characterized in that the image generation operation is performed. The method according to claim 1, And the output processor adds image-related index information to a Taylor, which is a rear end of the image, when the index information is collected on the generated image tile. The method according to claim 1, The large-capacity image processing apparatus is mounted on a web server, and the large-capacity image processing device bundles the generated image tiles into a predetermined number of image tiles and transmits them to the user terminal through a wired / wireless network according to a user terminal's connection and image transfer request. Image processing unit. A first step of creating an index for each partial region of the original image and a size of the original image to access the original image for each partial region; A second step of setting image conversion information including a minimum scale based on the original image related information; A third step of sequentially reducing the current scale according to the set reduction ratio until the minimum scale is set, and generating each scaled scale and tile arrangement information for each scale; And A fourth step of generating an image tile by reducing the image tiles of a higher scale to a predetermined ratio by calling image tiles of a higher scale in order to configure an image tile according to the tile arrangement information for each scale defined above; . The method according to claim 8, The image conversion information of the second step may include at least one of a tiling reference size for a size of a tile to be divided, a maximum scale for dividing an original image using the tiling reference size, and a minimum scale and a reduction ratio to be maximized. A large-capacity image processing method comprising at least one. The method according to claim 8, The third step further comprises the step of arranging a plurality of scales for each tile corresponding to each other. The method according to claim 8, And the maximum scale further comprises a virtual area other than the original image when tiling the original image to a tiling reference size. The method according to claim 8, Each scale is obtained by reducing the maximum scale by a set ratio. The method according to claim 8, And the tile arrangement information for each scale includes map information on image tiles corresponding to each scale. The method according to claim 8, And the image tile generated in the fourth step is collected with the index information related to the image and stored in the storage medium. The method according to claim 14, And when the index information is collected in the generated image tile, the image related index information is added to a Taylor which is a rear end of the image. The method according to claim 8, When generating the image tile in the fourth step, a large-scale image processing method, characterized in that for generating at least two or more tiles of the upper scale by collecting and down-sampling at a predetermined reduction ratio. The method according to claim 8, Each tile of the lower scale is a large-scale image processing method, characterized in that the image conversion and generation operations are performed independently of each other. The method according to claim 8, In the fourth step, if the tiles of the higher scale are not configured, it is determined whether the next higher scale tiles corresponding to the specific tile of the higher scale are configured according to the tile arrangement information for each scale, and the corresponding image tiles are called. Large-capacity image processing method characterized in that. The method according to claim 8, The image tile generated in the mass image processing method characterized in that the bundle is transmitted to a predetermined number of image tiles through a wired or wireless network at the request of the user terminal. The method according to claim 19, And transmitting additional files such as a video, an image, an animation, a sound, or a 3D linked to a specific location of an image when the image is transmitted in units of tile through the wired / wireless network. The method of claim 20, And the additional file can be reduced and enlarged at a scale ratio equivalent to that of the image.

Images