KR100772613B1 - Image compression method for image apparatus - Google Patents

Abstract

The present invention relates to a method for compressing an image suitable for a device which requires a high reproduction rate while having high compression efficiency.

To this end, the present invention, the first coding step of performing the coding using the difference value between the current pixel and the previous pixel for the input color image, and applying the neighboring color merging method to the input color image, and each of the first coding step Arranging the difference values stored in the pixels in order of high frequency, generating a difference table, dividing the stream of difference values according to a pattern, and optimally for the stream of difference values divided according to the pattern An image compression method comprising a second coding step of generating a code of.

Image Compression, Difference Value, Difference Value Table, Adjacent Color Merging

Description

Image compression method for image apparatus

1 is a flowchart illustrating an image compression process of an imaging apparatus according to the present invention.

TECHNICAL FIELD The present invention relates to image compression technology, and more particularly, to an image compression method suitable for a device requiring high reproduction speed and fast reproduction. In particular, it is an image compression method that is more suitable for a portable terminal having excellent portability but small storage capacity and low processing power.

Recently, with the development of the Internet and information and communication technology, personal portable terminals such as mobile phones, MP3 players, and PMPs have gained great popularity.

These portable terminals are in a trend of miniaturization and light weight due to the great advantage that they are easy to carry, and at the same time, consumers are demanding multi-function and large-capacity processing such as audio, video, video playback or Internet use in the portable terminals.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described needs, and an image compression method capable of showing fast reproduction while greatly improving the compression efficiency by encoding the difference between the current pixel and the immediately preceding pixel and applying a neighboring color merge method. The purpose is to provide.

In this way, it is possible to reduce the image data stored in the memory of the imaging device to maximize the data storage capacity, it is possible to provide the maximum satisfaction to the user through the fast playback capability.

In order to achieve the above object, according to an aspect of the present invention, the primary coding to use the difference value between the current pixel and the immediately preceding pixel for the input color image, and to perform coding by applying a neighboring color merging method to the input color image Generating a difference table by arranging the difference values stored in each pixel of the first coding step in the order of high frequency, dividing the stream of difference values according to a pattern, and And performing secondary coding to generate an optimal code for the separated stream of difference values.

In addition, the first coding step may include a first step of converting a pixel value consisting of a combination of three colors R, G, and B for the input color image into a difference value of the immediately preceding pixel value from the current pixel value; A second step of applying a neighboring color merging method for merging the end-digit values into a neighboring color for the pixel value converted in the first step; and a pixel value to which the neighboring color merging method in the second step is applied to the current pixel value. It characterized in that it comprises a third step of storing as.

The pattern may be divided into a coding unit displaying information on data to be coded and a data unit representing an actual difference value.

The coding unit may include: a first structure unit indicating what type of data is present; a second structure unit indicating how many times a value whose difference value is not '0' immediately after the current pixel is continuous; a difference value immediately before the current pixel And a data structure of the third structure portion which indicates how many times the value of '0' is continuous.

In addition, the first structure unit allocates two bits as bits for determining the type of data, and indicates a data type when the difference value is a difference value existing only once and the difference value is present in the difference value table. ',' If the difference value is not '0' only once, and if the difference value does not exist in the difference table, set the data type to '01', and the difference value is not '0'. It is characterized in that the data type in the case of appearing continuously is designated as '10', and the data type in the case where the sequence is 64 or more in the pattern of '10' is designated as '11'.

In addition, the second structure and the third structure may be extended by using the number of times for the remaining 6 bits of the 2 bits used in the pattern, or by continuously using the 6 bits and 1 byte immediately after that. It features.

In addition, when the pattern is '00', the data unit designates an index of a matching value on the difference value table as 1 byte, and when the pattern is '01', the difference value is assigned to 2 bytes as it is, and the pattern Is '10' and '11', if the difference value exists in the difference table, specify the index of the matching value as 1 byte, otherwise specify 1 byte as '0xff' and then 2 bytes It is characterized in that for specifying a difference value.

In addition, the second coding is treated as one pattern in the case of a specific pattern set, and adjusts the number of streams in which the difference value '0' appears continuously, and the difference value is not '0'. It is characterized by adjusting the number of streams which appear continuously.

The method may further include supporting a lossless mode for the stream of difference values when the neighboring color merging method is not applied.

In addition, in the case of the lossless mode support, one bit is further added as a sign bit to two bytes.

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

1 is a flowchart illustrating an image compression process of a portable terminal according to an exemplary embodiment of the present invention.

As shown in FIG. 1, first, a difference value between a current pixel and a previous pixel is obtained, and image encoding is performed by applying a neighboring color merging method to the difference value (S110).

At this time, the difference value table stored in each pixel is created in order of high frequency (S120).

Then, in order to increase the efficiency of compression, the stream of difference values to be coded are classified according to a pattern, and each pattern is a coding part for displaying information about data and a data part for indicating an actual difference value. divided into parts) (S130).

Thereafter, second order coding is applied to the pattern in order to increase the efficiency of compression (S140).

Subsequently, when the neighboring color merge method is not applied when encoding the image, the lossless mode is supported (S150).

As such, image data stored in the memory of the imaging apparatus may be reduced through the compression process.

Hereinafter, the above-described image compression process will be described in more detail.

1. First order coding.

1.1. Just before the current pixel Interpixel  Use difference value

In the case of a color image, one pixel is generally composed of three colors, R (Red), G (Green), and B (Blue).

In particular, the image used in the mobile phone uses two bytes (R: 5 bits, G: 6 bits, B: 5 bits). For example, in the case of an 8 * 8 color image, pixel values consisting of three colors R, G, and B are stored for each 64 pixels.

In the compression algorithm according to the present invention, a pixel value consisting of a combination of three colors R, G, and B is converted into a difference value obtained by subtracting the value of the immediately preceding pixel from the current pixel and stored in the current pixel value. (The first pixel that starts must have the same value as the first pixel of the image, so that you get the same result image as the original image.)

UI (User Interface) image data of an imaging device has a great similarity between adjacent pixels. Therefore, if the difference value stored in each pixel is sorted in order of frequency, most UI images are included in 256 indexes. Using these characteristics, if 256 color tables are created in order of frequency, the difference value of 2 bytes is matched to the color table index of 1 byte to improve compression efficiency. It can increase.

On the other hand, pixel values having a difference value of '0' need not be coded. However, you only need to specify the distance at which the pixel value appears, not the difference value '0', just before the current pixel position. That is, only the number of streams having a difference value of '0' needs to be specified.

1.2. Adjacent color merging

The neighboring color merge method is defined as the last digit for each color value of R, G, and B. Merge values into neighboring colors (i.e. end-points can appear from 0 to 9, but only show values of '1, 3, 5, 7, 9' (or '0, 2, 4, 6, 8') Merging to one side). By eliminating the less important parts (the values at the end), you can increase the rate of inclusion in the difference table.

If you apply the neighboring color merge method, there is a slight loss, but the compression efficiency can be greatly increased without being able to feel the difference from the original image because it is visually felt. In the case of an image having a gradation, the image quality is visually deteriorated. In such a case, a color having a gradation is solved by applying the neighbor color merge method only to the remaining colors without applying the neighbor color merge method.

An encoder and a decoder are generated by applying the algorithms of the difference value between the current pixel and the immediately preceding pixel and the neighbor color merging method.

2. Coding Rules

2.1. Difference table

As mentioned above, when the similarity between pixels is large, when coding using a difference value, most difference values are included in 256 difference table indexes. In other words, two-byte difference can be coded with an index of one byte, thus creating 256 color tables according to the frequency. Also, since the number of difference tables may vary depending on the image, a value that can specify how many difference tables are created is also written.

2.2. Coding

In order to increase the efficiency of compression, a stream of difference values to be coded is divided according to a pattern, and each pattern is divided into two parts (a coding part and a data part). The coding part displays information on data, and the data part is a part for indicating an actual difference value.

end. Coding part

When analyzing the stream of difference values, it is effective to divide the configuration of the coding unit into three. The first indicates what type of data it is, the second indicates how many times the value of the difference is non-zero immediately after the current pixel, and the third is zero at the current pixel position. It indicates how many times the value of 'is consecutive.

1) bits that determine the type of data

Analysis of the pattern of the stream of difference values is classified into three major categories. Therefore, 2 bits are allocated as bits for determining the type of data.

'00': Set the data type to '00' when the difference value is not '0' and the difference value exists only once in the difference table.

'01': Set the data type to '01' when the difference value is not '0' once and the difference value does not exist in the difference table.

'10': Specify '10' as the data type when the difference value is not '0'.

If there is no pattern to be allocated for the case of '11', to increase the efficiency of compression, the case where the number of consecutive occurrences of 64 or more among the patterns of '10' is 64 is designated as '11'.

In other words, the final pattern type is shown below.

'00': Set the data type to '00' when the difference value is not '0' and the difference value exists only once in the difference table.

'01': Set the data type to '01' when the difference value is not '0' once and the difference value does not exist in the difference table.

'10': Specify '10' as the data type when the difference value is not '0'.

'11': Specify '11' as the data type when the sequence is 64 or more times among the patterns of '10'.

2) Bit that specifies the number of times the difference value is not '0' immediately after the current pixel

Only in the case of '10' and '11' among the previous patterns. That is, in the patterns of '00' and '01', it is not necessary to allocate bits for coding '2)'. To improve the efficiency of compression, it is basically used to specify the number of times for the remaining 6 bits of the 2 bits used in the pattern. If the pattern is "10" if there can specify the number of 6 bits, "11", there can, using the 6-bit and the immediately following byte specify up to 16383 (2 ¹⁴ -1) of the number. To specify the number of times more than 16383, the first 1 bit of 1 byte immediately after 6 bits is used as an extension bit. In other words, if the bit is set to '1', 8192 (2 ¹³ ) or more times can be designated. You can continue to expand in this way.

3) Bit that designates the number of times the difference value is '0' immediately before the current pixel.

Specifies the number of consecutive times a value with a difference of '0' appears before the pattern is changed. Like this '2)', the bit for expansion can be designated separately and can be continuously extended.

I. Data section

The value to be inserted into the data part varies according to the pattern, so it is classified according to the pattern.

If the pattern is '00': Specify the index of the matching value in the difference table as 1 byte.

If the pattern is '01': Set the difference value to 2 bytes as it is.

If the pattern is '10' and '11': If the difference value exists in the difference value table, specify the index of the matching value as 1 byte, otherwise specify 1 byte as '0xff' and the remaining 2 Specifies the difference value in bytes. In order to improve the efficiency of compression, 1 byte is designated as '0xff', and the actual difference value table is 256 to 255 except for '0xff'. Even if the difference table is reduced from 256 to 255, the ratio included in the difference table does not vary significantly, and the compression efficiency is higher.

3. Secondary Coding

The following methods are additionally applied to increase the efficiency of compression.

3.1. For a specific set of patterns, treat them as one pattern

If the difference value is a pattern ('10' or '11') in which a value other than '0' is consecutive, the above pattern (' 10 'or' 11 '). For example, a stream such as "·· 0x0000 0x0001 0x0002 0x0000 0x0002 0x0003 0x0000 0x0002 0x0001 0x0000 ... It can be treated as one pattern.

3.2. Successive values with a difference of '0' Of stream  Adjust the frequency

A value with a difference of '0' must be seen at least two times in a row to be recognized as not a '10' (or '11') pattern. Can be taken.

3.3. Consecutive values other than zero Of stream  Adjust the frequency

For the same reason as '3.2', '-1' may be taken from the number of streams in which the difference value is not '0'.

2.4. Loss mode  Support (if no neighbor merging is applied)

When the neighboring color merge method is applied, the difference value can be expressed in two bytes. However, if the neighboring color merge method is not applied, one bit (sign bit) is required for the two bytes. To minimize memory consumption for these sign bits, allocate space to specify the sign bits of the difference table (up to 32 bytes), and set the sign bits for the difference values that do not exist in the difference table. Allocate space that you can specify. In general, since most difference values are included in the difference table, the overhead caused by the sign bits is not large.

In this way, the image data stored in the memory of the imaging apparatus can be reduced through the compression process, the decoded image is restored to the image before compression, and the reproduced image is reproduced through the display unit through decoding.

The compression method is applicable to various types of video devices as well as portable terminals such as MP3 players, mobile phones, and PMPs.

As described above, the image compression method according to the present invention can greatly improve the compression efficiency by encoding the difference between the current pixel and the immediately preceding pixel and applying the neighboring color merge method.

Through this, it is possible to reduce the image data stored in the memory of the imaging device to maximize the data storage capacity.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above-described embodiments, which can be variously modified and modified by those skilled in the art to which the present invention pertains. Modifications are possible.

Therefore, the spirit of the present invention should be grasped only by the claims described below, and all equivalent or equivalent modifications thereof will belong to the scope of the present invention.

Claims (10)

For each pixel of the pixel stream constituting the color image, the difference value between the current pixel and the previous pixel of the current pixel is converted into the value of the current pixel, and the neighboring color merging method is applied to the value of the converted pixel to obtain the first order. A primary coding step of performing coding; Creating a difference value table by sorting each pixel value of the primary coded pixel stream in order of high frequency; Detecting a pattern of the primary coded pixel stream; Secondary coding the primary coded pixel stream according to the detected pattern Image compression method comprising a. The method of claim 1, The first coding step, The pixel value is a combination of three colors R, G, and B, and the value of the current pixel is a difference value between the value of the current pixel and the value of the previous pixel of the current pixel, and has a value having one or more digits. A first step of converting; A second step of applying a neighboring color merging method of merging the last digit value of the converted current pixel values into a neighboring color; And storing the value to which the neighboring color merge method is applied as a value of the current pixel. The method of claim 1, The pattern is divided into a coding unit for displaying information on the data to be coded, and a data unit for representing the actual difference value. The method of claim 3, The coding unit may include: a first structure unit for indicating what type of data is present; a second structure unit for displaying how many times a difference value is not '0' immediately after the current pixel, and a difference value immediately before the current pixel And a data structure of a third structure portion indicating how many times a value of '0' is consecutive. The method of claim 4, wherein The first structure unit allocates 2 bits as bits for determining the type of data, Set the data type to '00' when the difference value is a non-zero value only once and the difference value exists in the difference table. Specify a data type of '01' when the difference value is a non-zero difference only once and the difference value does not exist in the difference table. Set the data type to '10' when the difference value is not '0' in succession. The data compression method of claim 11, wherein the data type is set to '11' when the sequence is 64 or more times among the patterns of '10'. The method of claim 5, The second structure portion and the third structure portion can be used to designate the number of times for the remaining six bits of the two bits used in the pattern, or can be extended in such a manner that the six bits and one byte immediately after that are used continuously. Image Compression Method. The method of claim 4, wherein If the pattern is' 00 ', the data index designates the index of the matching value on the difference value table as 1 byte, and if the pattern is' 01', the difference value is designated as 2 bytes and the pattern is' In case of 10 'and' 11 ', if the difference value exists in the difference table, the index of the matching value is designated as 1 byte; otherwise, 1 byte is designated as' 0xff' and then 2 bytes are specified. An image compression method characterized by specifying a difference value. The method of claim 1, In the second coding, a specific pattern set is treated as one pattern, and the number of streams in which the difference value '0' is continuously displayed is adjusted, and the difference value is not '0' is continuous. And adjusting the number of streams displayed. The method of claim 1, And if the neighboring color merge method is not applied, supporting a lossless mode for the stream of difference values. The method of claim 9, In the case of the lossless mode support, the image compression method characterized by adding one bit as a sign bit to two bytes.

Images