CN117544734A - Lossy compression method based on RGB three primary color images - Google Patents

Abstract

The invention discloses a method for lossy compression based on RGB three primary color images, which relates to the field of screen display and digital image processing, and is characterized in that: the method comprises the following steps: s1: sampling an input RGB image, performing RAM_buffer storage according to row arrangement, and setting a first row, a first column, a last row and a last column of pixel marks; s2: performing a small value marking by using the G component of each pixel to perform a near neighborhood; s3: the odd number columns of the odd number rows are RG, the even number columns of the odd number columns are GB, the even number columns of the even number columns are GB, and the even number columns of the even number columns are RG. The invention utilizes the transmission characteristic of the data pipeline type, through RAM_buffer buffer in the FPGA chip, acquires the pixel with the nearest value of the G component of the pixel by using a comparator and an arithmetic subtraction calculator, performs subtraction processing, compares the obtained 2-4 results, and takes the smallest comparison result, finally obtains the corresponding small mark, combines and replaces the original pixel data component value, combines the corresponding RG/GB value, and then buffers and outputs.

Description

Lossy compression method based on RGB three primary color images

Technical Field

The invention relates to the field of screen display and digital image processing, in particular to a method for lossy compression based on RGB three-primary-color images.

Background

The RGB colors are a color standard, various colors are obtained through superposition of three color channels of red, green and blue, the three primary colors are independent in nature, any color of the three primary colors cannot be composed of other two colors, the three primary colors have the largest mixed color gamut, other colors can be mixed by the three primary colors according to a certain proportion, and the maximum number of the obtained colors is obtained after mixing, just like the base in space, like x, y and z axes in a three-dimensional space;

however, as the resolution of the display device is higher, the image transmission speed of the device is faster, the pressure of the buffer space is higher, the image compression technology is generated, and the existing compression algorithm is complicated and occupies more resources.

Disclosure of Invention

Based on this, the present invention aims to provide a method for lossy compression based on RGB three primary color images, so as to solve the technical problems set forth in the background above.

In order to achieve the above purpose, the present invention provides the following technical solutions: a method for lossy compression based on RGB three primary color images comprises the following specific steps:

s1: sampling an input RGB image, performing RAM_buffer storage according to row arrangement, and setting a first row, a first column, a last row and a last column of pixel marks;

s2: performing a small value marking by using the G component of each pixel to perform a near neighborhood;

s3: the original value RGB of odd columns and odd columns is RG, the original value RGB of even columns and even columns of odd columns is GB, the value GB of even columns and the value RG of even columns and even columns are GB;

s4: replacing the minimum 2bit of the R component in the sample RG with the small mark, and replacing the minimum 2bit of the B component in the GB with the small mark corresponding to the value of the next sample;

s5: storing the combined rGGb/GbrG data into a cache from line sequence and outputting the data;

s6: the rGGb/GbrG data is taken out from the cache and is stored in a row RAM_buffer according to row arrangement;

s7: generating a first row of pixel marks, a first column of pixel marks, a last row of pixel marks and a last column of pixel marks of the boundary;

s8: taking the R/B component missing from the pixel point from the neighboring domain by using the boundary mark combination to take the small mark 2 bit;

s9: and outputting the restored RGB pixels to a display terminal according to the rows and the columns to display an image.

Preferably, the green G component which is sensitive to human eyes is used as a comparison factor, the neighborhood G component is arithmetically compared, and R/B components corresponding to the pixel are removed in a staggered mode according to the odd-even rows and the even-odd columns to be used as a compression mode, so that the purpose of 1/3 lossy compression is achieved.

Preferably, in the reduction process, a 2bit small flag in the R/B component is used to directly take a value of a B/R component in the corresponding field as a decompression mode, and the method and the steps include:

step one: inputting RGB three primary color images with m+1xn+1 resolution, caching the images into a RAM_buffer line by line, recording row and column sequence number marks, odd and even row and column marks, row sequence numbers 0-m and column sequence numbers 0-n;

step two: the second line of the input image data starts to carry out size comparison by using the pixel point G component;

(1) class: row 0, column 0 pixel f (0, 0) is compared with the G component to the right and below;

(2) class: row 0, column 1 through column n-1 pixel f (0, 1 through n-1) with its left, right, and lower G components involved in the comparison;

(3) class: row 0, column n pixel f (0, n) participates in the comparison with the left and lower G components;

(4) class: the 1 st row to the n-1 st row, the 0 th column pixels f (1-n-1, 0) are used above,

The lower and right G components participate in the comparison;

(5) class: the 1 st to m-1 st rows 1 st to n-1 st columns of pixels f (1 to m-1,1 to n-1) participate in comparison by using four G components of the pixels f up, down, left and right;

(6) class: the n-th column f (1-m-1, n) of the 1 st to m-1 st rows participates in comparison with the upper part, the left part and the lower part thereof;

(7) class: the m-th row, 0-th column pixel f (m, 0) participates in the comparison with its upper and right G components;

(8) class: the m-th row 1-n-1 column pixels f (m, 1-n-1) participate in comparison with their left, upper and right G components;

(9) class: the m-th row and n-th column pixels f (m, n) are used for the comparison by the upper and right G components to obtain a 2bit small mark, wherein binary 00 represents the upper direction of the value, 01 represents the right direction of the value, 10 represents the left direction of the value, and 11 represents the lower direction of the value

Step three: when the odd-numbered lines are arranged to take the value RG_GB, the even-numbered lines take the value GB_RG, when the odd-numbered lines are arranged to take the value GB_RG, the even-numbered lines take the value RG_GB, the staggered value is kept, the R component removed by the pixel point is kept to have the B component in the four directions of the neighborhood, the B component removed by the pixel point is kept to have the R component in the four directions of the neighborhood, the boundary indication method and the type II confirm the arrangement of the R/B component, after the small mark of 2 bits after arithmetic comparison is obtained, the G component is kept unchanged, the 2bit mark is replaced to the lowest bit 2bit of R/B in the reserved RG/GB, so that the compression process is completed by obtaining the compressed 1/3 image data, the image compressed data is stored in a buffer area in a pipelining mode and is sent to the next-stage hardware equipment

Step four: when the image is decompressed, the image data can be decompressed out from the first line when the image data is read to the second line, the buffer data is sequentially read, the R/B component missing from each pixel point is obtained according to the boundary mark type by utilizing the R/B component lowest 2bit mark in the received rG_Gb/Gb_rG, a complete RGB three-primary-color pixel is formed, when the small mark is 00, the R/B component corresponding to the sequence number of the pixel column in the last line is represented by the small mark, when the small mark is 01, the R/B component corresponding to the sequence number of the pixel column in the same line is represented by the small mark, namely the R/B component in the left side is represented by the small mark, and when the small mark is 11, the R/B component corresponding to the sequence number of the pixel column in the next line is represented by the small mark.

In summary, the invention has the following advantages:

the invention utilizes the data pipeline transmission characteristic, through RAM_buffer buffer in the FPGA chip, uses a comparator and an arithmetic subtraction calculator to obtain the pixel with the size closest to the G component value of the pixel, performs subtraction processing, compares the obtained 2-4 results, and obtains the minimum comparison result, finally obtains the corresponding small mark, combines and replaces the original pixel data component value, combines the corresponding RG/GB value and then buffers and outputs. The invention can largely reserve the original image information, the decompressed image has no color edge phenomenon, the image is not deformed, no edge blurring phenomenon, the color rendition is high, the whole process is simple, the algorithm is concise, few resources are occupied, the transmission speed of image data is reduced, the resources and the frequency cost are saved for a processor, the cost of a main chip is greatly reduced, and more selection space is reserved for chip device selection.

Drawings

FIG. 1 is a schematic flow chart of an embodiment of the present invention;

FIG. 2 is a schematic diagram of an algorithm architecture for implementing the present invention;

FIG. 3 is a schematic diagram of an algorithm decompression implemented according to the present invention.

In the figure: 201-primary image input RGB three primary color pixel arrangement;

202-the upper left corner of the original image, namely the first column of the first row, corresponds to a G component value taking model and an assignable small mark result of m+1χn+1 resolution rank serial number (0, 0) pixels;

203-the first row, second column to n-1 th column pixel type model of the original image;

204-the upper right corner of the original image, i.e., the first row, last column, corresponds to the row and column number (0, n) pixel type model;

205-the second row to the m-1 th row of the original image, the first column of pixel models;

206-second row to m-1 th row, second column to n-1 column pixel model of original image;

207-the second row to the m-1 row of the original image, the last column of pixel model;

208-the lower left corner of the original image, i.e., the last row, first column of pixel models;

209-the last row of the original image, second column to n-1 column pixel model;

210-the bottom right corner of the original image, i.e., the last row, last column of pixel models;

211-R/G/B digital bit distribution, high bit front, big end mode, 2bit small mark to replace the lowest two bits R10/B10 in R/B component, and obtain final result rG_Gb/Gb_rG.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the invention.

Hereinafter, an embodiment of the present invention will be described in accordance with its entire structure.

A method for lossy compression based on RGB three primary color images, as shown in fig. 1, specifically comprises the following steps:

s1: sampling an input RGB image, performing RAM_buffer storage according to row arrangement, and setting a first row, a first column, a last row and a last column of pixel marks;

s2: performing a small value marking by using the G component of each pixel to perform a near neighborhood;

s3: the original value RGB of odd columns and odd columns is RG, the original value RGB of even columns and even columns of odd columns is GB, the value GB of even columns and the value RG of even columns and even columns are GB;

s4: replacing the minimum 2bit of the R component in the sample RG with the small mark, and replacing the minimum 2bit of the B component in the GB with the small mark corresponding to the value of the next sample;

s5: storing the combined rGGb/GbrG data into a cache from line sequence and outputting the data;

s6: the rGGb/GbrG data is taken out from the cache and is stored in a row RAM_buffer according to row arrangement;

s7: generating a first row of pixel marks, a first column of pixel marks, a last row of pixel marks and a last column of pixel marks of the boundary;

s8: taking the R/B component missing from the pixel point from the neighboring domain by using the boundary mark combination to take the small mark 2 bit;

s9: and outputting the restored RGB pixels to a display terminal according to the rows and the columns to display an image.

As shown in fig. 2 and 3, the green G component which is sensitive to human eyes is used as a comparison factor, the neighborhood G component is arithmetically compared, and R/B components corresponding to the pixel are removed in a staggered manner according to the odd-even rows and the even-odd columns as a compression mode, so that the purpose of 1/3 lossy compression is achieved; the method and the steps of the method are as follows:

step one: inputting RGB three primary color images with m+1xn+1 resolution, caching the images into a RAM_buffer line by line, recording row and column sequence number marks, odd and even row and column marks, row sequence numbers 0-m and column sequence numbers 0-n;

step two: the second line of the input image data starts to carry out size comparison by using the pixel point G component;

(1) class: row 0, column 0 pixel f (0, 0) is compared with the G component to the right and below;

(2) class: row 0, column 1 through column n-1 pixel f (0, 1 through n-1) with its left, right, and lower G components involved in the comparison;

(3) class: row 0, column n pixel f (0, n) participates in the comparison with the left and lower G components;

(4) class: the 1 st row to the n-1 st row, the 0 th column pixels f (1-n-1, 0) are used above,

The lower and right G components participate in the comparison;

(5) class: the 1 st to m-1 st rows 1 st to n-1 st columns of pixels f (1 to m-1,1 to n-1) participate in comparison by using four G components of the pixels f up, down, left and right;

(6) class: the n-th column f (1-m-1, n) of the 1 st to m-1 st rows participates in comparison with the upper part, the left part and the lower part thereof;

(7) class: the m-th row, 0-th column pixel f (m, 0) participates in the comparison with its upper and right G components;

(8) class: the m-th row 1-n-1 column pixels f (m, 1-n-1) participate in comparison with their left, upper and right G components;

(9) class: the m-th row and n-th column pixels f (m, n) participate in comparison by using the upper and right G components thereof to obtain a 2-bit small mark, wherein binary 00 represents the upper direction of the value, 01 represents the right direction of the value, 10 represents the left direction of the value, and 11 represents the lower direction of the value, as shown in fig. 2;

step three: when the odd lines are arranged to take the value RG_GB, the even lines take the value GB_RG, when the odd lines are arranged to take the value GB_RG, the even lines take the value RG_GB, the staggered value is kept, the R component removed by the pixel point is kept to have the B component in the up, down, left and right directions of the neighborhood, the B component removed by the pixel point is kept to have the R component in the up, down, left and right directions of the neighborhood, the R/B component arrangement is confirmed by the boundary indication method and the type II, after the small mark of 2 bits after arithmetic comparison is obtained, the G component is kept unchanged, the 2bit mark is replaced to the lowest bit 2bit of R/B in the reserved RG/GB, so that the compression process is completed by obtaining the compressed 1/3 image data, the image compressed data is stored in a buffer area in a pipelining mode and is sent to the next-stage hardware equipment;

step four: when the image is decompressed, the image data can be decompressed out from the first line when the image data is read to the second line, the buffer data is sequentially read, the R/B component missing from each pixel point is obtained according to the boundary mark type by utilizing the R/B component lowest 2bit mark in the received rG_Gb/Gb_rG, a complete RGB three-primary-color pixel is formed, when the small mark is 00, the R/B component corresponding to the sequence number of the pixel column in the last line is represented by the small mark, when the small mark is 01, the R/B component corresponding to the sequence number of the pixel column in the same line is represented by the small mark, namely the R/B component in the left side is represented by the small mark, and when the small mark is 11, the R/B component corresponding to the sequence number of the pixel column in the next line is represented by the small mark.

Referring to fig. 1 to 3, the method is implemented by using an FPGA of MachXO3 series LCMXO3LF-2100C of Lattice company as a processor, inputting image data pixel by pixel in a pipeline manner, using an internal RAM module of the processor for line buffering, storing and reading data in the buffer in FIFO first-in first-out mode, starting when the second line of input data arrives, specifying a target G component value as G (x, y) corresponding to a line as G (x+1, y), a left neighborhood component value as G (x, y-1) of the same line, a right neighborhood value as G (x, y+1) of the same line, a next line as G (x+1, y) of the same line, and x and y as pixel coordinates, and includes:

step one, upper gvalue00

Gvalue 00=g (x-1, y) -G (x, y) for G (x-1, y) > = G (x, y)

Gvalue 00=g (x, y) -G (x-1, y) for G (x, y) <=g (x, y)

Left gvalue10

Gvalue 10=g (x, y-1) -G (x, y) for G (x, y-1) > = G (x, y)

Gvalue 10=g (x, y) -G (x, y-1) given that G (x, y-1) <=g (x, y)

Right gvalue01

Gvalue 01=g (x, y+1) -G (x, y)

Gvalue 01=g (x, y) -G (x, y+1) for G (x, y) <=g (x, y)

Lower gvalue11

Gvalue 11=g (x+1, y) -G (x, y) is given by G (x+1, y) > = G (x, y)

Gvalue 11=g (x, y) -G (x+1, y) is given by G (x+1, y) <=g (x, y)

Step two, the magnitude comparison and arithmetic subtraction are carried out to obtain a difference gvalue00/01/10/11 (two, three values are shown as the boundary values according to 202-210 of the graph_2) of the G component in four directions, the four values are used for carrying out the minimum value comparison again, and the final magnitude Mark is obtained, namely

mark=Min (gvalue 00, gvalue01, gvalue10, gvalue 11), the minimum value result corresponds to the Mark value, the situation that two values are equal in the process indicates that the gradient of the image in the direction is gentle, the pixel difference is small, the selected value has no influence on the final result, and the steps are carried out on pixel points by pixel point according to the priority value of a comparison method;

in the original image RGB sequence, each pixel directly discards an odd sequence B component value, an even sequence discards an R component value, the R component value is changed into rG and Gb after the combination is taken as a small mark, the R component value is discarded by the even sequence in the same way, the B component value is discarded by the odd sequence, the Gb and rG are changed into the odd sequence after the combination is taken as a small mark, the R/G/B bit width is the same as that of the original image, and the 1/3 lossy compression process of the image data is completed;

step four, as shown in fig. 3, the next stage of data transmission is a decompression process, the final restored image data is output to the display terminal, the decompression process is the inverse process of compression, the compressed data of the buffer block is read, the line-by-line FIFO storage is performed according to the pipeline mode, when the second data arrives, the lowest 2 bits of the r value of the odd line and the odd column, namely the Mark corresponding bit is read, if

Mark=00 indicates that the B component of the pixel point is a component value of the corresponding position of the last row, if

Mark=01 indicates that the B component of the pixel is a component value of the right corresponding position of the selected row, if

Mark=10 indicates that the B component of the pixel is a component value of the corresponding position on the left side of the line, if

Mark=11 indicates that the B component of the pixel point is a component value (the boundary value is a value according to the condition shown by 202-210 of fig. 2) of a corresponding position of the next row, and the lowest 2bit is directly assigned with 00;

reading the Mark with the lowest value of 2 bits of the b value of the even number columns of the odd number rows, if

Mark=00 indicates that the R component of the pixel is the component value of the corresponding position of the last row, if

Mark=01 indicates that the R component of the pixel is the component value of the right corresponding position of the selected row, if

Mark=10 indicates that the R component of the pixel is the component value of the corresponding position on the left side of the selected row, if

Mark=11 indicates that the R component of the pixel point is a component value (the boundary value is a value according to the condition shown by 202-210 of fig. 2) of a corresponding position of the next row, and the lowest 2bit is directly assigned with 00;

and similarly, taking the Mark with the same b/r value and the same initial arrangement of Gb_rG as the initial arrangement according to the method, and completing the decompression process.

Although embodiments of the invention have been shown and described, the detailed description is to be construed as exemplary only and is not limiting of the invention as the particular features, structures, materials, or characteristics may be combined in any suitable manner in any one or more embodiments or examples, and modifications, substitutions, variations, etc. may be made in the embodiments as desired by those skilled in the art without departing from the principles and spirit of the invention, provided that such modifications are within the scope of the appended claims.

Claims (3)

1. A method for lossy compression based on RGB three primary color images, characterized by: the method comprises the following steps:

s1: sampling an input RGB image, performing RAM_buffer storage according to row arrangement, and setting a first row, a first column, a last row and a last column of pixel marks;

s2: performing a small value marking by using the G component of each pixel to perform a near neighborhood;

s3: the original value RGB of odd columns and odd columns is RG, the original value RGB of even columns and even columns of odd columns is GB, the value GB of even columns and the value RG of even columns and even columns are GB;

s4: replacing the minimum 2bit of the R component in the sample RG with the small mark, and replacing the minimum 2bit of the B component in the GB with the small mark corresponding to the value of the next sample;

s5: storing the combined rGGb/GbrG data into a cache from line sequence and outputting the data;

s6: the rGGb/GbrG data is taken out from the cache and is stored in a row RAM_buffer according to row arrangement;

s7: generating a first row of pixel marks, a first column of pixel marks, a last row of pixel marks and a last column of pixel marks of the boundary;

s8: taking the R/B component missing from the pixel point from the neighboring domain by using the boundary mark combination to take the small mark 2 bit;

s9: and outputting the restored RGB pixels to a display terminal according to the rows and the columns to display an image.

2. A method of lossy compression based on RGB three primary color images as claimed in claim 1, characterized in that: the green G component sensitive to human eyes is used as a comparison factor, the neighborhood G component is arithmetically compared, and R/B components corresponding to the pixel are removed in a staggered mode according to the odd-even rows and the even-odd columns to be used as a compression mode, so that the purpose of 1/3 lossy compression is achieved.

3. A method of lossy compression based on RGB three primary color images as claimed in claim 1, characterized in that: the method and the steps of the method are as follows:

step one: inputting RGB three primary color images with m+1xn+1 resolution, caching the images into a RAM_buffer line by line, recording row and column sequence number marks, odd and even row and column marks, row sequence numbers 0-m and column sequence numbers 0-n;

step two: the second line of the input image data starts to carry out size comparison by using the pixel point G component;

(1) class: row 0, column 0 pixel f (0, 0) is compared with the G component to the right and below;

(2) class: row 0, column 1 through column n-1 pixel f (0, 1 through n-1) with its left, right, and lower G components involved in the comparison;

(3) class: row 0, column n pixel f (0, n) participates in the comparison with the left and lower G components;

(4) class: row 1 through row n-1, column 0 pixels f (1-n-1, 0) participate in the comparison with their upper, lower and right G components;

(5) class: the 1 st to m-1 st rows 1 st to n-1 st columns of pixels f (1 to m-1,1 to n-1) participate in comparison by using four G components of the pixels f up, down, left and right;

(6) class: the n-th column f (1-m-1, n) of the 1 st to m-1 st rows participates in comparison with the upper part, the left part and the lower part thereof;

(7) class: the m-th row, 0-th column pixel f (m, 0) participates in the comparison with its upper and right G components;

(8) class: the m-th row 1-n-1 column pixels f (m, 1-n-1) participate in comparison with their left, upper and right G components;

(9) class: the m-th row and n-th column pixels f (m, n) are used for the comparison by the upper and right G components to obtain a 2bit small mark, wherein binary 00 represents the upper direction of the value, 01 represents the right direction of the value, 10 represents the left direction of the value, and 11 represents the lower direction of the value

Step three: when the odd-numbered lines are arranged to take the value RG_GB, the even-numbered lines take the value GB_RG, when the odd-numbered lines are arranged to take the value GB_RG, the even-numbered lines take the value RG_GB, the staggered value is kept, the R component removed by the pixel point is kept to have the B component in the four directions of the neighborhood, the B component removed by the pixel point is kept to have the R component in the four directions of the neighborhood, the boundary indication method and the type II confirm the arrangement of the R/B component, after the small mark of 2 bits after arithmetic comparison is obtained, the G component is kept unchanged, the 2bit mark is replaced to the lowest bit 2bit of R/B in the reserved RG/GB, so that the compression process is completed by obtaining the compressed 1/3 image data, the image compressed data is stored in a buffer area in a pipelining mode and is sent to the next-stage hardware equipment

Step four: when the image is decompressed, the image data can be decompressed out from the first line when the image data is read to the second line, the buffer data is sequentially read, the R/B component missing from each pixel point is obtained according to the boundary mark type by utilizing the R/B component lowest 2bit mark in the received rG_Gb/Gb_rG, a complete RGB three-primary-color pixel is formed, when the small mark is 00, the R/B component corresponding to the sequence number of the pixel column in the last line is represented by the small mark, when the small mark is 01, the R/B component corresponding to the sequence number of the pixel column in the same line is represented by the small mark, namely the R/B component in the left side is represented by the small mark, and when the small mark is 11, the R/B component corresponding to the sequence number of the pixel column in the next line is represented by the small mark.