CN100578544C - Method for sampling the digital image with 9:4 scale - Google Patents

Abstract

The invention relates to a method for carrying out 9 to 4 down-sapling on a digital image which includes the following steps: A. an inputted image is divided into a plurality of image blocks with size of 3 multiplied by 3; B. 9 to 4 nonlinearity downsapling is carried out on the image obtained by step a to obtain image blocks with size of 2 multiplied by 2; C. the image blocks obtained by step B are arranged according to the positions of the corresponding image blocks before downsapling and recombining a complete output image. The invention aims at providing a method for carrying out 9 to 4 downsapling on the digital image which has small calculation capacity, can not loss the edge of a target and can effectively maintain the details in the image so as to lead the content of the image to adapt the requirements of a display with smaller resolution factor and other applications.

Description

A kind of method of digital picture being carried out 9 to 4 down-samplings

Technical field

The present invention relates to a kind of disposal route of digital picture, be specifically related to a kind of method of digital picture down-sampling.

Background technology

During video and rest image were handled, down-sampling and up-sampling technology were widely used in before the coding step or after the decoding step, so that make the content of image be fit to the spatial resolution of display or the needs of other application.At present, the most frequently used Downsapling method is neighbor interpolation method and bilinear interpolation.

The thought of neighbor interpolation method is: in calculating output image during the pixel value of specified point, the coordinate of this point is obtained a floating-point coordinate by reciprocal transformation, it is simply rounded, obtain an integer type coordinate, the pixel value of this integer type coordinate corresponding pixel in input picture is exactly the result who needs.When carrying out 9 to 4 down-samplings, its essence is exactly directly to extract the result of 4 pixels after as down-sampling according to certain rule in 9 pixels.The thought of bilinear interpolation is: in calculating output image during the pixel value of specified point, the coordinate of this point is obtained a floating-point coordinate by reciprocal transformation, the pixel value of 4 pixels around this floating-point coordinate in the input picture is carried out bilinear interpolation, thereby obtain the result of needs.

The details of above-mentioned two kinds of algorithms is seen 117 pages in " Digital Image Processing " (process plate) book the 8th chapter, the author: Kenneth R.Castleman, publishing house of Tsing-Hua University published in 1998.The advantage of neighbor interpolation method is that operand is little, and shortcoming is the situation that can there be the lose objects edge in the output image that obtains behind the down-sampling, influences the subjective effect of image.The bilinear interpolation advantage be can lose objects the edge, shortcoming is that the details in the image can fog, and operand is bigger.

Summary of the invention

The objective of the invention is to adapt to the needs of the less display of spatial resolution or other application for the content that makes image, it is little to propose a kind of operand, edge that can lose objects, what can effectively keep details in the image carries out 9 to 4 non-linear Downsapling methods to digital picture.

The present invention is achieved by the following technical solutions:

A kind of digital picture is carried out the method for 9 to 4 down-samplings, comprise digital picture is cut apart, further comprising the steps of:

(1) digital picture with input is divided into the image block of several 3*3 sizes;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁Element constitutes, b ₁₂Element is by a ₁₂, a ₁₃The element arithmetic mean constitutes, b ₂₁Element is by a ₂₁, a ₃₁The element arithmetic mean constitutes, b ₂₂Element is by a ₂₂, a ₂₃, a ₃₂, a ₃₃The element arithmetic mean constitutes;

(3) place the position of the image block of the 2*2 image block that step (2) is obtained before by the down-sampling of correspondence, is reassembled into a complete output image, gets final product.

A kind of digital picture is carried out the method for 9 to 4 down-samplings, the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₁₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃Element constitutes, b ₂₁Element is by a ₂₁, a ₂₂, a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₂₃, a ₃₃The element arithmetic mean constitutes.

A kind of digital picture is carried out the method for 9 to 4 down-samplings, the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₁₂, a ₂₁, a ₂₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃, a ₂₃The element arithmetic mean constitutes, b ₂₁Element is by a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₃₃Element constitutes.

A kind of digital picture is carried out the method for 9 to 4 down-samplings, the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₂₁The element arithmetic mean constitutes, b ₁₂Element is by a ₁₂, a ₁₃, a ₂₂, a ₂₃The element arithmetic mean constitutes, b ₂₁Element is by a ₃₁Element constitutes, b ₂₂Element is by a ₃₂, a ₃₃The element arithmetic mean constitutes.

The beneficial effect of the inventive method is, image down sampling can be grown up and wide all be the downscaled images of life size 2/3 size.Make the content of image can adapt to the needs of the less display of spatial resolution or other application.Compare with the operand of known Downsapling method, 4 points of the every computing of this method only need 5 sub-additions and 3 shift operations.1 point of the every computing of bilinear interpolation just needs 4 sub-additions, 4 subtractions and 4 multiplication (to see the formula 6 of 119 pages in " Digital Image Processing " (process plate) book the 8th chapter, author: Kenneth R.Castleman, publishing house of Tsing-Hua University, published in 1998), so this method operand is little a lot.Compare with the effect of known Downsapling method, the edge that this method can lose objects can effectively keep the details of image.

To utilize this that digital picture is carried out the progress of the method and the existing method of 9 to 4 down-samplings in order further proving, to have done following experiment, concrete outcome is referring to accompanying drawing 1～4.

In the accompanying drawing, the word segment that superposes in Fig. 2 image owing to lose the edge, is beyond recognition behind the down-sampling.The right side top of car is watched in Fig. 3 and Fig. 4 contrast, and the details of Fig. 4 is than the clarity of detail of Fig. 3; Die Jia word segment in addition, Fig. 4 and Fig. 3 can recognize, but the details of Fig. 4 is than the clarity of detail of Fig. 3.Therefore this experimental results show that the down-sampling better effects if of this method.

Description of drawings

Fig. 1 carries out the preceding image of digital picture down-sampling.

Fig. 2 is the output image of neighbor interpolation method down-sampling.

Fig. 3 is the output image of bilinear interpolation down-sampling.

Fig. 4 is the output image of the inventive method down-sampling.

Fig. 5 is the process flow diagram of the inventive method.

Embodiment

Referring to shown in Figure 5.

Embodiment 1

Digital picture is carried out the method for 9 to 4 down-samplings, may further comprise the steps:

(1) digital picture with input is divided into the image block of several 3*3 sizes; The method of cutting apart is similar to the piece that in the MPEG4 image compression algorithm image segmentation is become the 16*16 size, and difference only is that the size of this method image block becomes 3*3;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁Element constitutes, b ₁₂Element is by a ₁₂, a ₁₃The element arithmetic mean constitutes, b ₂₁Element is by a ₂₁, a ₃₁The element arithmetic mean constitutes, b ₂₂Element is by a ₂₂, a ₂₃, a ₃₂, a ₃₃The element arithmetic mean constitutes; The physical relationship expression formula is as follows:

[formula 1]

b ₁₁＝a ₁₁

[formula 2]

b ₁₂＝(a ₁₂+a ₁₃)/2

[formula 3]

b ₂₁＝(a ₂₁+a ₃₁)/2

[formula 4]

b ₂₂＝(a ₂₂+a ₂₃+a ₃₂+a ₃₃)/4

(3) place the position of the image block of the 2*2 image block that step (2) is obtained before by the down-sampling of correspondence, is reassembled into a complete output image, gets final product.

Embodiment 2

Digital picture is carried out the method for 9 to 4 down-samplings, may further comprise the steps:

(1) digital picture with input is divided into the image block of several 3*3 sizes; The method of cutting apart is similar to the piece that in the MPEG4 image compression algorithm image segmentation is become the 16*16 size, and difference only is that the size of this method image block becomes 3*3;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁, a ₁₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃Element constitutes, b ₂₁Element is by a ₂₁, a ₂₂, a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₂₃, a ₃₃The element arithmetic mean constitutes.The physical relationship expression formula is as follows:

[formula 1]

b ₁₁＝(a ₁₁+a ₁₂)/2

[formula 2]

b ₁₂＝a ₁₃

[formula 3]

b ₂₁＝(a ₂₁+a ₂₂+a ₃₁+a ₃₂)/4

[formula 4]

b ₂₂＝(a ₂₃+a ₃₃)/2

(3) place the position of the image block of the 2*2 image block that step (2) is obtained before by the down-sampling of correspondence, is reassembled into a complete output image, gets final product.

Embodiment 3

Digital picture is carried out the method for 9 to 4 down-samplings, may further comprise the steps:

(1) digital picture with input is divided into the image block of several 3*3 sizes; The method of cutting apart is similar to the piece that in the MPEG4 image compression algorithm image segmentation is become the 16*16 size, and difference only is that the size of this method image block becomes 3*3;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁, a ₁₂, a ₂₁, a ₂₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃, a ₂₃The element arithmetic mean constitutes, b ₂₁Element is by a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₃₃Element constitutes.The physical relationship expression formula is as follows:

[formula 1]

b ₁₁＝(a ₁₁+a ₁₂+a ₂₁+a ₂₂)/4

[formula 2]

b ₁₂＝(a ₁₃+a ₂₃)/2

[formula 3]

b ₂₁＝(a ₃₁+a ₃₂)/2

[formula 4]

b ₂₂＝a ₃₃

(3) place the position of the image block of the 2*2 image block that step (2) is obtained before by the down-sampling of correspondence, is reassembled into a complete output image, gets final product.

Embodiment 4

Digital picture is carried out the method for 9 to 4 down-samplings, may further comprise the steps:

(1) digital picture with input is divided into the image block of several 3*3 sizes; The method of cutting apart is similar to the piece that in the MPEG4 image compression algorithm image segmentation is become the 16*16 size, and difference only is that the size of this method image block becomes 3*3;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁, a ₂₁The element arithmetic mean constitutes, b ₁₂Element is by a ₁₂, a ₁₃, a ₂₂, a ₂₃The element arithmetic mean constitutes, b ₂₁Element is by a ₃₁Element constitutes, b ₂₂Element is by a ₃₂, a ₃₃The element arithmetic mean constitutes.The physical relationship expression formula is as follows:

[formula 1]

b ₁₁＝(a ₁₁+a ₂₁)/2

[formula 2]

b ₁₂＝(a ₁₂+a ₁₃+a ₂₂+a ₂₃)/4

[formula 3]

b ₂₁＝a ₃₁

[formula 4]

b ₂₂＝(a ₃₂+a ₃₃)/2

Claims (4)

1, a kind of digital picture is carried out the method for 9 to 4 down-samplings, comprises digital picture is cut apart, it is characterized in that may further comprise the steps:

(1) digital picture with input is divided into the image block of several 3*3 sizes;

(2) step (1) is cut apart the image block that obtains as the input data, is made as: $\left[\begin{array}{ccc}{a}_{11}& a_{12}& a_{13}\\ a_{21}& a_{22}& a_{23}\\ a_{31}& a_{32}& a_{33}\end{array}\right],$ Be designated as A; The output data of down-sampling is the image block of 2*2 size, is made as: $\left[\begin{array}{cc}{b}_{11}& b_{12}\\ b_{21}& b_{22}\end{array}\right],$ Be designated as B; The mode that sampling is chosen is: b ₁₁Element is by a ₁₁Element constitutes, b ₁₂Element is by a ₁₂, a ₁₃The element arithmetic mean constitutes, b ₂₁Element is by a ₂₁, a ₃₁The element arithmetic mean constitutes, b ₂₂Element is by a ₂₂, a ₂₃, a ₃₂, a ₃₃The element arithmetic mean constitutes;

(3) place the position of the image block of the 2*2 image block that step (2) is obtained before by the down-sampling of correspondence, is reassembled into a complete output image, gets final product.

2, according to claim 1ly a kind of digital picture is carried out the method for 9 to 4 down-samplings, it is characterized in that the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₁₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃Element constitutes, b ₂₁Element is by a ₂₁, a ₂₂, a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₂₃, a ₃₃The element arithmetic mean constitutes.

3, according to claim 1ly a kind of digital picture is carried out the method for 9 to 4 down-samplings, it is characterized in that the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₁₂, a ₂₁, a ₂₂The element arithmetic mean constitutes, b ₁₂Element is by a ₁₃, a ₂₃The element arithmetic mean constitutes, b ₁₁Element is by a ₃₁, a ₃₂The element arithmetic mean constitutes, b ₂₂Element is by a ₃₃Element constitutes.

4, according to claim 1ly a kind of digital picture is carried out the method for 9 to 4 down-samplings, it is characterized in that the mode that sampling is chosen described in the described step (2) is: b ₁₁Element is by a ₁₁, a ₂₁The element arithmetic mean constitutes, b ₁₂Element is by a ₁₂, a ₁₃, a ₂₂, a ₂₃The element arithmetic mean constitutes, b ₂₁Element is by a ₃₁Element constitutes, b ₂₂Element is by a ₃₂, a ₃₃The element arithmetic mean constitutes.

Images