CN105761208A - Picture fusing and splicing method - Google Patents

Abstract

The invention discloses a picture fusing and splicing method, and relates to the field of computer technology. The method comprises the following steps that (1) the picture amount G*G is calculated, G=int(N/A), and each splicing unit is processed in a circulated manner; (2) in each basic splicing unit, pictures are spliced and fused in the horizontal direction, picture data in the same row is merged into one picture, the pictures of the different rows are merged successively, pictures are then spliced and fused in the vertical direction, and A*A pictures are merged into one picture; and (3) when N cannot be divided by A exactly, data of N%A rows is remained according to a A*A splicing rule, the picture amount is maintained in a manner that G*G=int(N/A)*int(N/A), and thus, the pictures are taken into the splicing units nearby, namely the size of the splicing unit in the edge ranges from A*A and 2A*2A. The picture fusing and splicing method has the advantages that occupied memory during picture capture is reduced and the stability of picture capture is high.

Description

A kind of image co-registration joining method

Technical field

The present invention relates to field of computer technology, particularly relate to a kind of image co-registration joining method.

Background technology

Magnanimity three-dimensional data is carried out high definition sectional drawing by conventional needle, it is necessary to data are loaded completely, sometimes will exhausting because of memory source, cause that the exception of picture effect or picture can not normally generate.In high definition sectional drawing scheme, it is multi-block data sectional drawing respectively by screen area cutting, it is general way, but the premise that realizes of this way is that the data for small pieces carry out, it may be possible to the amplification to a piece of community, it is also likely to be the amplification to a house, building, it is not managed for internal memory, therefore, in these sectional drawing processes, being difficult to occur memory source to take too high problem, stability can be guaranteed.But, in whole city, the data in the even whole nation are when needing sectional drawing, the big data quantity bottleneck problem that can run into, the problem that there will be EMS memory occupation height, sectional drawing poor stability in high definition sectional drawing process.

Summary of the invention

The present invention is directed in prior art the shortcoming that EMS memory occupation is high, sectional drawing is unstable, it is provided that a kind of image co-registration joining method.

In order to solve above-mentioned technical problem, the present invention is addressed by following technical proposals:

A kind of image co-registration joining method, comprises the following steps:

(1) picture is N*N dispersion file, and basic splicing unit is A*A, calculates picture number G*G, G=int (N/A) according to ranks number N*N, basic splicing unit A*A, then starts the cycle over and each splicing unit is processed；

(2) in each basic splicing unit, the picture splicing first carried out in horizontal direction is merged, and the image data in same a line is merged into a pictures, and each row is sequentially carried out merging；After row used has merged, the picture splicing then carried out in vertical direction is merged, and A*A pictures synthesizes a pictures the most at last；

(3) when N does not divide exactly A time, splicing rule according to A*A, have more the data of N%A ranks, final picture number to be maintained at G*G=int (N/A) * int (N/A), therefore they are grouped in the concatenation unit that they are neighbouring, are namely in the size of concatenation unit at edge between A*A and 2A*2A.

Due to the fact that and have employed above technical scheme, there is significant technique effect: picture fusion method effectively processes the relation between the final number of achievement picture and individual resolution, the process facilitating the later stage operates, it is to avoid in sectional drawing process, EMS memory occupation is high and cause the phenomenon of instability.

Detailed description of the invention

The present invention is described in further detail by example below.

Embodiment 1

A kind of image co-registration joining method, screen area view data is carried out N etc. horizontally and vertically and divides operation, divide the on-screen data region unit obtaining the sizes such as N*N, each on-screen data region unit is referred to as elementary area, it is identified with n row n row, because sectional drawing is the scene image data intercepting whole screen, need to be divided into screen N*N block, and result map chip resolution is necessary for N times of screen size, then on each on-screen data region unit, by arranging specific sectional drawing parameter, the data of this on-screen data region unit are amplified N times, sectional drawing, the picture size generated is identical with whole screen size, also it is that effect effect is amplified N times, thus realizing the function of high definition sectional drawing, further comprising the steps of:

Pre-treatment step (1): arranging the projection relation of n row n row elementary area, screen resolution, projection relation is rectangular projection, and calculate the viewport size of n row n row elementary area, what comes into a driver's body size；

Pre-treatment step (2): in the elementary area of certain row row, carries out data area segmentation；

Pre-treatment step (3): data area segmentation includes following control parameter: V represents that data are predetermined and loads threshold value, and m represents the length starting zoning in vertical direction from viewpoint position；First, in from distance viewpoint 0 to the scope of m, carry out the preload operation of a secondary data, when the data actual loaded numerical value C obtained exceedes the predetermined loading threshold value V of setting, reduce m value, carrying out repartitioning of region, if after repartitioning, C value still exceedes V-value, again carry out reducing m Value Operations, so repeatedly, until C value stabilization is within the scope of V-value, m value obtains the m ' value determined；When C value is within the scope of V-value, then proceeding by the loading of three-dimensional data, after data have loaded, render, rendering result is saved in image frame buffer；

Pre-treatment step (4): then data area continues to divide, and scope is from m ' to 2m ', it is judged that whether C value is within the scope of V-value, if it is, data area divides scope to adjacent 2m ' to 3m '；Scope is from 2m ' to 3m ', it is judged that whether C value is within the scope of V-value, if it is, data area divides scope to adjacent 3m ' to 4m ', and repeats this operation；If C value exceedes V-value, then carrying out the operation of step 3, next region divides the m ' value of scope and confirms according to the m ' value of upper and adjacent area, and stop when ranging up to critical distance set in advance until dividing, then single image unit sectional drawing completes；

Pre-treatment step (5): the image information being saved in by elementary area in image frame buffer processes, after the every frame of image frame buffer receives image information, image frame buffer preserves the depth buffer of image, color cache information, the image information that contrast receives, by carrying out pixel replacement according to depth buffer information, the pixel of same position, it is little that what depth information was big can not replace depth information, namely depth information is more little, and it is more high that pixel replaces priority；

Pre-treatment step (6): it is exactly the final image-forming information of this elementary area that image frame buffer processes final result, is written in picture file by the image-forming information of this elementary area and preserves, with the data that this picture file of tense marker is n row n row；

Pre-treatment step (7): carry out the shot operation of next elementary area, repeats the pre-treatment step 3 operation to pre-treatment step 6；

, can there is the N*N picture opened in pre-treatment step (8): after all of elementary area is all write as picture file；

Step (1): picture is N*N dispersion file, and basic splicing unit is A*A, calculates picture number G*G, G=int (N/A) according to ranks number N*N, basic splicing unit A*A, then starts the cycle over and each splicing unit is processed；

Step (2): in each basic splicing unit, the picture splicing first carried out in horizontal direction is merged, and the image data in same a line is merged into a pictures, and each row is sequentially carried out merging；After row used has merged, the picture splicing then carried out in vertical direction is merged, and A*A pictures synthesizes a pictures the most at last；Step (3): when N does not divide exactly A time, splicing rule according to A*A, have more the data of N%A ranks, final picture number to be maintained at G*G=int (N/A) * int (N/A), therefore they are grouped in the concatenation unit that they are neighbouring, are namely in the size of concatenation unit at edge between A*A and 2A*2A.

Embodiment 2

With embodiment 1, institute is perspective projection the difference is that projection relation.

Embodiment 3

With embodiment 1, institute is the difference is that, after certain sub-region divides, when the ratio of C value and V-value is equal to 0.2, carrying out m value and increase.

Embodiment 4

A kind of image co-registration joining method, screen area view data is carried out 18 grades horizontally and vertically and divides operation, divide the on-screen data region unit obtaining the sizes such as 18*18, each on-screen data region unit is referred to as elementary area, it is identified with n row n row, because sectional drawing is the scene image data intercepting whole screen, need to be divided into screen 18*18 block, and result map chip resolution is necessary for 18 times of screen size, then on each on-screen data region unit, by arranging specific sectional drawing parameter, the data of this on-screen data region unit are amplified 18 times, sectional drawing, the picture size generated is identical with whole screen size, also it is that effect effect is amplified 18 times, thus realizing the function of high definition sectional drawing, further comprising the steps of:

Pre-treatment step (1): arranging the projection relation of n row n row elementary area, screen resolution, projection relation is rectangular projection, and calculate the viewport size of n row n row elementary area, what comes into a driver's body size；

Pre-treatment step (2): in the elementary area of certain row row, carries out data area segmentation；

Pre-treatment step (3): data area segmentation includes following control parameter: V represents that data are predetermined and loads threshold value, m represents the length starting zoning in vertical direction from viewpoint position, the present embodiment m is 500 meters, V represents predetermined loading data chained list membership, and predetermined loading data chained list membership is 500；First, in the scope of distance viewpoint 0 to 500 meters, carry out the preload operation of a secondary data, when the data actual loaded numerical value C obtained exceedes the predetermined loading threshold value 500 of setting, C is 800, reduces m value, carries out repartitioning of region, when m value reduces, C value also declines accordingly, if after repartitioning, C value still exceedes V-value, again carry out reducing m Value Operations, so repeatedly, until C value stabilization is within the scope of V-value, m value obtains the m ' value determined, m ' value is 250 meters, and C value also drops to 400 accordingly；When C value is within the scope of V-value, then proceeding by the loading of three-dimensional data, after data have loaded, render, rendering result is saved in image frame buffer；

Pre-treatment step (4): then data area continues to divide, and scope is from m ' to 2m ', it is judged that whether C value is within the scope of V-value, if it is, data area divides scope to adjacent 2m ' to 3m '；Scope is from 2m ' to 3m ', it is judged that whether C value is within the scope of V-value, if it is, data area divides scope to adjacent 3m ' to 4m ', and repeats this operation；If C value exceedes V-value, then carrying out the operation of step 3, next region divides the m ' value of scope and confirms according to the m ' value of upper and adjacent area, and stop when ranging up to critical distance set in advance until dividing, then single image unit sectional drawing completes；

Pre-treatment step (5): the image information being saved in by elementary area in image frame buffer processes, after the every frame of image frame buffer receives image information, image frame buffer preserves the depth buffer of image, color cache information, the image information that contrast receives, by carrying out pixel replacement according to depth buffer information, the pixel of same position, it is little that what depth information was big can not replace depth information, namely depth information is more little, and it is more high that pixel replaces priority；

Pre-treatment step (6): it is exactly the final image-forming information of this elementary area that image frame buffer processes final result, is written in picture file by the image-forming information of this elementary area and preserves, with the data that this picture file of tense marker is n row n row；

Pre-treatment step (7): carry out the shot operation of next elementary area, repeats the pre-treatment step 3 operation to pre-treatment step 6；

, can there is the 18*18 picture opened in pre-treatment step (8): after all of elementary area is all write as picture file；

Step (1): picture is 18*18 dispersion file, basic splicing unit is 5*5, calculate picture number G*G, G=int (18/5)=3 according to ranks number 18*18, basic splicing unit 5*5, then start the cycle over and each splicing unit is processed；

Step (2): in each basic splicing unit, the picture splicing first carried out in horizontal direction is merged, and the image data in same a line is merged into a pictures, and each row is sequentially carried out merging；After row used has merged, the picture splicing then carried out in vertical direction is merged, and 25 pictures synthesize a pictures the most at last；

Step (3): when 18 do not divide exactly 5 time, splicing rule according to 5*5, have more the data of 3 row 3 row, final picture number to be maintained at G*G=3*3=9, therefore they are grouped in the concatenation unit that they are neighbouring, are namely in the size of concatenation unit at edge between 5*5 and 10*10.

In a word, the foregoing is only presently preferred embodiments of the present invention, all equalizations made according to the present patent application the scope of the claims change and modify, and all should belong to the covering scope of patent of the present invention.

Claims (1)

1. an image co-registration joining method, it is characterised in that comprise the following steps:

(1) picture is N*N dispersion file, and basic splicing unit is A*A, calculates picture number G*G, G=int (N/A) according to ranks number N*N, basic splicing unit A*A, then starts the cycle over and each splicing unit is processed；

(2) in each basic splicing unit, the picture splicing first carried out in horizontal direction is merged, and the image data in same a line is merged into a pictures, and each row is sequentially carried out merging；After row used has merged, the picture splicing then carried out in vertical direction is merged, and A*A pictures synthesizes a pictures the most at last；

(3) when N does not divide exactly A time, splicing rule according to A*A, have more the data of N%A ranks, final picture number to be maintained at G*G=int (N/A) * int (N/A), therefore they are grouped in the concatenation unit that they are neighbouring, are namely in the size of concatenation unit at edge between A*A and 2A*2A.