CN103530639B - A kind of orderly point set extracting method of image outline - Google Patents
A kind of orderly point set extracting method of image outline Download PDFInfo
- Publication number
- CN103530639B CN103530639B CN201310522543.5A CN201310522543A CN103530639B CN 103530639 B CN103530639 B CN 103530639B CN 201310522543 A CN201310522543 A CN 201310522543A CN 103530639 B CN103530639 B CN 103530639B
- Authority
- CN
- China
- Prior art keywords
- idy
- idx
- broken
- point
- selsete
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of orderly point set extracting method of image outline, it is characterised by：Based on a kind of logical topological structure, each point eight broken points of correspondence, a broken point represents a UNICOM direction of point；All broken points are attached, its method is, if a broken point correspondence UNICOM direction reaches UNICOM's condition, otherwise next broken point on the clockwise for the broken point that broken point is pointed into correspondence UNICOM direction is pointed to next broken point on the clockwise of this point.Described clockwise is included clockwise and anticlockwise.Described clockwise is all unified to all points.Novelty of the invention is that：The present invention is a kind of orderly point set extracting method of new image outline.The practicality of the present invention is：A kind of orderly point set extracting method of image outline is provided for computer vision developer, facilitate computer vision developer exploitation to be adapted to the computer vision system of parallel multiprocessor, the orderly point set data of image outline can be more quickly and easily obtained compared with the prior art.
Description
Technical field
The invention belongs to computer vision field, a kind of orderly point set extracting method of image outline is more particularly related to, is belonged to
In (2006.01) No. I classification of G06K9/46 of patent classification entitled " extraction of characteristics of image or characteristic ".
Background technology
The extraction of image outline is a basic core procedure in computer vision, is needed in many occasions to image
Contours extract is simultaneously converted to orderly point set.
The existing orderly point set extracting method of computer picture profile typically using worm with method, raster scanning method etc., these
Method is all difficult to concurrent operation when running, the small image in image inside is more, takes bigger, it is difficult to predict the consumption of contour extraction
When；These methods are difficult to the border that parsing intersects；These method convergence properties are not sufficiently stable, and may be fallen into so-called " trap ".
The content of the invention
The purpose of the present invention is to solve the shortcomings of the prior art proposing a kind of orderly point set extracting method of image outline, solve
Certainly the operation of conventional images profile orderly point set extracting method is slow, the problem of be difficult to parse the border intersected, restrain unstable,
Computer vision developer is conducive to develop computer vision system.
A kind of orderly point set extracting method of image outline, based on a kind of logical topological structure, each point eight broken points of correspondence,
One broken point represents a UNICOM direction of point；All broken points are attached, its method is：If a broken point correspondence UNICOM
Direction reaches UNICOM's condition, then next broken point on the clockwise for the broken point that broken point is pointed to correspondence UNICOM direction, otherwise
Pointed to next broken point on the clockwise of this point.
Further illustrate 1：Described clockwise includes counter clockwise direction and clockwise direction.
Further illustrate 2：Described clockwise is all unified to all points, that is to say, that in a picture processing
The middle clockwise used a little must be consistent, that is to say, that not allow any one point to use different from other points
Clockwise；(referring to Fig. 2).
Further illustrate 3：Described UNICOM's condition be it is variable, by using the present invention computer vision developer according to
Demand and define.
Further illustrate 4：Described unified clockwise is not the clockwise of the boundary chain formed to broken point connection
Restriction, that is to say, that the broken chain that the connection of broken point is produced after finishing can be different clockwises, in fact of the invention broken
The broken chain that point connection is formed, its direction is existing also to be had counterclockwise clockwise.
Further illustrate 5：Due to carrying out after broken concatenation operation, broken of border chain is just ticked, but due to broken point and
Point is corresponding, that is to say, that ticking for broken of border chain, for computer vision developer, is equal to orderly point set
Tick；（Referring to Fig. 3, Fig. 4, Fig. 5）.
Further illustrate 6：Computer vision developer the understanding of the present invention and utilization for convenience, specially makes non-
The basis for estimation of the supplement of invention essential feature --- acnode：Acnode is external connectionless point, broken chain its include
All broken points all correspond to same point, then the broken chain of namely one acnode of this broken chain（Referring to Fig. 3, point P8）.
Further illustrate 7：Computer vision developer the understanding of the present invention and utilization for convenience, specially makes non-
The supplement of invention essential feature --- the basis for estimation 1 of the non-broken chain in border：Because all connections all occur in pairs, then when one
When two adjacent UNICOM directions of individual point are all externally first connected, broken chain where broken point rearward is ranked on clockwise
It must be broken of non-border chain；Referring to Fig. 3, broken chain P0（S6）、P1（S3,S4）, P4 (S1) and broken chain P1（S5,S6）、P2
(S3,S4)、P4(S7,S0)、P5(S5,S2)。
Further illustrate 8：Computer vision developer the understanding of the present invention and utilization for convenience, specially makes non-
The supplement of invention essential feature --- the basis for estimation 2 of the non-broken chain in border：In the case that acnode not can be regarded as effective image, due to
One point eight broken point of correspondence, then the broken points of latter of connection effectively broken chain are certain to be more than eight, the invalid broken points of broken chain
It is certain to be less than or equal to eight；Referring to Fig. 3, broken chain P0（S6）、P1（S3,S4）, P4 (S1) and broken chain P1（S5,S6）、P2(S3,
S4)、P4(S7,S0)、P5(S5,S2)。
Further illustrate 9：Computer vision developer the understanding of the present invention and utilization for convenience, specially makes non-
The supplement of invention essential feature --- the method for simplifying of broken chain, with put for unit along conterclockwise reverse direction successively computing
Eight broken points, if the external not UNICOM of certain broken point, is just pointed to the sensing for being changed to next broken point on clockwise, and will
Next broken point is removed；The number of times that broken point after computing in same point is appeared in same broken chain is reduced；It is worth note
Meaning, can be eliminated because all broken points of acnode are externally connectionless after simplification（Referring to Fig. 5）.
Further illustrate 10：Because UNICOM's condition of all broken points is consistent, that is to say, that on a UNICOM direction
Connection is paired appearance；Because the closure rule for being connected into pairs appearance and all points are used is identical（Clockwise）, institute
It is closure link that the connection one formed with all broken points, which is established a capital, so without worrying convergence problem.
Further illustrate 11：Due to each broken one UNICOM direction of point correspondence, the connection of broken point only with this direction whether
Reach that condition is related, so when two boundary-intersecteds, the present invention can also be detected respectively, without judging conflict.
Further illustrate 12：Because broken concatenation operation on each UNICOM direction is independent, and other UNICOM directions
Situation is unrelated, and the data dependence between broken point is smaller, so the present invention is well suited for concurrent operation.
Further illustrate 13：It understanding of knowable to further illustrating after 1-12, though the present invention creates for parallelization computing,
Its speed of service of relatively conventional method is very fast, the advantage intersected border, do not have to concern convergence problem can be detected, in non-parallel meter
Equally existed in calculation.
Novelty of the invention is that：The present invention is a kind of orderly point set extracting method of new image outline.
The practicality of the present invention is：A kind of orderly point set extraction side of image outline is provided for computer vision developer
Method, facilitates computer vision developer exploitation to be adapted to the computer vision system of parallel multiprocessor, compared with the prior art may be used
More quickly and easily to obtain the orderly point set data of image outline.
Brief description of the drawings
Fig. 1 is one example of logical topological structure proposed by the present invention, and point P abscissa is x, and ordinate is y, color
It is worth for c, the corresponding eight broken points of point P are S0-S7.
Fig. 2 be the present invention an embodiment, embodiment be a 3*3 pixel view data be imported into after logic
Topological structure.
Fig. 3 is another state of Fig. 2 embodiments, embodies logical topology knots of the Fig. 2 after broken concatenation operation
Structure.
Fig. 4 is another state of Fig. 2 embodiments, embodies Fig. 3 and removes the logical topology knot after non-broken chain in border
Structure.
Fig. 5 is another state of Fig. 2 embodiments, embodies Fig. 4 and simplifies broken chain and remove the logic after acnode and opens up
Flutter structure.
Embodiment
Below in conjunction with embodiment, the present invention will be described.
It is an a kind of implementation of the orderly point set extracting method of image outline of the invention as shown in Fig. 2, Fig. 3, Fig. 4, Fig. 5
Example, having handled one has 3*3 pixels, the image with 0,1,3 three kind of color value；UNICOM's condition that this example is used is face
Color is identical.
As shown in Fig. 2 be this embodiment after view data importing, the logical topology knot before broken concatenation operation
Structure；Wherein broken point a little all constitute a clockwise round closure chain；Wherein all auxiliary magnet P (- 1, -1, -1)
Composition prevents ' enclosure wall ' of calculation overflow together, because the broken point in ' enclosure wall ' is not engaged in computing, for clear, the letter of view
It is clean, do not pointed to and drawn in figure.
As shown in figure 3, being logical topological structures of the Fig. 2 after broken concatenation operation, P0（S6）、P1（S3,S4）、P4
(S1) broken of the non-border chain of one " strip " is constituted；P1（S5,S6）、P2(S3,S4)、P4(S7,S0)、P5(S5,S2)
Constitute broken of " cross " non-border chain.
As shown in figure 3, although the border for the image that P3, P7 are constituted and the border of image where P6 are to intersect border, but warp
It is two broken chains that them, which are crossed after broken concatenation operation, by detection respectively.
As shown in figure 3, acnode P8 is externally without UNICOM.
It is that Fig. 3 rejects the logical topological structure after non-broken chain in border, non-broken chain in border is eliminated such as Fig. 4（Mark
It is negative, clear for view is not drawn）.
It is that Fig. 4 carries out the logical topological structure after broken chain simplified operation, the broken point of same point is in same broken point such as Fig. 5
The number of times occurred on chain is reduced, and acnode is also eliminated（I.e. labeled as negative, for not drawing clearly for view）.
As shown in figure 5, the border ordered set that this example is drawn is：（P3,P7）With（P0,P4,P6,P4,P5,P2,P1）, P4
Occur in that twice, tally with the actual situation in corresponding profile ordered set.
The following is this embodiment based on the tall and handsome partial code up to company's CUDA parallel computings, this hair is embodied
The characteristic of bright suitable concurrent operation；It is that computer vision developer understands and using the present invention that the purpose for providing these codes, which is,
Reference is provided.
Struct FR_POINT { // broken points
long next_x;The x values of // broken point pointed to
long next_y;The y values of // broken point pointed to
long next_s;The numbering of // broken the point pointed to
};
struct POINT{
long X;// abscissa
long y;// ordinate
long c;// color value
long selsete[4];// be used to select, it can be combined instead of if sentences according to different situations
Computing, produces different value
FR_POINT S[8];// broken point set
};
POINT P[5][5];// construction one is by dot-patterned structure group
for(i=0;i<5;i++){
// construction enclosure wall, prevents calculation overflow, and its color value -1 is invalid value, to prevent being connected
P[0][i].c = -1;
P[4][i].c = -1;
P[i][0].c = -1;
P[i][4].c = -1;
}
idx=threadIdx.x+1;// thread number x, idx scope 1-3
idy=threadIdx.y+1;// thread number y, idy scope 1-3
if（idx<4 && idy<4）{ // idx scope 1-3, idy scope 1-3
for(t=0;t<8;t++){
P[idx][idy].S[t].next_x=idx;// all broken points are pointed into this point,
P[idx][idy].S[t].next_y=idy;
P[idx][idy].S[t].next_s=（t+1）%8;// broken point is constituted into close ring clockwise
Shape chain, is as a result shown in Fig. 2；
}
P[idx][idy].c=IMAGE_COLOR[idx][idy];// import color of image value
// broken put connects, and concatenation operation is concurrent operation, and operation result is shown in Fig. 3
P[idx][idy].selsete[3] = !P[idx][idy].c-P[idx][idy-1].c；
UNICOM's situation in // preservation S0 correspondence UNICOMs direction, the equal then connection status of color is 1, is otherwise zero
P[idx][idy].S[0].next_y -= P[idx][idy].selsete[3];
If the external UNICOMs of // S0 change y value, the point abscissa in S0 correspondence UNICOMs direction is consistent with S0 points；S0 pairs
The point abscissa x values in the UNICOM direction answered and the identical of this point are so without change
P[idx][idy].S[0].next_s = 4 * P[idx][idy].selsete[3]+1;
If the external UNICOMs of // S0 become S value into 5, S values are 1 if not UNICOM
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx-1][idy-1].c);
P[idx][idy].S[1].next_x -= P[idx][idy].selsete[3];
P[idx][idy].S[1].next_y -= P[idx][idy].selsete[3];
P[idx][idy].S[1].next_s = 4 * P[idx][idy].selsete[3]+2;
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx-1][idy].c);
P[idx][idy].S[2].next_x -= P[idx][idy].selsete[3];
P[idx][idy].S[2].next_s = 4 * P[idx][idy].selsete[3]+3;
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx-1][idy+1].c);
P[idx][idy].S[3].next_x -= P[idx][idy].selsete[3];
P[idx][idy].S[3].next_y += P[idx][idy].selsete[3];
P[idx][idy].S[3].next_s = 4 * !P[idx][idy].selsete[3];
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx][idy+1].c);
P[idx][idy].S[4].next_y += P[idx][idy].selsete[3];
P[idx][idy].S[4].next_s = 4 * !P[idx][idy].selsete[3]+1;
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx+1][idy+1].c)；
P[idx][idy].S[5].next_x += P[idx][idy].selsete[3];
P[idx][idy].S[5].next_y += P[idx][idy].selsete[3];
P[idx][idy].S[5].next_s = 4 * !P[idx][idy].selsete[3]+2;
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx+1][idy].c);
P[idx][idy].S[6].next_x += P[idx][idy].selsete[3];
P[idx][idy].S[6].next_s = 4 * !P[idx][idy].selsete[3]+3;
P[idx][idy].selsete[3] = !(P[idx][idy].c-P[idx+1][idy-1].c);
P[idx][idy].S[7].next_x += P[idx][idy].selsete[3];
P[idx][idy].S[7].next_y += P[idx][idy].selsete[3];
P[idx][idy].S[7].next_s = 4 * P[idx][idy].selsete[3];
// negative is labeled as the following is by broken of non-border chain, to remove broken of non-border chain, operation result joins Fig. 4
P[idx][idy].selsete[0] = -10;
P[idx][idy].selsete[1] = 0;
P[idx][idy].selsete[2] = 0;
P[idx][idy].selsete[3]=!(P[idx][idy].S[0].next_s-1)+ !(P[idx][idy].S
[1].next_s-2);
// judge the adjacent two broken external UNICOM's situation of point S0, S1
P[idx][idy].S[1].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
If all external UNICOM of // S0, S1, S1.next_s is pointed to and is changed to negative value, and representative is eliminated
P[idx][idy].selsete[3]=!(P[idx][idy].S[1].next_s-2)+ !(P[idx][idy].S
[2].next_s-3);
P[idx][idy].S[2].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[2].next_s-3)+ !(P[idx][idy].S
[3].next_s-4);
P[idx][idy].S[3].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[3].next_s-4)+ !(P[idx][idy].S
[4].next_s-5);
P[idx][idy].S[4].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[4].next_s-5)+ !(P[idx][idy].S
[5].next_s-6);
P[idx][idy].S[5].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[5].next_s-6)+ !(P[idx][idy].S
[6].next_s-7);
P[idx][idy].S[6].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[6].next_s-7)+ !(P[idx][idy].S
[7].next_s-0);
P[idx][idy].S[7].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3]=!(P[idx][idy].S[7].next_s-0)+ !(P[idx][idy].S
[0].next_s-1);
P[idx][idy].S[0].next_s += P[idx][idy].selsete[P[idx][idy].selsete
[3]];
// the following is may be not fully erased to computing above broken of non-border chain (such as P1 (S3）) be purged;
P[idx][idy].selsete[3]=!(P[idx][idy].S[0].next_s-1)+ !(P[idx][idy].S
[2].next_s-3);
// the external UNICOM's situation of S0, S2 is preserved, income value is zero if all external UNICOM
P[idx][idy].S[1].next_s += P[idx][idy].selsete[ P[idx][idy].selsete
[3] ];
If the external UNICOM of // S0, S2, S1 sensing is changed into negative
P[idx][idy].selsete[3]=!(P[idx][idy].S[2].next_s-3)+ !(P[idx][idy].S
[4].next_s-5);
P[idx][idy].S[3].next_s += P[idx][idy].selsete[ P[idx][idy].selsete
[3] ];
P[idx][idy].selsete[3]=!(P[idx][idy].S[4].next_s-5)+ !(P[idx][idy].S
[6].next_s-7);
P[idx][idy].S[5].next_s += P[idx][idy].selsete[ P[idx][idy].selsete
[3] ];
P[idx][idy].selsete[3]=!(P[idx][idy].S[6].next_s-0)+ !(P[idx][idy].S
[0].next_s-1);
P[idx][idy].S[7].next_s += P[idx][idy].selsete[ P[idx][idy].selsete
[3] ];
// be below the broken chain in border simplification, operation result is referring to Fig. 5
P[idx][idy].selsete[3] = !（P[idx][idy].S[7].next_s-0）;
// if 0 is pointed to, illustrate the external not UNICOMs of S [7], if the external sensing that S [0] is pointed to not UNICOM,
And remove S [0]
P[idx][idy].selsete[0] = P[idx][idy].S[7].next_x;
Externally the UNICOM abscissa x that then S7 is pointed to is constant by //S [7]
P[idx][idy].selsete[1] = P[idx][idy].S[0].next_x;
Externally the UNICOM abscissa x that then S7 is pointed to is not changed into the abscissa of S [0] sensings to //S [7]
P[idx][idy].S[7].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
// Selecting operation, if
P[idx][idy].selsete[0] = P[idx][idy].S[7].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[0].next_y;
P[idx][idy].S[7].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[7].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[0].next_s;
P[idx][idy].S[7].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
If // S [7] is externally connected, its sensing is not changed in above several line codes, it is necessary to retain S [0]
P[idx][idy].selsete[1] = -10;
If // S [7] is not connected externally, it, which is pointed to, to change in above several line codes, it is necessary to remove S [0], negative value generation
Table is removed
P[idx][idy].S[0].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
// make a choice computing,
P[idx][idy].selsete[3] = !（P[idx][idy].S[6].next_s-7）;// if 7 are pointed to,
Illustrate the external not UNICOMs of S [6]
P[idx][idy].selsete[0] = P[idx][idy].S[6].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[7].next_x;
P[idx][idy].S[6].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[6].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[7].next_y;
P[idx][idy].S[6].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[6].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[7].next_s;
P[idx][idy].S[6].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[7].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[5].next_s-6）;// if 6 are pointed to,
Illustrate the external not UNICOMs of S [5]
P[idx][idy].selsete[0] = P[idx][idy].S[5].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[6].next_x;
P[idx][idy].S[5].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[5].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[6].next_y;
P[idx][idy].S[5].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[5].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[6].next_s;
P[idx][idy].S[5].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[6].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[4].next_s-5）;// if 5 are pointed to,
Illustrate the external not UNICOMs of S [4]
P[idx][idy].selsete[0] = P[idx][idy].S[4].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[5].next_x;
P[idx][idy].S[4].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[4].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[5].next_y;
P[idx][idy].S[4].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[4].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[5].next_s;
P[idx][idy].S[4].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[5].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[3].next_s-4）;// if 4 are pointed to,
Illustrate the external not UNICOMs of S [3]
P[idx][idy].selsete[0] = P[idx][idy].S[3].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[4].next_x;
P[idx][idy].S[3].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[3].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[4].next_y;
P[idx][idy].S[3].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[3].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[4].next_s;
P[idx][idy].S[3].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[4].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[2].next_s-3）;// if 3 are pointed to,
Illustrate the external not UNICOMs of S [2]
P[idx][idy].selsete[0] = P[idx][idy].S[2].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[3].next_x;
P[idx][idy].S[2].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[2].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[3].next_y;
P[idx][idy].S[2].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[2].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[3].next_s;
P[idx][idy].S[2].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[3].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[1].next_s-2）;// if 2 are pointed to,
Illustrate the external not UNICOMs of S [1]
P[idx][idy].selsete[0] = P[idx][idy].S[1].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[2].next_x;
P[idx][idy].S[1].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[1].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[2].next_y;
P[idx][idy].S[1].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[1].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[2].next_s;
P[idx][idy].S[1].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[2].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[3] = !（P[idx][idy].S[0].next_s-1）;// if 1 is pointed to,
Illustrate the external not UNICOMs of S [0]
P[idx][idy].selsete[0] = P[idx][idy].S[0].next_x;
P[idx][idy].selsete[1] = P[idx][idy].S[1].next_x;
P[idx][idy].S[0].next_x = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[0].next_y;
P[idx][idy].selsete[1] = P[idx][idy].S[1].next_y;
P[idx][idy].S[0].next_y = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = P[idx][idy].S[0].next_s;
P[idx][idy].selsete[1] = P[idx][idy].S[1].next_s;
P[idx][idy].S[0].next_s = P[idx][idy].selsete[P[idx][idy].selsete
[3]];
P[idx][idy].selsete[0] = 1;
P[idx][idy].selsete[1] = -10;
P[idx][idy].S[1].next_s *=P[idx][idy].selsete[P[idx][idy].selsete
[3]];
}.
The orderly point set extracting method of a kind of image outline that the present invention is provided, is a kind of method invention, it can pass through
The computer hardwares such as CPU, GPU, MCU, DSP, FPGA are realized.
It can be seen from content of the invention explanation and embodiment, typical embodiment of the invention is not the limit to the present invention
It is fixed, it is any to utilize logical topological structure proposed by the present invention, and carry out broken point company according to broken connection method proposed by the present invention
The orderly point set extracting method of image outline connect, should all belong to scope of the invention.
Claims (2)
1. a kind of orderly point set extracting method of image outline, it is characterised by：Based on a kind of logical topological structure, each
Point eight broken points of correspondence, a broken point represents a UNICOM direction of point；All broken points are attached, its method is,
If a broken point correspondence UNICOM direction reaches UNICOM's condition, broken point is pointed to the hour hands side of the broken point in correspondence UNICOM direction
Upward next broken point, is otherwise pointed to next broken point on the clockwise of this point；
Described clockwise is all unified to all points；
Judge in accordance with the following methods and eliminate broken of non-border chain：If two adjacent UNICOM directions of a point all externally first connect
Connect, then broken chain where broken point rearward is ranked on clockwise must be broken of non-border chain；
Simplify broken of border chain in accordance with the following methods：It is that unit is broken along conterclockwise reverse direction successively computing eight to put
Point, if the external not UNICOM of certain broken point, is just pointed to the sensing for being changed to next broken point on clockwise, and by under
One broken point is removed；The number of times that broken point after computing in same point is appeared in same broken chain is reduced；Due to
All broken points of acnode are externally connectionless to be eliminated after simplification.
2. as described in claim 1, a kind of orderly point set extracting method of image outline is characterised by：Described hour hands
Direction is included clockwise and anticlockwise.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710172510.0A CN106960209B (en) | 2013-10-30 | 2013-10-30 | Image contour ordered point set extraction method |
| CN201310522543.5A CN103530639B (en) | 2013-10-30 | 2013-10-30 | A kind of orderly point set extracting method of image outline |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310522543.5A CN103530639B (en) | 2013-10-30 | 2013-10-30 | A kind of orderly point set extracting method of image outline |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710172510.0A Division CN106960209B (en) | 2013-10-30 | 2013-10-30 | Image contour ordered point set extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103530639A CN103530639A (en) | 2014-01-22 |
| CN103530639B true CN103530639B (en) | 2017-10-31 |
Family
ID=49932636
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710172510.0A Active CN106960209B (en) | 2013-10-30 | 2013-10-30 | Image contour ordered point set extraction method |
| CN201310522543.5A Expired - Fee Related CN103530639B (en) | 2013-10-30 | 2013-10-30 | A kind of orderly point set extracting method of image outline |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710172510.0A Active CN106960209B (en) | 2013-10-30 | 2013-10-30 | Image contour ordered point set extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN106960209B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106960209B (en) * | 2013-10-30 | 2020-08-14 | 浩德科技股份有限公司 | Image contour ordered point set extraction method |
| CN107704861B (en) * | 2017-10-23 | 2021-07-27 | 九竹物联技术有限公司 | Method for constructing transformation template of image contour data structure for computer vision system |
| CN107818539B (en) * | 2017-10-23 | 2021-04-23 | 湖南轻创科技有限公司 | Construction method of conversion template of image contour data structure |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101119429A (en) * | 2006-08-01 | 2008-02-06 | 北京北大方正电子有限公司 | Digital watermark embedded and extracting method and device |
| CN103353991A (en) * | 2013-06-24 | 2013-10-16 | 西安交通大学 | Method for generating contour fragments using iteration analysis |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020191812A1 (en) * | 2001-04-24 | 2002-12-19 | Nam-Deuk Kim | Object edge watermarking |
| CN100407231C (en) * | 2004-06-10 | 2008-07-30 | 中国煤炭地质总局航测遥感局 | Optimizing method for image transfigure border side tracking |
| CN100380406C (en) * | 2006-12-29 | 2008-04-09 | 四川大学 | Contour projection method of tumour 3D anatomy structure in radiotherapy plan system |
| CN100520811C (en) * | 2007-07-11 | 2009-07-29 | 苏州大学 | Method for recognizing outline shape of digital image |
| CN101520851A (en) * | 2008-02-29 | 2009-09-02 | 富士通株式会社 | Character information identification device and method |
| CN101414358B (en) * | 2008-11-18 | 2011-03-16 | 广东威创视讯科技股份有限公司 | Method for detecting and extracting chromosome contour based on directional searching |
| US8238658B2 (en) * | 2009-01-21 | 2012-08-07 | The United States Of America, As Represented By The Secretary Of The Navy | Boundary extraction method |
| CN102270299B (en) * | 2011-08-24 | 2013-06-12 | 复旦大学 | Edge connection algorithm realized in parallel based on breakpoints |
| US8873865B2 (en) * | 2011-10-10 | 2014-10-28 | Qualcomm Incorporated | Algorithm for FAST corner detection |
| CN103268384B (en) * | 2013-06-01 | 2016-02-17 | 哈尔滨工业大学 | A kind of method of orderly extraction structure outline |
| CN106960209B (en) * | 2013-10-30 | 2020-08-14 | 浩德科技股份有限公司 | Image contour ordered point set extraction method |
-
2013
- 2013-10-30 CN CN201710172510.0A patent/CN106960209B/en active Active
- 2013-10-30 CN CN201310522543.5A patent/CN103530639B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101119429A (en) * | 2006-08-01 | 2008-02-06 | 北京北大方正电子有限公司 | Digital watermark embedded and extracting method and device |
| CN103353991A (en) * | 2013-06-24 | 2013-10-16 | 西安交通大学 | Method for generating contour fragments using iteration analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106960209B (en) | 2020-08-14 |
| CN103530639A (en) | 2014-01-22 |
| CN106960209A (en) | 2017-07-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103530639B (en) | A kind of orderly point set extracting method of image outline | |
| CN106358003A (en) | Video analysis and accelerating method based on thread level flow line | |
| RU2008107746A (en) | TRAINING OF CONVOLUTIONAL NEURAL NETWORKS ON GRAPHIC PROCESSORS | |
| CN107256406A (en) | Overlapping fibers image partition method, device, storage medium and computer equipment | |
| CN106548179A (en) | The detection method of object and dress ornament key point, device and electronic equipment | |
| CN102064977A (en) | Graphics processing unit (GPU) based method for detecting message content of high-speed network | |
| CN104732490A (en) | CUDA-based quick bilateral filtering method | |
| CN112991231B (en) | Single-image super-image and perception image enhancement joint task learning system | |
| CN110472539A (en) | A kind of Method for text detection, device and computer storage medium | |
| CN101727580A (en) | Image processing apparatus, electronic medium, and image processing method | |
| CN109740508A (en) | A kind of image processing method and nerve network system based on nerve network system | |
| CN107066979A (en) | A kind of human motion recognition method based on depth information and various dimensions convolutional neural networks | |
| CN109740529A (en) | A kind of virtual fit method drawn based on neural network | |
| CN109922346A (en) | A kind of convolutional neural networks for the reconstruct of compressed sensing picture signal | |
| CN107169977A (en) | Adaptive threshold color image edge detection method based on FPGA and Kirsch | |
| CN103942788B (en) | High-spectrum remote sensing feature extracting method and device | |
| CN106775579B (en) | Floating-point operation accelerator module based on configurable technology | |
| Shahzad et al. | Enhanced watershed image processing segmentation | |
| Lee et al. | Fast stereoscopic rendering on mobile ray tracing GPU for virtual reality applications | |
| Sánchez et al. | A parallel method for impulsive image noise removal on hybrid CPU/GPU systems | |
| CN112270704B (en) | Star point centroid extraction method based on data linked list structure | |
| CN113840169B (en) | Video processing method, device, computing equipment and storage medium | |
| CN107622037A (en) | The method and apparatus that a kind of Matrix Multiplication for improving graphics processing unit calculates performance | |
| CN110108200B (en) | Laser speckle phase unwrapping method based on improved branch cutting method | |
| CN209708124U (en) | The layout structure of dedicated IC chip calculating kernel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20170927 Address after: 410200, No. 4, No. 4100, B1 District, bay area international business logistics park, Bay Road International Business Logistics Park, Wangcheng District, Hunan, Changsha Applicant after: HUNAN QINGCHUANG SCIENCE AND TECHNOLOGY CO., LTD. Address before: 410600 Hunan province Changsha city Ningxiang County Shuang Fu Pu Zhen Su Xi Cun Xian Chong Li group Applicant before: Liu Wei |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20181024 Address after: 554300 No. 39 Chuang guest house, North Industrial Park, Dalong Economic Development Zone, Tongren, Guiyang, Guizhou Patentee after: Guizhou Cang Chuang Incubator Co., Ltd. Address before: 410200 No. 4100, B1 District, zone 4, ceramic trade zone, Changsha, Wangcheng, Hunan. Patentee before: HUNAN QINGCHUANG SCIENCE AND TECHNOLOGY CO., LTD. |
|
| TR01 | Transfer of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171031 Termination date: 20181030 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |