CN105447495A - Automatic tracking apparatus realizing rapid tracking - Google Patents
Automatic tracking apparatus realizing rapid tracking Download PDFInfo
- Publication number
- CN105447495A CN105447495A CN201610008846.9A CN201610008846A CN105447495A CN 105447495 A CN105447495 A CN 105447495A CN 201610008846 A CN201610008846 A CN 201610008846A CN 105447495 A CN105447495 A CN 105447495A
- Authority
- CN
- China
- Prior art keywords
- curvature
- point
- characteristic area
- contour
- noise
- 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.)
- Pending
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/34—Smoothing or thinning of the pattern; Morphological operations; Skeletonisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an automatic tracking apparatus realizing rapid tracking. The apparatus comprises a common automatic tracking device and an object identification device installed on the automatic tracking device, wherein the identification device comprises a modeling module, a segmentation module, a merging module and a filtering module. According to the invention, through additional arrangement of the object identification apparatus on the automatic tracking device, the tracking capability of the automatic tracking device can be effectively enhanced, the automatic tracking device identifies an object through an object contour, object contour noise can be effectively filtered in the identification process, the type of the object is correctly identified, and the object is further tracked.
Description
Technical field
The present invention relates to autotracker field, be specifically related to a kind of autotracker of quick tracking.
Background technology
Can provide the space orientation of moving target, attitude, structure behavior and performance from motion tracking, along with the progress of society and science and technology, autotracker is applied to increasing field.But current autotracker is serious by noise, cannot Quick Catch target, slow to target recognition speed, make tracking effect undesirable.
Objective contour identification is as the important means of target identification, owing to being subject to the impact of the factor such as noise, quantization error in practical application, objective contour inevitably produces distortion, and in order to accurate description contour feature, the filtering process of objective contour is very necessary.At present, scholars propose the filtering algorithm of many noisy profiles, but ubiquity calculated amount is huge, noise reduction is undesirable, the excessive filtering of easy generation causes the problems such as target distortion.
Summary of the invention
For the problems referred to above, the invention provides a kind of autotracker of quick tracking.
Object of the present invention realizes by the following technical solutions:
A kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Width x width is selected to be window function W (n) of D, to curvature k
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1compare, determine noisy contour curvature k ' according to comparative result
nt (), that is: as | k|
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, can effectively reduce the distortion phenomenon after merging; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area;
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, in order to improve the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
The present invention by installing Target Identification Unit additional on autotracker, effectively can strengthen the tracking power of autotracker, autotracker is by objective contour identification target, can effective filtering objective contour noise in identifying, thus correct identification is made to targeted species, and further tracking target.
Accompanying drawing explanation
The invention will be further described to utilize accompanying drawing, but the embodiment in accompanying drawing does not form any limitation of the invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, can also obtain other accompanying drawing according to the following drawings.
Fig. 1 is the structured flowchart of the autotracker of quick tracking of the present invention.
Embodiment
The invention will be further described with the following Examples.
Fig. 1 is structured flowchart of the present invention, and it comprises: MBM, segmentation module, merging module, filtration module.
Embodiment 1: a kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Owing to being subject to the impact of noise, noisy profile G
nthe curvature value k of (t) upper part unique point
nt () accurately can not represent profile information, in order to obtain curvature accurately, select width to be that { window function W (n) of 7,9}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1=0.24 compares, and determines noisy contour curvature k ' according to comparative result
n(t), that is: as | k1N (t)-k2N (t) | > T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1;
The distribution of the unique point that obtains and non-unique point after classification is also discontinuous, cannot carry out effective contour smoothing by selecting filter to it.In order to obtain good contour smoothing effect, be necessary to carry out merging treatment to point of the same type.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, in this embodiment, and S=17,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, the length that the place that curvature is large needs is less, and the length that the place that curvature is little needs more greatly, can effectively reduce the distortion phenomenon after merging like this; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area.
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, pay close attention to the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
In this embodiment, S=17, threshold value T
1=0.24, window function width D ∈ { 7,9}, to noise intensity I ∈, { the noisy image of 10dB, 20dB} has good smooth effect, and autotracker is by objective contour identification target, can effective filtering objective contour noise in identifying, improve 0.2s to the recognition speed of target, discrimination improves 10%, and tracing process is stablized.
Embodiment 2: a kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Owing to being subject to the impact of noise, noisy profile G
nthe curvature value k of (t) upper part unique point
nt () accurately can not represent profile information, in order to obtain curvature accurately, select width to be that { window function W (n) of 10,12}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1=0.24 compares, and determines noisy contour curvature k ' according to comparative result
nt (), that is: as | k
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1;
The distribution of the unique point that obtains and non-unique point after classification is also discontinuous, cannot carry out effective contour smoothing by selecting filter to it.In order to obtain good contour smoothing effect, be necessary to carry out merging treatment to point of the same type.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, in this embodiment S=19,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, the length that the place that curvature is large needs is less, and the length that the place that curvature is little needs more greatly, can effectively reduce the distortion phenomenon after merging like this; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area.
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, pay close attention to the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
In this embodiment, S=19, threshold value T
1=0.24, window function width D ∈ { 10,12}, to noise intensity I ∈, { the noisy image of 20dB, 30dB} has good smooth effect, and autotracker is by objective contour identification target, can effective filtering objective contour noise in identifying, improve 0.25s to the recognition speed of target, discrimination improves 9%, and tracing process is stablized.
Embodiment 3: a kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Owing to being subject to the impact of noise, noisy profile G
nthe curvature value k of (t) upper part unique point
nt () accurately can not represent profile information, in order to obtain curvature accurately, select width to be that { window function W (n) of 13,14}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1=0.26 compares, and determines noisy contour curvature k ' according to comparative result
nt (), that is: as | k
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1;
The distribution of the unique point that obtains and non-unique point after classification is also discontinuous, cannot carry out effective contour smoothing by selecting filter to it.In order to obtain good contour smoothing effect, be necessary to carry out merging treatment to point of the same type.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, in this embodiment S=21,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, the length that the place that curvature is large needs is less, and the length that the place that curvature is little needs more greatly, can effectively reduce the distortion phenomenon after merging like this; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area.
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, pay close attention to the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
In this embodiment, S=21, threshold value T
1=0.26, window function width D ∈ { 13,14}, to noise intensity I ∈, { the noisy image of 30dB, 40dB} has good smooth effect, calculated amount and detailed information retain situation all in zone of acceptability in and obtain and preferably balance, autotracker, can effective filtering objective contour noise in identifying by objective contour identification target, improves 0.3s to the recognition speed of target, discrimination improves 8%, and tracing process is stablized.
Embodiment 4: a kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Owing to being subject to the impact of noise, noisy profile G
nthe curvature value k of (t) upper part unique point
nt () accurately can not represent profile information, in order to obtain curvature accurately, select width to be that { window function W (n) of 15,17}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1=0.28 compares, and determines noisy contour curvature k ' according to comparative result
nt (), that is: as | k
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1;
The distribution of the unique point that obtains and non-unique point after classification is also discontinuous, cannot carry out effective contour smoothing by selecting filter to it.In order to obtain good contour smoothing effect, be necessary to carry out merging treatment to point of the same type.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, in this embodiment S=23,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, the length that the place that curvature is large needs is less, and the length that the place that curvature is little needs more greatly, can effectively reduce the distortion phenomenon after merging like this; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area.
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, pay close attention to the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
In this embodiment, S=23, threshold value T
1=0.28, { 15,17}, to noise intensity I ∈ { 40dB for window function width D ∈, although the noisy image of 50dB} adds part calculated amount, but have excellent smooth effect to the image in this interval, and detailed information retains situation better, autotracker is by objective contour identification target, can effective filtering objective contour noise in identifying, improve 0.3s to the recognition speed of target, discrimination improves 9%, and tracing process is stablized.
Embodiment 5: a kind of autotracker of quick tracking, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module;
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Owing to being subject to the impact of noise, noisy profile G
nthe curvature value k of (t) upper part unique point
nt () accurately can not represent profile information, in order to obtain curvature accurately, select width to be that { window function W (n) of 17,19}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1=0.26 compares, and determines noisy contour curvature k ' according to comparative result
nt (), that is: as | k
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t) otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k, when | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0 otherwise, fundamental function f (t)=1;
The distribution of the unique point that obtains and non-unique point after classification is also discontinuous, cannot carry out effective contour smoothing by selecting filter to it.In order to obtain good contour smoothing effect, be necessary to carry out merging treatment to point of the same type.
Merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, in this embodiment S=25,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, the length that the place that curvature is large needs is less, and the length that the place that curvature is little needs more greatly, can effectively reduce the distortion phenomenon after merging like this; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area.
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, pay close attention to the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
In this embodiment, S=25, threshold value T
1=0.26, window function width D ∈ { 17,19}, to noise intensity I ∈, { the noisy image of 50dB, 60dB} has preferably smooth effect, and detailed information reservation situation is better, autotracker, can effective filtering objective contour noise in identifying by objective contour identification target, improves 0.1s to the recognition speed of target, discrimination improves 10%, and tracing process is stablized.
Finally should be noted that; above embodiment is only in order to illustrate technical scheme of the present invention; but not limiting the scope of the invention; although done to explain to the present invention with reference to preferred embodiment; those of ordinary skill in the art is to be understood that; can modify to technical scheme of the present invention or equivalent replacement, and not depart from essence and the scope of technical solution of the present invention.
Data simulation
The beneficial effect of this autotracker is: for the diversity of noise type and the unicity of current denoising method, adopts a kind of novel repeatedly filter, and proposes new contour segmentation, merging means and filter function; Calculated amount is relative and uncomplicated, considers the factor of global characteristics and local feature and level and smooth effective except making an uproar simultaneously; Consider the otherness of profile between dissimilar region, between restraint speckle and reservation details, obtain good balance; Different according to the curvature of difference, development length is automatic adaptive change correspondingly, effectively reduces the distortion phenomenon after merging.
By emulation, this device and other device is adopted to compare under noise intensity N, to the discrimination of target as following table:
| Noise intensity | 0.03 | 0.06 | 0.09 | 0.12 |
| This device embodiment 1 | 94％ | 93％ | 92％ | 90％ |
| This device embodiment 2 | 93％ | 94％ | 92％ | 90％ |
| This device embodiment 3 | 92％ | 91％ | 93％ | 90％ |
| This device embodiment 4 | 91％ | 90％ | 91％ | 92％ |
| This device embodiment 5 | 90％ | 90％ | 92％ | 93％ |
| Other device | 84％ | 83％ | 81％ | 80％ |
Claims (2)
1. the autotracker followed the tracks of fast, comprise common autotracker and be arranged on the Target Identification Unit on autotracker, this autotracker has very strong tracking power, Target Identification Unit can identify target according to objective contour, it is characterized in that, comprise MBM, segmentation module, merging module and filtration module; Wherein,
MBM, for setting up the parametrization equation of objective contour: for given objective contour G (t), its arc length parameterized the Representation Equation is G (t)=(x (t), y (t)), wherein x (t) and y (t) represents the coordinate of point respectively, t represents the parameter of contour curve equation, and t ∈ [0,1];
The arc length parameterized the Representation Equation of noisy profile is: G
n(t)=G (t)+N
1(t)+N
2(t) G (t), wherein additive noise part N
1(t)=N
1(x
1(t), y
1(t)), multiplicative noise part N
2(t)=N
2(x
2(t), y
2(t));
Segmentation module, the segmentation for profile: objective contour G (t) and noisy profile G
nt the curvature corresponding to () is respectively k (t) and k
n(t); Select width to be window function W (n) of D, { 7,9}, to curvature k for D ∈
nt () carries out neighborhood averaging, obtain mean curvature k
1N(t), simultaneously to the curvature value sequence in window, selected intermediate value curvature k
2Nt (), by mean curvature k
1N(t) and intermediate value curvature k
2Nabsolute value and the selected threshold value T of (t) difference
1compare, determine noisy contour curvature k ' according to comparative result
n(t), T
1=0.2, that is:
When | k
1N(t)-k
2N(t) | >T
1time, k '
n(t)=k
1N(t)
Otherwise, k '
n(t)=k
2N(t);
Because point that curvature value is larger reflects the notable feature of target, usually according to k '
nt point all in profile are divided into unique point or non-unique point by (), setting variable weight T
k, by judging that objective contour feature is how many, adaptive decision T
k,
When | k '
n(t) | <T
k* max|k '
n(t) | time, fundamental function f (t)=0
Otherwise, fundamental function f (t)=1.
2. autotracker according to claim 1, be further characterized in that, merge module: for rejecting the pseudo-random numbers generation because noise produces, and union operation is carried out to the unique point and non-unique point that cannot form continuum, thus obtain effective characteristic area and non-characteristic area: a selected starting point O, profile starting point extends the adjacent point of merging to both sides, using this starting point type as this region preset kind, extend each S × μ to both sides
0in time, stops, and wherein S is default minimum length, if S=15,
for the real-time curvature correction factor at O point place,
represent the radius-of-curvature of O point,
represent the mean radius of curvature of the O point obtained by above-mentioned window function, real-time curvature correction factor μ
0different for the curvature according to difference, auto modification development length, can effectively reduce the distortion phenomenon after merging; Calculate number N+1 and the N-1 of dissimilarity in two side areas respectively, if the number of dissimilarity is less than the minimum number of the type dissimilarity of setting, then this region is identical with preset kind, otherwise, contrary with preset kind; Again with two halt O
+ 1with an O
-1restart to calculate as starting point, extend S × μ laterally
o+1or S × μ
o-1in time, stops, wherein μ
o+1and μ
o-1representative point O respectively
+ 1with an O
-1the real-time curvature correction factor at place, O
+ 1in two side areas, dissimilarity number is N
+ 2, O
-1in two side areas, dissimilarity number is N
-2, according to above-mentioned decision condition, determine each section of types of profiles successively, the part of curtailment S calculates dissimilarity number according to the ratio of itself and S, counts corresponding characteristic area; Adjacent region of the same type is merged, obtains continuous print characteristic area and non-characteristic area;
Filtration module: multiplicative noise is owing to being relevant with picture signal, change with the change of picture signal, adopt Wiener filtering to carry out first-level filtering and remove, now image information also includes remaining multiplicative noise, by F wave filter F (x, y)=q × exp (-(x
2+ y
2)/β
2carry out secondary filtering, wherein q is by the coefficient of function normalization, that is: ∫ ∫ q × exp (-(x
2+ y
2)/β
2) dxdy=1, β be image template parameter;
After multiplicative noise filtering, the arc length parameterized the Representation Equation of noisy objective contour is G
n(t) '=G (t)+N
1(t); Suppose that additive noise is white Gaussian noise: x
n(t) '=x (t)+g
1(t, σ
2), y
n(t) '=y (t)+g
2(t, σ
2), wherein x
n(t) ' and y
nt () ' represents respectively and to remove after multiplicative noise each point coordinate on noisy profile, g
1(t, σ
2) and g
2(t, σ
2) be average be respectively zero, variance is σ
2white Gaussian noise, for simulating the additive noise in noisy objective contour;
Adopt function
smoothing to noisy profile, called after K wave filter, through point classification and Region dividing, noisy profile G
nt () ' is expressed as the combination of dissimilar contour segmentation:
wherein
represent the contour segmentation comprising characteristic area,
represent the contour segmentation comprising non-characteristic area, choose the parameter of K wave filter according to contour feature distribution, consider global characteristics and local characteristic factor simultaneously, at characteristic area, in order to retain detailed information, order
at non-characteristic area, in order to improve the effect of restraint speckle, order
the wherein overall variance that obtains for priori estimation of σ ', σ
1for the priori estimation variance of selected characteristic area, σ
0for the priori estimation variance of selected non-characteristic area,
for the average real-time curvature correction factor of selected characteristic area,
for the average real-time curvature correction factor of selected non-characteristic area; In order to reach good smooth effect, choose the length of half as K wave filter 85% fiducial interval of every type region minimum length S, thus the K wave filter of length self-adaptation different parameters according to two class regions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610008846.9A CN105447495A (en) | 2016-01-07 | 2016-01-07 | Automatic tracking apparatus realizing rapid tracking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610008846.9A CN105447495A (en) | 2016-01-07 | 2016-01-07 | Automatic tracking apparatus realizing rapid tracking |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105447495A true CN105447495A (en) | 2016-03-30 |
Family
ID=55557655
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610008846.9A Pending CN105447495A (en) | 2016-01-07 | 2016-01-07 | Automatic tracking apparatus realizing rapid tracking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105447495A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106934402A (en) * | 2017-03-10 | 2017-07-07 | 周艳 | Indoor moving video tracking positions auxiliary shooting method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673345A (en) * | 2009-07-01 | 2010-03-17 | 北京交通大学 | Method for extracting target closed contour based on shape prior |
| CN103123723A (en) * | 2013-01-23 | 2013-05-29 | 中国人民解放军信息工程大学 | Flowage line extracting method based on Canny edge detection and active contour model |
| CN103903251A (en) * | 2012-12-30 | 2014-07-02 | 南京理工大学 | Night vision image salient contour extracting method based on non-classical receptive field composite modulation |
-
2016
- 2016-01-07 CN CN201610008846.9A patent/CN105447495A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673345A (en) * | 2009-07-01 | 2010-03-17 | 北京交通大学 | Method for extracting target closed contour based on shape prior |
| CN103903251A (en) * | 2012-12-30 | 2014-07-02 | 南京理工大学 | Night vision image salient contour extracting method based on non-classical receptive field composite modulation |
| CN103123723A (en) * | 2013-01-23 | 2013-05-29 | 中国人民解放军信息工程大学 | Flowage line extracting method based on Canny edge detection and active contour model |
Non-Patent Citations (1)
| Title |
|---|
| 史思琦: "《基于轮廓特征的目标识别研究》", 《中国博士学位论文全文数据库》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106934402A (en) * | 2017-03-10 | 2017-07-07 | 周艳 | Indoor moving video tracking positions auxiliary shooting method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105913396A (en) | Noise estimation-based image edge preservation mixed de-noising method | |
| CN105528590A (en) | Quickly-alarming alarm device | |
| CN105678262A (en) | Tunnel geological monitoring equipment capable of conducting autonomous operation | |
| CN105414774A (en) | Laser cutting device capable of achieving autonomous cutting | |
| CN102324099A (en) | Step edge detection method oriented to humanoid robot | |
| CN106600610B (en) | FCM image segmentation method and device | |
| KR20110080246A (en) | Apparatus and method for data classification based on support vector machine | |
| CN109636822B (en) | Improved Canny self-adaptive edge extraction method based on newly-constructed membership function | |
| CN103207997B (en) | Kernel density estimation-based license plate character segmentation method | |
| CN109087347B (en) | Image processing method and device | |
| CN105447495A (en) | Automatic tracking apparatus realizing rapid tracking | |
| JP4994955B2 (en) | Mobile object identification device and mobile object identification program | |
| CN105447485A (en) | Landscape modeling detection tool realizing rapid detection | |
| Das et al. | Histogram equalization techniques for contrast enhancement: a review | |
| CN105459901A (en) | Vehicle driving restriction warning device capable of autonomously operating | |
| CN116912115A (en) | Underwater image self-adaptive enhancement method, system, equipment and storage medium | |
| CN105488514A (en) | Welding device with autonomous welding function | |
| CN105678770A (en) | Wall crack detection device excellent in profile identification and filtering performances | |
| CN105678841A (en) | Rapidly modeling type three-dimensional map acquisition device | |
| CN106683107A (en) | SAR image edge detection method based on ROEWA improvement | |
| CN105654443A (en) | High-performance flight device | |
| CN114066779B (en) | Depth map filtering method and device, electronic equipment and storage medium | |
| CN105631460A (en) | Automobile auxiliary brake device capable of fast braking | |
| CN105512642A (en) | Packing device capable of packing rapidly | |
| CN105677986A (en) | Stamping die good in contour recognition filtering performance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160330 |