CN101996312A - Method and device for tracking targets - Google Patents

Abstract

The invention relates to a method and device for tracking targets. The method comprises the following steps: calculating a plurality of similarity curved surfaces related to a plurality of characteristics of the targets, wherein the similarity curved surfaces are curved surfaces which are formed by similarity between a group of candidate targets and target templates; calculating reliability of the characteristics based on parameters related to the similarity curved surfaces; synthesizing the plurality of similarity curved surfaces based on the reliability of each characteristic to obtain the synthesized similarity curved surfaces; and determining the position of the targets based on the synthesized similarity curved surfaces. The method and device for tracking the targets of the invention can effectively track the targets.

Description

The method and apparatus of tracking target

Technical field

The application relates to computer vision field, relates more specifically to the tracking of target.

Background technology

Target following is a major issue of computer vision field, and it is behavioural analysis, and the basis of various higher layer applications such as abnormality detection has great significance in video monitoring system.

The factor that influences the target following effect is a lot, the wherein variation of illumination, and the variation of dbjective state, and to block (comprise block between target and target is blocked by background) be most important three kinds of situations.For accurate robust ground tracking target in video, people introduce various features, such as color histogram, gradient orientation histogram (HOG), Haar wavelet character, edge feature, LBP (Local BinaryPattern, local binary pattern) feature or the like.Because between the feature relative merits are arranged respectively, Feature Fusion has just become important problem of target tracking domain.

Summary of the invention

Provide hereinafter about brief overview of the present invention, so that basic comprehension about some aspect of the present invention is provided.Should be appreciated that this general introduction is not about exhaustive general introduction of the present invention.It is not that intention is determined key of the present invention or pith, neither be intended to limit scope of the present invention.Its purpose only is to provide some notion with the form of simplifying, with this as the preorder in greater detail of argumentation after a while.

One object of the present invention is to provide a kind of method and apparatus of new tracking target.

According to an aspect of the present invention, a kind of method of tracking target comprises: calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, this similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes; Come the confidence level of calculated characteristics based on the parameter of relevant similarity curved surface; Confidence level based on each feature synthesizes a plurality of similarity curved surfaces to obtain synthetic similarity curved surface; And the position of determining target based on synthetic similarity curved surface.

According to a further aspect in the invention, a kind of device of tracking target comprises: similarity curved surface computing unit, be configured to calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, this similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes; The confidence level computing unit is configured to come based on the parameter of relevant similarity curved surface the confidence level of calculated characteristics; Similarity curved surface synthesis unit is configured to based on the confidence level of each feature a plurality of similarity curved surfaces be synthesized to obtain synthetic similarity curved surface; And the target location determining unit, be configured to determine the position of target based on synthesizing the similarity curved surface.

According to a further aspect of the invention, a kind of computer program also is provided, be used to realize following method: calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, this similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes; Come the confidence level of calculated characteristics based on the parameter of relevant similarity curved surface; Confidence level based on each feature synthesizes a plurality of similarity curved surfaces to obtain synthetic similarity curved surface; And the position of determining target based on synthetic similarity curved surface.

According to a further aspect of the invention, also provide a kind of computer program of computer-readable medium form at least, recorded the computer program code of the method that is used to realize above-mentioned tracking target on it.

The present invention is tracking target effectively.

Description of drawings

With reference to below in conjunction with the explanation of accompanying drawing, can understand above and other purpose of the present invention, characteristics and advantage more easily to embodiment of the present invention.Parts in the accompanying drawing are just in order to illustrate principle of the present invention.In the accompanying drawings, same or similar technical characterictic or parts will adopt identical or similar Reference numeral to represent.

Fig. 1 shows the schematic diagram according to the Feature Fusion of present embodiment;

Fig. 2 shows the process flow diagram according to the method for the tracking target of first embodiment;

Fig. 3 shows the synoptic diagram of the method for tracking target second embodiment of the invention;

Fig. 4 shows the synoptic diagram of the multi-peaks structure that the similarity curved surface of HOG feature presented;

Fig. 5 shows the synoptic diagram according to the method for the tracking target of the 3rd embodiment of the present invention;

Fig. 6 shows the synoptic diagram according to the method for the tracking target of the 4th embodiment of the present invention;

Fig. 7 shows the synoptic diagram according to the method for the tracking target of the 5th embodiment of the present invention;

Fig. 8 illustrates the structure of giving an example of computing equipment of the device of the tracking target that can be used to realize embodiments of the present invention;

Fig. 9 shows the device according to the tracking target of the 6th embodiment of the present invention.

Embodiment

Embodiments of the present invention are described with reference to the accompanying drawings.Element of describing in an accompanying drawing of the present invention or a kind of embodiment and feature can combine with element and the feature shown in one or more other accompanying drawing or the embodiment.Should be noted that for purpose clearly, omitted the parts that have nothing to do with the present invention, those of ordinary skills are known and the expression and the description of processing in accompanying drawing and the explanation.

The method of tracking target

First embodiment

Fig. 1 shows the schematic diagram according to the Feature Fusion of present embodiment.Shown in Fig. 1 is fusion to the two-way feature.Calculate the similarity curved surface of first feature and the similarity curved surface of second feature respectively for objects in video (being the pedestrian shown in Fig. 1), the similarity curved surface with two-way is weighted summation then, obtains finally synthetic similarity curved surface.Thereby obtain the position of target probability of occurrence maximum.To further explain above-mentioned principle below.

In order to utilize the complementarity of different features, can comprise at least one global characteristics and at least one local feature in a plurality of features.Global characteristics can be color histogram (Histogram of Color, HC), local feature can be HOG, Haar wavelet character, edge feature, local binary pattern feature (Local Binary Pattern, LBP).For example, the major advantage of HOG feature is insensitive to illumination variation, but its discriminate individuals does not have separating capacity for different pedestrians.And the color histogram feature generally can discriminate individuals, but often is subjected to the influence of illumination variation easily.

Fig. 2 shows the process flow diagram according to the method for the tracking target of first embodiment.

Suppose known target template and initial position thereof.This To Template and initial position thereof can be to mark by hand or obtain by a reliable object detector.For example, can use HOG (the Histogram of Oriented Gradients that obtains through training, gradient orientation histogram) feature and SVM (Support Vector Machine) sorter (being designated hereinafter simply as the HOG sorter) detects initial object appearing in the video as object detector, should detected target being defined as To Template, the initial position that occurs of target is defined as initial position.Because target generally occupies certain area in image, for convenience, also the center of target (also can be other specified points, for example peak etc.) be called the position of target below.With the example that is fused to of two-way feature present embodiment is introduced below, be it should be understood that present embodiment also can be applied to the situation that three or more features are merged.

In step 202, calculate relevant with first feature of target respectively similarity curved surface with second feature.

The similarity curved surface is meant the curved surface that similarity constituted between one group of candidate target and the To Template.

Candidate target can be determined by the following method.In the process of following the tracks of, for present frame, (for former frame is under the situation of initial frame based on the position of the target of determining in the former frame, as mentioned above, the position of target can obtain by manual mark or by a reliable object detector) and the maximum translational speed of the target of estimating, judge the zone that target may occur.Below zone that the target of judging may be occurred be called the region of search.The region of search has determined the coordinate range of similarity curved surface.The region of search can be with the target location that former frame is determined be the center of circle, to multiply by the distance that the mistiming between two frames determines with the maximum translational speed of the target of estimating be the circle of radius.The region of search can also be with the target location that former frame is determined be the center, to multiply by the distance that the mistiming between two frames determines with the target translational speed of estimating be the square of the length of side.The maximum translational speed of the target of above-mentioned estimation can be an empirical value, or multiply by a coefficient again by the value that the distance between the position of the determined target of front cross frame obtained divided by the mistiming between two frames.More simply, the region of search can also be that the target location of determining with former frame is the center of circle, is the circle of radius with the preset distance, or the target location of determining with former frame is the center, is the square of the length of side with the preset distance.Those skilled in the art also it is contemplated that the method for other definite region of search.The size of candidate target can be identical with To Template with shape.Perhaps, the size of candidate target can be carried out convergent-divergent with respect to To Template along with the variation of target of estimating and the distance between the camera lens.Can be with each point in the region of search all as the position of candidate target.In order to reduce calculated amount, also can in the region of search, select the position of some points at random as candidate target according to probabilistic model.Also can be earlier based on the position of the target of determining in the former frame, select at random according to probabilistic model that some points are as the position of candidate target in the image, a zone that will comprise these points then is defined as the region of search.The detail of more than determining candidate target and hunting zone is that those skilled in the art can realize, is not described in detail here.

After having determined candidate target, can the calculated candidate target and To Template between similarity.Similarity is relevant with clarification of objective, and therefore, similarity is actually the similarity that the characteristic model with candidate target and To Template compares.The characteristic model of To Template can obtain in advance.For example, can obtain the HC characteristic model based on the known target template for the HC feature.And for the HOG feature, the HOG characteristic model that can obtain by the method for sample training, and needn't the based target template obtain the HOG characteristic model.Especially, if adopt the HOG sorter to detect initial object appearing in the video as mentioned above, then can be with this HOG sorter as the HOG characteristic model.The concrete calculating of similarity is that those skilled in the art can realize, for example can adopt the Euclidean distance method, and Pasteur (Bhattacharyya) distance method or the like is not described in detail here.

After the similarity that calculates between each candidate target and the To Template, just can determine the similarity curved surface according to the value of the coordinate of the position of candidate target and corresponding similarity.Some points in region of search as mentioned above are not used as under the situation of position of candidate target, can utilize the method for interpolation to obtain the value of the corresponding similarity of these points.

In step 204, come the confidence level of calculated characteristics based on the average similarity of relevant similarity curved surface.

In embodiments of the present invention, determine the confidence level of the pairing feature of this similarity curved surface based on the similarity curved surface that obtains above.Give bigger weight for the big feature of confidence level when the Feature Fusion, thereby can determine the position of target more accurately.

Particularly, can from the similarity curved surface, extract the parameter of the confidence level that can reflect feature.

Extract the parameter of the average similarity of similarity curved surface in the present embodiment as the confidence level of reflection feature.Average similarity has reflected the average similarity degree between characteristic model and each candidate target, and on average similarity degree is high more, illustrates that this feature can describe the target of current tracking more accurately, and just this feature is credible more.That is to say that the confidence level of feature becomes positive correlation with average similarity.

In order to reduce The noise, can only average the similarity of similarity in the similarity curved surface greater than the point of predetermined threshold.Average similarity can be calculated as follows:

$\stackrel{\‾}{d}=\frac{1}{\left|R\right|}\underset{(x,y)\∈R}{\mathrm{\Σ}}D(x,y)$

Wherein,

Represent average similarity, R represents that similarity in the region of search surpasses the set of the point of threshold value, (x, the y) point among the expression R, D (x, y) the expression point (x, the similarity value of y) locating, | R| represents to gather the element number among the R.Above-mentioned threshold value can be scheduled to, and also can determine according to the parameter of similarity curved surface.For example this threshold value can be determined by the value at the top of similarity curved surface, particularly this threshold value can be the top value 1/2, or the value behind the value at the top decay 3dB.

Because the confidence level Conf of average similarity as feature can be directly adopted in average similarity and confidence level positive correlation, that is to say:

$\mathrm{Conf}=\stackrel{\‾}{d}$

Also the function of the average similarity confidence level Conf as feature for example can be able to be defined

$\mathrm{Conf}=e^{k\stackrel{\‾}{d}}$

Wherein k is the constant greater than 0, for example can be 3.

In step 206, a plurality of similarity curved surfaces are synthesized to obtain synthetic similarity curved surface based on the confidence level of each feature.

The confidence level of supposing first feature is Conf ₁, the confidence level of second feature is Conf ₂, can determine that then the weight of the pairing similarity curved surface of first feature is:

${\mathrm{<figure-callout\; id="1"\; label="w"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">w</figure-callout>}}_1=\frac{{\mathrm{<figure-callout\; id="1"\; label="Conf"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">Conf</figure-callout>}}_1}{{\mathrm{Conf}}_1+{\mathrm{Conf}}_2}$

The weight of the pairing similarity curved surface of second feature is:

$w_2=\frac{{\mathrm{Conf}}_2}{{\mathrm{Conf}}_1+{\mathrm{Conf}}_2}$

So, synthetic similarity curved surface D _cBe the similarity curved surface D of first feature ₁Similarity curved surface D with second feature ₂Weighted sum, promptly

D _c＝w ₁D ₁+w ₂D ₂

In step 208, determine the position of target based on synthetic similarity curved surface.For example, can be defined as the position of target in the present frame with synthesizing the pairing position of maximal value in the similarity curved surface.

By carrying out above step 202,204,206 and 208 continuously, can determine the position of target continuously, thereby realize the tracking of target.

Second embodiment

Fig. 3 shows the synoptic diagram of the method for tracking target second embodiment of the invention.Step 302 among Fig. 3,306 and 308 and Fig. 2 in step 202,206 and 208 similar, be not described in detail here.

The difference of second embodiment and first embodiment is the step of the confidence level of calculated characteristics.In the step 304 of Fig. 3, based on the confidence level of the multimodal degree calculated characteristics of relevant similarity curved surface.

Fig. 4 shows the synoptic diagram of the multi-peaks structure that the similarity curved surface of HOG feature presented.In the process of tracking target, do not having under the situation about disturbing, the similarity curved surface often presents unimodal structure.And under the situation that the existence disturbed is arranged, the similarity curved surface then presents multi-peaks structure, often presents two peak structure such as the HOG feature under the approaching situation in two people positions, and the color histogram feature is in the next multi-peaks structure that presents easily of situation of having powerful connections and disturbing.Than multi-peaks structure, we think that the distribution of unimodal structure is more credible.

Can be by the multimodal degree of following formula definition similarity curved surface:

$\mathrm{\&gamma;}=\frac{d_{\mathrm{peak}}^{-}}{d_{\max }}$

Wherein, γ represents the multimodal degree of similarity curved surface, Second peak in the expression similarity curved surface

Value, d _MaxThe value at the top in the expression similarity curved surface.

In the formula of above-mentioned multimodal degree, the value of γ increases along with the increase of the value on first peak, reduces along with the reducing of value on second peak, that is to say that γ is more little, illustrates that this feature confidence level is high more.That is to say feature confidence level and multimodal degree negative correlation.When the similarity curved surface presented unimodal structure, γ was zero.

Because γ is more little, this feature confidence level is high more, can define confidence level to be:

$\mathrm{Conf}=\frac{1}{\mathrm{\&gamma;}}$

In this definition, γ can not be zero.For fear of this situation, can consider another kind of definition:

Conf＝e ^-cγ

Wherein c is the constant greater than 0, for example can be 0.5.

The 3rd embodiment

Fig. 5 shows the synoptic diagram according to the method for the tracking target of the 3rd embodiment of the present invention.Step 502 among Fig. 5,506 and 508 and Fig. 2 in step 202,206 and 208 similar, be not described in detail here.

The difference of the 3rd embodiment and first embodiment is the step of the confidence level of calculated characteristics.In the step 504 of Fig. 5, based on the confidence level of the smooth degree calculated characteristics of relevant similarity curved surface.The reflection of the smooth degree of similarity curved surface the ability of feature differentiation target and background.Curved surface is smooth more, illustrates that the ability of this feature differentiation target and background is poor more, and promptly this feature confidence level is low more.And curved surface is precipitous more, and it is strong more to illustrate that this feature is distinguished the ability of target and background more, and just this feature confidence level is high more.That is to say the smooth degree negative correlation of feature confidence level and similarity curved surface.

Can be by the smooth degree of following formula definition similarity curved surface:

$S=\frac{1}{\left|P\right|}\underset{(x,y)\&Element;P}{\mathrm{\&Sigma;}}{I}_{D(x,y)>\mathrm{th}}$

Wherein, S represents smooth degree, and P represents the region of search, | P| represents the number of whole region of search mid point, (x, the y) point of expression among the P, D (x, y) the expression point (x, the similarity value of y) locating,

Th represents threshold value.This threshold value can be scheduled to, and also can determine according to the parameter of similarity curved surface.For example this threshold value can be determined by the value at the top of similarity curved surface, particularly this threshold value can be the top value 1/2, or the value behind the value at the top decay 3dB.

Because S is more little, this feature confidence level is high more, can define confidence level to be:

$\mathrm{Conf}=\frac{1}{S}$

The 4th embodiment

Fig. 6 shows the synoptic diagram according to the method for the tracking target of the 4th embodiment of the present invention.Step 602 among Fig. 6,606 and 608 and Fig. 2 in step 202,206 and 208 similar, be not described in detail here.

The difference of the 4th embodiment and first embodiment is the step of the confidence level of calculated characteristics.In the step 604 of Fig. 6, based on the confidence level of wantonly two or three calculated characteristics in average similarity, multimodal degree and the smooth degree of relevant similarity curved surface.

For example, can utilize a confidence level of coming calculated characteristics in the following formula:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}{e}^{k\stackrel{\&OverBar;}{d}}}{S}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}}{S}$

Perhaps:

$\mathrm{Conf}=\frac{e^{k\stackrel{\&OverBar;}{d}}}{S}$

Perhaps:

$\mathrm{Conf}=e^{-\mathrm{c\&gamma;}}{e}^{k\stackrel{\&OverBar;}{d}}$

The definition of the symbol in above-mentioned four formula is same as above, no longer is repeated in this description here.

The 5th embodiment

Fig. 7 shows the synoptic diagram according to the method for the tracking target of the 5th embodiment of the present invention.Step 702 among Fig. 7,706 and 708 and Fig. 2 in step 202,206 and 208 similar, be not described in detail here.

The difference of the 5th embodiment and first embodiment is the step of the confidence level of calculated characteristics.In the step 704 of Fig. 7, based on the parameter of relevant similarity curved surface with come the confidence level of calculated characteristics with reference to the parameter of similarity curved surface.

For example, can obtain with reference to the similarity curved surface based on initial target location.Particularly, can will be the center with initial target location, have the square of the predetermined length of side as the region of search.Perhaps, can will be the center of circle with initial target location, have the circle of predetermined radii as the region of search.Calculating is similar with reference to the process of all the other processes of similarity curved surface and the calculating similarity curved surface described in first embodiment, is not described in detail here.Calculating can be carried out after determining initial target location with reference to the similarity curved surface with reference to the process of the parameter of similarity curved surface, and can only follow the tracks of execution at first once in tracing process.

After having introduced, can revise the formula of the calculated characteristics confidence level in first embodiment to the, five embodiments with reference to the similarity curved surface.Particularly, in the 5th embodiment, can with any or a plurality of similarity curved surface parameter in each the feature confidence level computing formula in first to the 4th embodiment with combine with reference to similarity curved surface parameter value accordingly, thereby the feature confidence level computing formula that obtains revising.

For example, can utilize a confidence level of coming calculated characteristics in the following formula:

$\mathrm{Conf}=e^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}$

Perhaps:

$\mathrm{Conf}=e^{c(1-\min (1,\mathrm{\&gamma;}/{\mathrm{\&gamma;}}_0))}$

Perhaps:

$\mathrm{Conf}=\frac{1}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

Perhaps:

$\mathrm{Conf}={e^{-\mathrm{c\&gamma;}}e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}$

Perhaps:

$\mathrm{Conf}={e^{c(1-\min (1,\mathrm{\&gamma;}/{\mathrm{\&gamma;}}_0))}e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}}{S}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-c(1-\min (1,\mathrm{\&gamma;}/{\mathrm{\&gamma;}}_0))}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}}{S}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

Perhaps:

$\mathrm{Conf}=\frac{e^{c(1-\min (1,\mathrm{\&gamma;}/{\mathrm{\&gamma;}}_0))}}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

Perhaps:

$\mathrm{Conf}=\frac{e^{-c(1-\min (1,\mathrm{\&gamma;}/{\mathrm{\&gamma;}}_0))}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">d</figure-callout>}}_0}))}}{S/{\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="H2009101659424E0000012.png"\; state="\{\{state\}\}">S</figure-callout>}}_0}$

In the above-mentioned formula

S ₀, γ ₀Represent the average similarity with reference to the similarity curved surface, smooth degree and multimodal degree respectively,

Expression 1 He In less one, min (1, γ/γ ₀) expression 1 and γ/γ ₀In less one.The definition of other symbols is same as above, no longer is repeated in this description here.

In above formula, to utilize at the same time under the situation of average similarity and two parameters of smooth degree, these two parameters all relate to the selection of the threshold value of similarity.In order to reduce calculated amount, can select identical threshold value at these two parameters.Certainly, also can select different threshold values.

The device of tracking target

Fig. 8 illustrates the structure of giving an example of computing equipment of the device of the tracking target that can be used to realize embodiments of the present invention.

In Fig. 8, CPU (central processing unit) (CPU) 801 carries out various processing according to program stored among ROM (read-only memory) (ROM) 802 or from the program that storage area 808 is loaded into random-access memory (ram) 803.In RAM 803, also store data required when CPU 801 carries out various processing or the like as required.

CPU 801, ROM 802 and RAM 803 are connected to each other via bus 804.Input/output interface 805 also is connected to bus 804.

Following parts are connected to input/output interface 805: importation 806 comprises keyboard, mouse or the like; Output 807 comprises display, such as cathode ray tube (CRT) display, LCD (LCD) or the like and loudspeaker or the like; Storage area 808 comprises hard disk or the like; With communications portion 809, comprise that network interface unit is such as LAN card, modulator-demodular unit or the like.Communications portion 809 is handled such as the Internet executive communication via network.

As required, driver 810 also is connected to input/output interface 805.Detachable media 811 is installed on the driver 810 as required such as disk, CD, magneto-optic disk, semiconductor memory or the like, makes the computer program of therefrom reading be installed to as required in the storage area 808.

Can from network such as the Internet or storage medium such as detachable media 811 installation procedure to computing equipment.

It will be understood by those of skill in the art that this storage medium is not limited to shown in Figure 8 wherein having program stored therein, distribute separately so that the detachable media 811 of program to be provided to the user with equipment.The example of detachable media 811 comprises disk (comprising floppy disk (registered trademark)), CD (comprising compact disc read-only memory (CD-ROM) and digital universal disc (DVD)), magneto-optic disk (comprising mini-disk (MD) (registered trademark)) and semiconductor memory.Perhaps, storage medium can be hard disk that comprises in ROM 802, the storage area 808 or the like, computer program stored wherein, and be distributed to the user with the equipment that comprises them.

To the various embodiments of the device of tracking target of the present invention be described in detail below.Wherein, when the aspect that related in the description that relates in front the method for tracking target, will no longer be repeated in this description for brevity.

The 6th embodiment

Fig. 9 shows the device 900 according to the tracking target of the 6th embodiment of the present invention.Device 900 comprises similarity curved surface computing unit 902, confidence level computing unit 904, similarity curved surface synthesis unit 906 and target location determining unit 908.

Each unit in the device 900 can be respectively applied for each step of the method for the tracking target of carrying out first embodiment to the, five embodiments.Particularly, similarity curved surface computing unit 902 is configured to calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces.Confidence level computing unit 904 is configured to come based on the parameter of relevant similarity curved surface the confidence level of calculated characteristics.Similarity curved surface synthesis unit 906 is configured to will to state a plurality of similarity curved surfaces based on the confidence level of each feature and synthesizes to obtain synthetic similarity curved surface.Target location determining unit 908 is configured to determine based on synthetic similarity curved surface the position of target.

In one example, confidence level computing unit 904 is configured to based on the parameter of relevant similarity curved surface and comes the confidence level of calculated characteristics with reference to the parameter of similarity curved surface.At this moment, similarity curved surface computing unit 902 can be further configured to calculating with reference to the similarity curved surface.

In one example, confidence level computing unit 904 is configured to the one or more confidence levels of coming calculated characteristics in the smooth degree based on relevant similarity curved surface, average similarity and the multimodal degree.

Similarity curved surface and/or can be carried out by similarity curved surface computing unit 902 with reference to the calculating of the concrete parameter of similarity curved surface (for example smooth degree, average similarity and multimodal degree) also can be carried out by confidence level computing unit 904.

Further details about the operation of installing each unit of 900 can no longer repeat here with reference to the description of first embodiment to the, five embodiments.

Above some embodiments of the present invention are described in detail.To understand as those of ordinary skill in the art, whole or any steps or the parts of method and apparatus of the present invention, can be in the network of any computing equipment (comprising processor, storage medium etc.) or computing equipment, realized with hardware, firmware, software or their combination, this is that those of ordinary skills' their basic programming skill of utilization under the situation of understanding content of the present invention just can be realized, does not therefore need to specify at this.

In addition, it is evident that, when relating to possible peripheral operation in the superincumbent explanation, will use any display device and any input equipment, corresponding interface and the control program that link to each other with any computing equipment undoubtedly.Generally speaking, the hardware of the various operations in the related hardware in computing machine, computer system or the computer network, software and the realization preceding method of the present invention, firmware, software or their combination promptly constitute equipment of the present invention and each building block thereof.

Therefore, based on above-mentioned understanding, purpose of the present invention can also realize by program of operation or batch processing on any messaging device.Described messaging device can be known common apparatus.Therefore, purpose of the present invention also can be only by providing the program product that comprises the program code of realizing described method or equipment to realize.That is to say that such program product also constitutes the present invention, and storage or the medium that transmits such program product also constitute the present invention.Obviously, described storage or transmission medium can be well known by persons skilled in the art, and perhaps therefore the storage or the transmission medium of any kind that is developed in the future also there is no need at this various storages or transmission medium to be enumerated one by one.

The present invention also provides a kind of computer program, is used to realize following method: calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, this similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes; Come the confidence level of calculated characteristics based on the parameter of relevant similarity curved surface; Confidence level based on each feature synthesizes a plurality of similarity curved surfaces to obtain synthetic similarity curved surface; And the position of determining target based on synthetic similarity curved surface.

The present invention also provides a kind of computer program of computer-readable medium form at least, records the computer program code of the method that is used to realize above-mentioned tracking target on it.

In equipment of the present invention and method, obviously, after can decomposing, make up and/or decompose, each parts or each step reconfigure.These decomposition and/or reconfigure and to be considered as equivalents of the present invention.The step that also it is pointed out that the above-mentioned series of processes of execution can order following the instructions naturally be carried out in chronological order, but does not need necessarily to carry out according to time sequencing.Some step can walk abreast or carry out independently of one another.Simultaneously, during specific description of embodiments of the present invention in the above, can in one or more other embodiment, use in identical or similar mode at the feature that a kind of embodiment is described and/or illustrated, combined with the feature in other embodiment, or the feature in alternative other embodiment.

Should emphasize that term " comprises/comprise " existence that refers to feature, key element, step or assembly when this paper uses, but not get rid of the existence of one or more further feature, key element, step or assembly or additional.

Though described the present invention and advantage thereof in detail, be to be understood that and under not exceeding, can carry out various changes, alternative and conversion by the situation of the appended the spirit and scope of the present invention that claim limited.And the application's scope is not limited only to the embodiment of the described process of instructions, equipment, means, method and step.The one of ordinary skilled in the art will readily appreciate that from disclosure of the present invention, can use according to the present invention and carry out and process, equipment, means, method or step essentially identical function of corresponding embodiment described herein or acquisition result essentially identical with it, existing and that will be developed in the future.Therefore, appended claim is intended to comprise such process, equipment, means, method or step in their scope.

The present invention can be applicable to fields such as video monitoring, is used to monitor pedestrian, animal or other targets.The present invention can realize better Feature Fusion effect by assessing the confidence level of each road feature more accurately, thus tracking target more effectively.

Claims (15)

1. the method for a tracking target comprises:

Calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, described similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes;

Come the confidence level of calculated characteristics based on the parameter of relevant similarity curved surface;

Confidence level based on each described feature synthesizes described a plurality of similarity curved surfaces to obtain synthetic similarity curved surface; And

Determine the position of described target based on described synthetic similarity curved surface.

2. the method for claim 1, wherein the step of the confidence level of calculated characteristics comprises based on the parameter of relevant similarity curved surface and comes the confidence level of calculated characteristics with reference to the parameter of similarity curved surface.

3. the method for claim 1, wherein the confidence level of calculated characteristics comprises the one or more confidence levels of coming calculated characteristics in the smooth degree based on relevant similarity curved surface, average similarity and the multimodal degree.

4. method as claimed in claim 3, the smooth degree of wherein said similarity curved surface is determined greater than the number of the point of predetermined threshold based on similarity in the similarity curved surface.

5. method as claimed in claim 3, the average similarity of wherein said similarity curved surface be in the similarity curved surface similarity greater than the average similarity of the point of predetermined threshold.

6. method as claimed in claim 3, the multimodal degree of wherein said similarity curved surface are the ratios of value with the value at top on second peak in the similarity curved surface.

7. method as claimed in claim 3, one or more during the step of the confidence level of wherein said calculated characteristics comprises in the following manner determined confidence level: confidence level and described smooth degree negative correlation; Confidence level and described average similarity positive correlation; Confidence level and described multimodal degree negative correlation.

8. method as claimed in claim 3, the step of the confidence level of wherein said calculated characteristics comprises the confidence level of coming calculated characteristics by following formula:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{d_0}))}}{S/S_0}$

The confidence level of Conf representation feature wherein,

The average similarity of expression similarity curved surface, Expression is with reference to the average similarity of similarity curved surface, and S represents the smooth degree of similarity curved surface, S ₀Expression is with reference to the smooth degree of similarity curved surface, and γ represents the multimodal degree of similarity curved surface, and c, k are the constant more than or equal to 0,

Expression 1 He

In less one.

9. the method for claim 1, wherein said a plurality of features comprise at least one global characteristics and at least one local feature.

10. method as claimed in claim 9, wherein said global characteristics are color histogram, and described local feature is selected from: gradient orientation histogram, Haar wavelet character, edge feature, local binary pattern feature.

11. the device of a tracking target comprises:

Similarity curved surface computing unit is configured to calculate relevant with a plurality of features of target respectively a plurality of similarity curved surfaces, and described similarity curved surface is the curved surface that the similarity between one group of candidate target and the To Template constitutes;

The confidence level computing unit is configured to come based on the parameter of relevant similarity curved surface the confidence level of calculated characteristics;

Similarity curved surface synthesis unit is configured to based on the confidence level of each described feature described a plurality of similarity curved surfaces be synthesized to obtain synthetic similarity curved surface; And

The target location determining unit is configured to determine based on described synthetic similarity curved surface the position of described target.

12. device as claimed in claim 11, wherein said confidence level computing unit are configured to based on the parameter of relevant similarity curved surface and come the confidence level of calculated characteristics with reference to the parameter of similarity curved surface.

13. device as claimed in claim 11, wherein said confidence level computing unit are configured to based on the one or more confidence levels of coming calculated characteristics in the smooth degree of relevant similarity curved surface, average similarity and the multimodal degree.

14. device as claimed in claim 13, one or more during wherein said confidence level computing unit is configured in the following manner determined confidence level: confidence level and described smooth degree negative correlation; Confidence level and described average similarity positive correlation; Confidence level and described multimodal degree negative correlation.

15. device as claimed in claim 13, wherein said confidence level computing unit is configured to come by following formula the confidence level of calculated characteristics:

$\mathrm{Conf}=\frac{e^{-\mathrm{c\&gamma;}}{e}^{-k(1-\min (1,\stackrel{\&OverBar;}{d}/\stackrel{\&OverBar;}{d_0}))}}{S/S_0}$

The confidence level of Conf representation feature wherein,

The average similarity of expression similarity curved surface,

Expression is with reference to the average similarity of similarity curved surface, and S represents the smooth degree of similarity curved surface, S ₀Expression is with reference to the smooth degree of similarity curved surface, and γ represents the multimodal degree of similarity curved surface, and c, k are the constant more than or equal to 0,

Expression 1 He

In less one.

Images