CN105300390A - Method and device for determining moving trace of obstacle - Google Patents

Abstract

The invention discloses a method and a device for determining a moving trace of an obstacle, and belongs to the field of intelligent traffic. The method comprises the following steps: determining a transverse coordinate of the obstacle in each frame of image of continuous n frames of images acquired by a vision sensor, wherein n is an integer larger than 1; performing curve fitting on the transverse coordinates of the obstacle in the continuous n frames of images according to a preset curve fitting algorithm, thus obtaining the moving trace of the obstacle. According to the method and the device, the images acquired by the vision sensor arranged on a vehicle are subjected to analytic processing, so that the moving trace of the obstacle is determined; the problem that of high complexity during data processing in the relevant technology is solved, and the effects of reducing the detection cost of the obstacle and improving the data processing efficiency of the sensor are achieved. The method is used for determining the moving trace of the obstacle.

Description

The defining method of obstructing objects movement locus and device

Technical field

The present invention relates to intelligent transportation field, particularly a kind of defining method of obstructing objects movement locus and device.

Background technology

In order to detect the road environment of vehicle periphery, in vehicle, be generally provided with the multiple sensors such as vision sensor, millimetre-wave radar sensor and laser radar sensor.This multiple sensors can be monitored the orientation of the obstructing objects of vehicle periphery and translational speed, so that driver can judge more accurately to the transport condition of obstructing objects.

In correlation technique, vehicle needs to carry out fusion treatment to the data of the multiple sensors collections such as vision sensor, millimetre-wave radar sensor and laser radar sensor, namely the measurement data gathered the plurality of sensor carries out complementation and optimal combination to generate more reliable more accurate information, and then improves vehicle to the detectivity of surrounding road environment.

Although the accuracy of detection of millimetre-wave radar sensor and laser radar sensor is higher, the cost of these two kinds of sensors is higher; Further, when the kind of sensor that vehicle is arranged is more, vehicle needs to carry out fusion treatment to the data of multiple sensors, and complexity during data processing is higher, and the efficiency of data processing is lower.

Summary of the invention

In order to solve the problem of prior art, embodiments provide a kind of defining method and device of obstructing objects movement locus.Described technical scheme is as follows:

On the one hand, provide a kind of defining method of obstructing objects movement locus, described method comprises:

Determine the horizontal ordinate in each two field picture in the continuous n two field picture that obstructing objects obtains at vision sensor, described n be greater than 1 integer;

Wherein, the t two field picture I of described obstructing objects in described continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, described t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' for described in belong to same level straight line the horizontal ordinate of candidate point in described t two field picture and ordinate;

Based on the described candidate point set Z belonging to same level straight line, determine that described obstructing objects is at described t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \[\underset{y=-h/2}{\overset{h/2}{\Σ}}{W}_t(x,y)L_t(x,y)\],\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\∈Z\\ L_t(x,y)=0,(x,y)\∉Z\end{array};$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of described t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in described t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in described t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture belongs to, and L _t(x, y) value is 1, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture does not belong to, and L _t(x, y) value is 0;

Curve fitting algorithm according to presetting carries out curve fitting to the horizontal ordinate of described obstructing objects in described continuous n two field picture, obtains the movement locus of described obstructing objects.

Optionally, described according to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged in (x, y) }, comprising:

Calculate described t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\′}}_t(x,y)=\∂I_t(x,y)/\∂y;$

By described differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in described alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

Optionally, described determine horizontal ordinate x (t) of described obstructing objects in described t two field picture after, described method also comprises:

Obtain the horizontal velocity component of described obstructing objects in described t two field picture:

$v\left(t\right)=\frac{fT{x}_{}\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of described vision sensor, and Z (t) is described obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between described actual scene coordinate system and described vision sensor coordinate system and Z-direction, R _yfor the yaw rate of described vision sensor.

Optionally, after the horizontal velocity component of the described obstructing objects of described acquisition in described t two field picture, described method also comprises:

According to horizontal ordinate x (t) of described obstructing objects in described t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of described obstructing objects;

Under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

Under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Judge that described obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that described obstructing objects is background _t);

As the posterior probability P (C that described obstructing objects is target disorders object _t) be greater than the posterior probability P (B that described obstructing objects is background _t) time, described obstructing objects is defined as target disorders object.

Optionally, described described obstructing objects is defined as target disorders object after, described method also comprises:

Calculate the curvature of the movement locus of described target disorders object;

Determine the transport condition of described target disorders object according to described curvature, described transport condition comprises: turn left, turn right, keep straight on or turn around.

Optionally, the described transport condition judging described target disorders object according to described curvature, comprising:

When described curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of described target disorders object is for turning right;

When described curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of described target disorders object is for keeping straight on;

When described curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of described target disorders object is for turning left;

When described curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of described target disorders object is for turning around.

On the other hand, provide a kind of determining device of obstructing objects movement locus, described device comprises:

First determination module, for determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, described n be greater than 1 integer;

Wherein, the t two field picture I of described obstructing objects in described continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, described t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' for described in belong to same level straight line the horizontal ordinate of candidate point in described t two field picture and ordinate;

Based on the described candidate point set Z belonging to same level straight line, determine that described obstructing objects is at described t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \[\underset{y=-h/2}{\overset{h/2}{\Σ}}{W}_t(x,y)L_t(x,y)\],\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\∈Z\\ L_t(x,y)=0,(x,y)\∉Z\end{array};$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of described t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in described t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in described t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture belongs to, and L _t(x, y) value is 1, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture does not belong to, and L _t(x, y) value is 0;

Fitting module, for carrying out curve fitting to the horizontal ordinate of described obstructing objects in described continuous n two field picture according to the curve fitting algorithm preset, obtains the movement locus of described obstructing objects.

Optionally, described first determination module also for:

Calculate described t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\′}}_t(x,y)=\∂I_t(x,y)/\∂y;$

By described differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in described alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

Optionally, described device also comprises:

Acquisition module, for obtaining the horizontal velocity component of described obstructing objects in described t two field picture:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of described vision sensor, and Z (t) is described obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between described actual scene coordinate system and described vision sensor coordinate system and Z-direction, R _yfor the yaw rate of described vision sensor.

Optionally, described device also comprises:

Set up module, for according to horizontal ordinate x (t) of described obstructing objects in described t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of described obstructing objects;

Second determination module, under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

3rd determination module, under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Judge module, for judging that described obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that described obstructing objects is background _t);

4th determination module, for when described obstructing objects being the posterior probability P (C of target disorders object _t) be greater than the posterior probability P (B that described obstructing objects is background _t) time, described obstructing objects is defined as target disorders object.

Optionally, described device also comprises:

Computing module, for calculating the curvature of the movement locus of described target disorders object;

5th determination module, for determining the transport condition of described target disorders object according to described curvature, described transport condition comprises: turn left, turn right, keep straight on or turn around.

Optionally, described 5th determination module, also for:

When described curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of described target disorders object is for turning right;

When described curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of described target disorders object is for keeping straight on;

When described curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of described target disorders object is for turning left;

When described curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of described target disorders object is for turning around.

The beneficial effect that the technical scheme that the embodiment of the present invention provides is brought is:

The defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides and device, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, reduce the cost that vehicle detects obstructing objects, and vehicle only need process the data that vision sensor obtains, and improves the efficiency of data processing.

Accompanying drawing explanation

In order to be illustrated more clearly in the technical scheme in the embodiment of the present invention, below the accompanying drawing used required in describing embodiment is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1-1 is the application scenarios figure involved by defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides;

Fig. 1-2 is the process flow diagram of the defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides;

Fig. 2-1 is the process flow diagram of the defining method of the another kind of obstructing objects movement locus that the embodiment of the present invention provides;

Fig. 2-2 is a kind of method flow diagrams determining the horizontal ordinate of obstructing objects in t two field picture that the embodiment of the present invention provides;

Fig. 3-1 is the schematic diagram of the determining device of a kind of obstructing objects movement locus that the embodiment of the present invention provides;

Fig. 3-2 is schematic diagram of the determining device of the another kind of obstructing objects movement locus that the embodiment of the present invention provides.

Embodiment

For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.

Embodiments provide a kind of application scenarios figure involved by defining method of obstructing objects movement locus, see Fig. 1-1, this application scenarios can comprise this car 01 and obstacle vehicle 02, wherein this car 01 is provided with vision sensor, the image that this car 01 can be obtained by this vision sensor, the movement locus of this obstacle vehicle 02 is detected, and the image that can obtain according to vision sensor, determine the transport condition of this obstacle vehicle 02 further, to help the driver of this car to carry out anticipation more accurately to the running environment of vehicle.

Embodiments provide a kind of defining method of obstructing objects movement locus, see Fig. 1-2, the method comprises:

Step 101, determine the horizontal ordinate in each two field picture in the continuous n two field picture that obstructing objects obtains at vision sensor, this n be greater than 1 integer.

Wherein, the t two field picture I of this obstructing objects in this continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

(English: Harris) Corner Feature algorithm is to this t two field picture I according to Harris _t(x, y) carries out horizontal properties extraction, obtains this t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, this t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' are the horizontal ordinate of this candidate point belonging to same level straight line in this t two field picture and ordinate;

Based on the candidate point set Z that this belongs to same level straight line, determine that this obstructing objects is at this t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \[\underset{y=-h/2}{\overset{h/2}{\Σ}}{W}_t(x,y)L_t(x,y)\],\left\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\∈Z\\ L_t(x,y)=0,(x,y)\∉Z\end{array}\right.;$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of this t two field picture, and this highly refers to the length of the vertical direction of the image that vision sensor obtains.W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in this t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when the coordinate points (x, y) in this t two field picture belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture do not belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 0;

The curve fitting algorithm that step 102, basis are preset carries out curve fitting to the horizontal ordinate of this obstructing objects in this continuous n two field picture, obtains the movement locus of this obstructing objects.

In sum, the defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, reduce the cost that vehicle detects obstructing objects, and vehicle only need process the data that vision sensor obtains, and improves the efficiency of data processing.

Optionally, this according to Harris's Corner Feature algorithm to this t two field picture I _t(x, y) carries out horizontal properties extraction, obtains this t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged in (x, y) }, comprising:

Calculate this t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\′}}_t(x,y)=\∂I_t(x,y)/\∂y;$

By this differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in this alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

Optionally, after this determines horizontal ordinate x (t) of this obstructing objects in this t two field picture, the method also comprises:

Obtain the horizontal velocity component of this obstructing objects in this t two field picture:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of this vision sensor, and Z (t) is this obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between this actual scene coordinate system and this vision sensor coordinate system and Z-direction, R _yfor the yaw rate of this vision sensor.

Optionally, after the horizontal velocity component of this obstructing objects of this acquisition in this t two field picture, the method also comprises:

According to horizontal ordinate x (t) of this obstructing objects in this t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of this obstructing objects;

Under determining this observation model (x (t), v (t)), this obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

Under determining this observation model (x (t), v (t)), this obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Judge that this obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that this obstructing objects is background _t);

As the posterior probability P (C that this obstructing objects is target disorders object _t) be greater than the posterior probability P (B that this obstructing objects is background _t) time, this obstructing objects is defined as target disorders object.

Optionally, after this obstructing objects is defined as target disorders object by this, the method also comprises:

Calculate the curvature of the movement locus of this target disorders object;

Determine the transport condition of this target disorders object according to this curvature, this transport condition comprises: turn left, turn right, keep straight on or turn around.

Optionally, this judges the transport condition of this target disorders object according to this curvature, comprising:

When this curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of this target disorders object is for turning right;

When this curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of this target disorders object is for keeping straight on;

When this curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of this target disorders object is for turning left;

When this curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of this target disorders object is for turning around.

In sum, the defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, reduce the cost that vehicle detects obstructing objects, and vehicle only need process the data that vision sensor obtains, and improves the efficiency of data processing.

Fig. 2-1 is the defining method of the another kind of obstructing objects movement locus that the embodiment of the present invention provides, and as shown in Fig. 2-1, the method comprises:

Step 201, determine the horizontal ordinate in each two field picture in the continuous n two field picture that obstructing objects obtains at vision sensor, this n be greater than 1 integer.Perform step 202.

In embodiments of the present invention, vehicle only can be provided with vision sensor, each two field picture in the continuous n two field picture that vehicle can obtain this vision sensor processes, and determines the horizontal ordinate of obstructing objects in this each two field picture.

Fig. 2-2 is a kind of method flow diagrams determining the horizontal ordinate of obstructing objects in t two field picture that the embodiment of the present invention provides, and wherein, this t, for being more than or equal to 1, is less than or equal to the integer of n, and as shown in Fig. 2-2, the method comprises:

Step 2011, calculate t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y).

In embodiments of the present invention, due to vehicle front obstructing objects be other vehicles time, can affect greatly the transport condition of this car, and the afterbody of each car contains roof substantially, vehicle window, bumper etc. are horizontal linear feature comparatively significantly.Therefore vehicle can carry out horizontal properties extraction to the image that vision sensor obtains, and according to the horizontal properties that this extracts, determines this obstructing objects coordinate in the images.Wherein, horizontal properties is carried out to image and extracts and can realize with reference to Harris Corner Feature algorithm, in embodiments of the present invention, owing to only needing to extract the horizontal linear feature in image, therefore can directly to this t two field picture I _t(x, y) differentiates I' in the vertical direction _t(x, y):

Step 2012, the I ' of predetermined threshold value δ will be greater than in this differential _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points.

In embodiments of the present invention, in vehicle, threshold value δ can be preset with, for image I _t(x, y) differential I ' in the vertical direction _ythe I ' of this predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y), vehicle can confirm that this coordinate points is the coordinate points that contrast is higher, and this coordinate points is defined as alternative coordinate points.It should be noted that, in actual applications, this predetermined threshold value δ can comprise two, i.e. δ ₁and δ ₂, wherein δ ₂< δ ₁, when being greater than predetermined threshold value δ in image ₁coordinate points less time, can predetermined threshold value δ be passed through ₂determine alternative coordinate set further.

Step 2013, by this alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

In embodiments of the present invention, for same two field picture, its horizontal properties might not on same a line pixel, therefore can calculate the difference of this alternative coordinate points pixel in vertical direction, coordinate points pixel difference being less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line, and then obtain the candidate point set Z={ (x' belonging to same level straight line, y') }, x', y' are the horizontal ordinate of this candidate point belonging to same level straight line in this t two field picture and ordinate.Example, this presetted pixel point threshold value can be two or three pixels, and the coordinate points namely vertical direction differing two or three pixels can be defined as the candidate point belonging to same level straight line.

Step 2014, belong to the candidate point set of same level straight line based on this, determine horizontal ordinate x (t) of this obstructing objects in this t two field picture.

This obstructing objects is at this t two field picture I _thorizontal ordinate x (t) in (x, y) can be obtained by following formulae discovery:

$x\left(t\right)=\max \&lsqb;\underset{y=-h/2}{\overset{h/2}{\&Sigma;}}{W}_t(x,y)L_t(x,y)\&rsqb;,\left\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\&Element;Z\\ L_t(x,y)=0,(x,y)\&NotElement;Z\end{array}\right.;$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of this t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in this t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when the coordinate points (x, y) in this t two field picture belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture do not belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 0.Wherein, this central point set preset is preset according to the actual travel situation of vehicle, in the image that this central point set can get for vision sensor, near the set of the coordinate points of central area, by the weight matrix that this is preset, can filter in image some less marginal points of traveling state of vehicle impact, and then complexity during minimizing data processing.

The curve fitting algorithm that step 202, basis are preset carries out curve fitting to the horizontal ordinate of this obstructing objects in this continuous n two field picture, obtains the movement locus of this obstructing objects.Perform step 203.

In embodiments of the present invention, after vehicle gets the horizontal ordinate of obstructing objects in continuous n two field picture in each two field picture, according to the curve fitting algorithm preset, matching can be carried out to the horizontal ordinate of this obstructing objects in each two field picture, and then obtains the movement locus of this obstructing objects.Example, this curve fitting algorithm preset can be least square fitting algorithm, and the process carried out curve fitting by the coordinate of least square fitting algorithm to obstructing objects can with reference to correlation technique, and the embodiment of the present invention does not repeat at this.

Step 203, obtain the horizontal velocity component v (t) of this obstructing objects in this t two field picture.Perform step 204.

In embodiments of the present invention, for arbitrary obstructing objects P of vehicle front, suppose when vision sensor gets t two field picture, the coordinate of this obstructing objects P in actual scene coordinate system is (X (t), Y (t), Z (t)), then the coordinate of this obstructing objects P in the coordinate system of vision sensor (x (t), y (t)) can be expressed as:

$\left\{\begin{array}{c}x\left(t\right)=fX\left(t\right)/Z\left(t\right)\\ y\left(t\right)=fY\left(t\right)/Z\left(t\right)\end{array}\right.$

Wherein, f is camera focus, and the coordinate transform coefficient between this actual scene coordinate system and this vision sensor coordinate system is (T _x(t), T _y(t), T _z(t)).In embodiments of the present invention, the horizontal ordinate of the obstructing objects determined in the horizontal ordinate of this obstructing objects P in the coordinate system of vision sensor and step 202 can contrast by vehicle, when the horizontal ordinate in the coordinate system of this obstructing objects P at vision sensor is identical with the horizontal ordinate of the obstructing objects determined in step 202, then calculate the horizontal velocity component of this obstructing objects.Suppose the angle of pitch of vision sensor, crab angle, yaw velocity is (R _x(t), R _y(t), R _z(t)), then this obstructing objects can be expressed as relative to the relative velocity of vision sensor:

(V _x(t),V _y(t),V _z(t))＝(T _x(t),T _y(t),T _z(t))+(X(t),Y(t),Z(t))×(R _x(t),R _y(t),R _z(t))

Due on straight road, rate of pitch and the yaw velocity of vision sensor are approximately 0, i.e. R _x(t) ≈ 0, R _z(t) ≈ 0, then the horizontal velocity component v (t) of this obstructing objects in this t two field picture can be obtained by following formula:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, Z (t) is this obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between this actual scene coordinate system and this vision sensor coordinate system and Z-direction, R _yfor the yaw rate of this vision sensor.

Step 204, according to horizontal ordinate x (t) of this obstructing objects in this t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of this obstructing objects.Perform step 205.

In embodiments of the present invention, when after horizontal ordinate x (t) in t two field picture and horizontal velocity component v (t) of vehicle determination obstructing objects, the observation model (x (t), v (t)) of this obstructing objects can be set up.

Step 205, determine this observation model (x (t), v (t)) under, this obstructing objects is the posterior probability P (C of target disorders object _t).Perform step 206.

Wherein, obstructing objects is the posterior probability P (C of target disorders object _t) can be expressed as:

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]。

Wherein, P (C _t)+P (B _t)=1, obstructing objects is respectively from t-1 two field picture to the state transition probability of t two field picture:

Obstructing objects is background in t-1 two field picture, is the transition probability of target disorders object: P (C in t two field picture _t| B _t-1)=0.5, obstructing objects is background in t-1 two field picture, is the transition probability of background: P (B in t two field picture _t| B _t-1)=0.5, obstructing objects is target disorders object in t-1 two field picture, is the transition probability of target disorders object: P (C in t two field picture _t| C _t-1)=0.8, obstructing objects is target disorders object in t-1 two field picture, is the transition probability of background: P (B in t two field picture _t| C _t-1)=0.2, and under original state, to arrange this obstructing objects be the posterior probability of target disorders object is P (C ₀)=0.7, this obstructing objects is the posterior probability of background is P (B ₀)=0.3.

Step 206, determine this observation model (x (t), v (t)) under, this obstructing objects is the posterior probability P (B of background _t).Perform step 207.

Wherein, obstructing objects is the posterior probability P (B of background _t) can be expressed as:

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Step 207, judge that this obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that this obstructing objects is background _t).Perform step 208.

As the posterior probability P (C that this obstructing objects is target disorders object _t) be greater than the posterior probability P (B that this obstructing objects is background _t) time, perform step 208; As the posterior probability P (C that this obstructing objects is target disorders object _t) be not more than the posterior probability P (B that this obstructing objects is background _t) time, perform step 211.

Step 208, this obstructing objects is defined as target disorders object.Perform step 209.

As the posterior probability P (C that this obstructing objects is target disorders object _t) be greater than the posterior probability P (B that this obstructing objects is background _t) time, this obstructing objects can be defined as target disorders object by vehicle.Namely this obstructing objects may be moveable object such as other vehicle or pedestrians etc.

Step 209, calculate the curvature of the movement locus of this target disorders object.Perform step 210.

After vehicle determination target disorders object, according to the movement locus of the obstructing objects obtained in above-mentioned steps 202, the movement locus of this target disorders object can be obtained, and calculates the curvature of the movement locus of this target disorders object.

Step 210, determine the transport condition of this target disorders object according to this curvature.

This transport condition comprises: turn left, turn right, keep straight on or turn around.

In embodiments of the present invention, after vehicle calculates the curvature of target disorders object, can also this judge according to this curvature the transport condition of this target disorders object to carry out anticipation more accurately to help driver to the road environment of vehicle periphery.Example, vehicle is specifically as follows according to the process of the transport condition of the curvature determination target disorders object of movement locus:

When this curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of this target disorders object is for turning right; When this curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of this target disorders object is for keeping straight on; When this curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of this target disorders object is for turning left; When this curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of this target disorders object is for turning around.

Step 211, this obstructing objects is defined as background.

As the posterior probability P (C that this obstructing objects is target disorders object _t) be not more than the posterior probability P (B that this obstructing objects is background _t) time, this obstructing objects can be defined as background by vehicle, and now vehicle can ignore the impact of this obstructing objects on vehicle running state.

In sum, the defining method of a kind of obstructing objects movement locus that the embodiment of the present invention provides, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, and the obstructing objects can analyzing vehicle front is that target disorders object is still for background, not only reduce the cost that vehicle detects obstructing objects, and improve efficiency and the precision of data processing.

It should be noted that, the sequencing of the step of the defining method of the obstructing objects movement locus that the embodiment of the present invention provides can suitably adjust, and step also according to circumstances can carry out corresponding increase and decrease.Anyly be familiar with those skilled in the art in the technical scope that the present invention discloses, the method changed can be expected easily, all should be encompassed within protection scope of the present invention, therefore repeat no more.

Fig. 3-1 is the determining device of a kind of obstructing objects movement locus that the embodiment of the present invention provides, and as shown in figure 3-1, this device comprises:

First determination module 301, for determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, this n be greater than 1 integer.

Wherein, the t two field picture I of this obstructing objects in this continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to this t two field picture I _t(x, y) carries out horizontal properties extraction, obtains this t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, this t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' are the horizontal ordinate of this candidate point belonging to same level straight line in this t two field picture and ordinate;

Based on the candidate point set Z that this belongs to same level straight line, determine that this obstructing objects is at this t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \&lsqb;\underset{y=-h/2}{\overset{h/2}{\&Sigma;}}{W}_t(x,y)L_t(x,y)\&rsqb;,\left\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\&Element;(x,y)\&Element;Z\\ L_t(x,y)=0,(x,y)\&Element;(x,y)\&NotElement;Z\end{array}\right.;$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of this t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in this t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when the coordinate points (x, y) in this t two field picture belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture do not belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 0.

Fitting module 302, for carrying out curve fitting to the horizontal ordinate of this obstructing objects in this continuous n two field picture according to the curve fitting algorithm preset, obtains the movement locus of this obstructing objects.

In sum, the determining device of a kind of obstructing objects movement locus that the embodiment of the present invention provides, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, reduce the cost that vehicle detects obstructing objects, and vehicle only need process the data that vision sensor obtains, and improves the efficiency of data processing.

Fig. 3-2 is the determining devices of the another kind of obstructing objects movement locus that the embodiment of the present invention provides, and as shown in figure 3-2, this device comprises:

First determination module 301, for determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, this n be greater than 1 integer.

Wherein, the t two field picture I of this obstructing objects in this continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to this t two field picture I _t(x, y) carries out horizontal properties extraction, obtains this t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, this t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' are the horizontal ordinate of this candidate point belonging to same level straight line in this t two field picture and ordinate;

Based on the candidate point set Z that this belongs to same level straight line, determine that this obstructing objects is at this t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \&lsqb;\underset{y=-h/2}{\overset{h/2}{\&Sigma;}}{W}_t(x,y)L_t(x,y)\&rsqb;,\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\&Element;Z\\ L_t(x,y)=0,(x,y)\&NotElement;Z\end{array};$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of this t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in this t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when the coordinate points (x, y) in this t two field picture belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 1, when the coordinate points (x, y) in this t two field picture do not belong to this belong to the candidate point set Z of same level straight line time, L _t(x, y) value is 0.

Fitting module 302, for carrying out curve fitting to the horizontal ordinate of this obstructing objects in this continuous n two field picture according to the curve fitting algorithm preset, obtains the movement locus of this obstructing objects.

Acquisition module 303, for obtaining the horizontal velocity component of this obstructing objects in this t two field picture:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of this vision sensor, and Z (t) is this obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between this actual scene coordinate system and this vision sensor coordinate system and Z-direction, R _yfor the yaw rate of this vision sensor.

Set up module 304, for according to horizontal ordinate x (t) of this obstructing objects in this t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of this obstructing objects.

Second determination module 305, under determining this observation model (x (t), v (t)), this obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]。

3rd determination module 306, under determining this observation model (x (t), v (t)), this obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3.

Judge module 307, for judging that this obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that this obstructing objects is background _t).

4th determination module 308, for when this obstructing objects being the posterior probability P (C of target disorders object _t) be greater than the posterior probability P (B that this obstructing objects is background _t) time, this obstructing objects is defined as target disorders object.

Computing module 309, for calculating the curvature of the movement locus of this target disorders object.

5th determination module 310, for determining the transport condition of this target disorders object according to this curvature, this transport condition comprises: turn left, turn right, keep straight on or turn around.

Optionally, this first determination module 301, also for:

Calculate this t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\&prime;}}_t(x,y)=\&part;I_t(x,y)/\&part;y;$

By this differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in this alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

Optionally, the 5th determination module 310, also for:

When this curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of this target disorders object is for turning right;

When this curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of this target disorders object is for keeping straight on;

When this curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of this target disorders object is for turning left;

When this curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of this target disorders object is for turning around.

In sum, the determining device of a kind of obstructing objects movement locus that the embodiment of the present invention provides, by determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, and according to the curve fitting algorithm preset, the horizontal ordinate of this obstructing objects in this continuous n two field picture is carried out curve fitting, obtain the movement locus of this obstructing objects.The method can detect the movement locus of vehicle peripheral obstacle body by means of only the vision sensor that vehicle is arranged, reduce the cost that vehicle detects obstructing objects, and vehicle only need process the data that vision sensor obtains, and improves the efficiency of data processing.

Those skilled in the art can be well understood to, and for convenience and simplicity of description, the device of foregoing description and the specific works process of module, with reference to the corresponding process in preceding method embodiment, can not repeat them here.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (12)

1. a defining method for obstructing objects movement locus, is characterized in that, described method comprises:

Determine the horizontal ordinate in each two field picture in the continuous n two field picture that obstructing objects obtains at vision sensor, described n be greater than 1 integer;

Wherein, the t two field picture I of described obstructing objects in described continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, described t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' for described in belong to same level straight line the horizontal ordinate of candidate point in described t two field picture and ordinate;

Based on the described candidate point set Z belonging to same level straight line, determine that described obstructing objects is at described t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=max\&lsqb;\underset{y=-h/2}{\overset{h/2}{\&Sigma;}}{W}_t(x,y)L_t(x,y)\&rsqb;,\left\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\&Element;Z\\ L_t(x,y)=0,(x,y)\&NotElement;Z\end{array}\right.;$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of described t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in described t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in described t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture belongs to, and L _t(x, y) value is 1, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture does not belong to, and L _t(x, y) value is 0;

Curve fitting algorithm according to presetting carries out curve fitting to the horizontal ordinate of described obstructing objects in described continuous n two field picture, obtains the movement locus of described obstructing objects.

2. method according to claim 1, is characterized in that, described according to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged in (x, y) }, comprising:

Calculate described t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\&prime;}}_t(x,y)=\&part;I_t(x,y)/\&part;y;$

By described differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in described alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

3. method according to claim 1 and 2, is characterized in that, described determine horizontal ordinate x (t) of described obstructing objects in described t two field picture after, described method also comprises:

Obtain the horizontal velocity component of described obstructing objects in described t two field picture:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of described vision sensor, and Z (t) is described obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between described actual scene coordinate system and described vision sensor coordinate system and Z-direction, R _yfor the yaw rate of described vision sensor.

4. method according to claim 3, is characterized in that, after the horizontal velocity component of the described obstructing objects of described acquisition in described t two field picture, described method also comprises:

According to horizontal ordinate x (t) of described obstructing objects in described t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of described obstructing objects;

Under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

Under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Judge that described obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that described obstructing objects is background _t);

As the posterior probability P (C that described obstructing objects is target disorders object _t) be greater than the posterior probability P (B that described obstructing objects is background _t) time, described obstructing objects is defined as target disorders object.

5. method according to claim 4, is characterized in that, described described obstructing objects is defined as target disorders object after, described method also comprises:

Calculate the curvature of the movement locus of described target disorders object;

Determine the transport condition of described target disorders object according to described curvature, described transport condition comprises: turn left, turn right, keep straight on or turn around.

6. method according to claim 5, is characterized in that, the described transport condition judging described target disorders object according to described curvature, comprising:

When described curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of described target disorders object is for turning right;

When described curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of described target disorders object is for keeping straight on;

When described curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of described target disorders object is for turning left;

When described curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of described target disorders object is for turning around.

7. a determining device for obstructing objects movement locus, is characterized in that, described device comprises:

First determination module, for determining the horizontal ordinate in the continuous n two field picture that obstructing objects obtains at vision sensor in each two field picture, described n be greater than 1 integer;

Wherein, the t two field picture I of described obstructing objects in described continuous n two field picture is determined _tthe process of the horizontal ordinate in (x, y) comprises:

According to Harris's Corner Feature algorithm to described t two field picture I _t(x, y) carries out horizontal properties extraction, obtains described t two field picture I _tthe candidate point set Z={ (x', y') of same level straight line is belonged to } in (x, y), wherein, described t, for being more than or equal to 1, is less than or equal to the integer of n, x', y' for described in belong to same level straight line the horizontal ordinate of candidate point in described t two field picture and ordinate;

Based on the described candidate point set Z belonging to same level straight line, determine that described obstructing objects is at described t two field picture I _thorizontal ordinate x (t) in (x, y):

$x\left(t\right)=\max \&lsqb;\underset{y=-h/2}{\overset{h/2}{\&Sigma;}}{W}_t(x,y)L_t(x,y)\&rsqb;,\left\{\begin{array}{c}{L}_t(x,y)=1,(x,y)\&Element;Z\\ L_t(x,y)=0,(x,y)\&NotElement;Z\end{array}\right.;$

Wherein, ∈ represents and belongs to, represent and do not belong to, h is the height of described t two field picture, W _t(x, y) weight matrix for presetting, when the coordinate points (x, y) in described t two field picture belongs to default central point set, W _t(x, y) value is 1, when the coordinate points (x, y) in described t two field picture does not belong to default central point set, and W _t(x, y) value is 0, L _t(x, y) is horizontal segmentation coefficient, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture belongs to, and L _t(x, y) value is 1, when belonging to the candidate point set Z of same level straight line described in the coordinate points (x, y) in described t two field picture does not belong to, and L _t(x, y) value is 0;

Fitting module, for carrying out curve fitting to the horizontal ordinate of described obstructing objects in described continuous n two field picture according to the curve fitting algorithm preset, obtains the movement locus of described obstructing objects.

8. device according to claim 7, is characterized in that, described first determination module also for:

Calculate described t two field picture I _t(x, y) differential I' in the vertical direction _t(x, y):

${I^{\&prime;}}_t(x,y)=\&part;I_t(x,y)/\&part;y;$

By described differential I ' _ythe I ' of predetermined threshold value δ is greater than in (x, y) _ycoordinate points corresponding to (x, y) is defined as alternative coordinate points;

By in described alternative coordinate points, the coordinate points that the pixel difference in vertical direction is less than presetted pixel point threshold value is defined as belonging to the candidate point of same level straight line.

9. the device according to claim 7 or 8, is characterized in that, described device also comprises:

Acquisition module, for obtaining the horizontal velocity component of described obstructing objects in described t two field picture:

$v\left(t\right)=\frac{{\mathrm{fT}}_x\left(t\right)-x\left(t\right)T_z\left(t\right)}{Z\left(t\right)}-\frac{x^2\left(t\right)+f^2}{f}{R}_y\left(t\right);$

Wherein, f is the focal length of described vision sensor, and Z (t) is described obstructing objects coordinate along Z-direction in actual scene, T _xand T _zbe respectively the coordinate transform coefficient of X-direction between described actual scene coordinate system and described vision sensor coordinate system and Z-direction, R _yfor the yaw rate of described vision sensor.

10. device according to claim 9, is characterized in that, described device also comprises:

Set up module, for according to horizontal ordinate x (t) of described obstructing objects in described t two field picture and horizontal velocity component v (t), set up the observation model (x (t), v (t)) of described obstructing objects;

Second determination module, under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (C of target disorders object _t):

P(C _t)＝max[P(B _t-1)P(C _t|B _t-1)p(x(t),v(t)|C _t),P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|C _t)]；

3rd determination module, under determining described observation model (x (t), v (t)), described obstructing objects is the posterior probability P (B of background _t):

P(B _t)＝max[P(B _t-1)P(B _t|B _t-1)p(x(t),v(t)|B _t)P(C _t-1)P(B _t|C _t-1)p(x(t),v(t)|B _t)]；

Wherein, P (C _t)+P (B _t)=1, and P (C _t| B _t-1)=0.5, P (B _t| B _t-1)=0.5, P (C _t| C _t-1)=0.8, P (B _t| C _t-1)=0.2, under original state, P (C ₀)=0.7, P (B ₀)=0.3;

Judge module, for judging that described obstructing objects is the posterior probability P (C of target disorders object _t) whether be greater than the posterior probability P (B that described obstructing objects is background _t);

4th determination module, for when described obstructing objects being the posterior probability P (C of target disorders object _t) be greater than the posterior probability P (B that described obstructing objects is background _t) time, described obstructing objects is defined as target disorders object.

11. devices according to claim 10, is characterized in that, described device also comprises:

Computing module, for calculating the curvature of the movement locus of described target disorders object;

5th determination module, for determining the transport condition of described target disorders object according to described curvature, described transport condition comprises: turn left, turn right, keep straight on or turn around.

12. devices according to claim 11, is characterized in that, described 5th determination module, also for:

When described curvature is more than or equal to 10 degree, and when being less than 60 degree, determine that the transport condition of described target disorders object is for turning right;

When described curvature is more than or equal to 60 degree, and when being less than 120 degree, determine that the transport condition of described target disorders object is for keeping straight on;

When described curvature is more than or equal to 120 degree, and when being less than 150 degree, determine that the transport condition of described target disorders object is for turning left;

When described curvature is more than or equal to 150 degree, and when being less than 180 degree, determine that the transport condition of described target disorders object is for turning around.