CN104228767B - A kind of automobile starting method based on palmprint authentication - Google Patents

Abstract

The present invention relates to automotive field, particularly relate to a kind of automobile starting method based on palmprint authentication.The present invention provides a kind of automobile starting method based on palmprint authentication, comprise the following steps 1, use palmmprint acquisition instrument that the palmmprint of staff is acquired, then the palmprint information being collected being passed to palmmprint process and identifier, palmmprint processes and pre-processes palmprint information with identifier；2, use the Eigenvector grader of neighborhood that palmprint information carries out certification of classifying.The invention has the beneficial effects as follows that employing popular nearest tagsort carries out palmprint information differentiating certification, car owner has only to input the palmprint information of oneself and just can realize automobile starting by certification, and other people can not start automobile, also improve extension inaccuracy and the computation complexity problem of nearest feature line.

Description

A kind of automobile starting method based on palmprint authentication

Technical field

The present invention relates to automotive field, particularly relate to a kind of automobile starting method based on palmprint authentication.

Background technology

Vehicle startup system is a subsystem the most crucial in the middle of current onboard system, at a lot of vapour Car manufacturer and R&D institution are of considerable interest.Vehicle startup system is mainly used in antitheft and convenient Drive.

Automobile starting technology currently mainly mainly has: by an automobile key by automobile starting or Automobile no-key based on RFID enters startup system.Vehicle startup system based on key, Chang Hui Because leaving behind key, or lose key and allow driver have a headache very much.Automobile based on RFID without Although key starts system, to have only to band mobile phone the most permissible, it is to avoid the multiple carry-on object of key handset, But mobile phone also be possible to omit or lose.

Grader occupies critically important position in PRS.Nearest neighbor classifier (NN) and Center grader (NM) is two graders that comparison is famous recently.As nearest neighbor classifier Extension, nearest feature line (NFL) grader is suggested.Although nearest feature line grader successfully changes Enter classification performance relative to nearest neighbor classifier, but NFL has still come with some shortcomings, such as, There is the coarse problem of extension and bigger computation complexity problem in it.

In order to solve computation complexity problem, some nearest feature line graders improved are suggested, example Such as nearest neighbor line classifier, nearest midpoint grader, nearest neighbor classifier based on center etc..For Improving extension inaccuracy problem, some graders strengthened are suggested, and eigencenter divides such as recently Class device, the nearest feature line grader of restriction, the shortest Eigenvector grader etc..

Summary of the invention

For defect or deficiency present in prior art, the technical problem to be solved is: Thering is provided a kind of automobile starting method based on palmprint authentication, this method can be in the situation without key (or other certification article) realize the startup of automobile, and making driver thoroughly break away from will be with very The worry of many article, also improves extension inaccuracy and the computation complexity problem of nearest feature line.

The technical scheme that the present invention takes, for providing a kind of automobile starting method based on palmprint authentication, is wrapped Include following steps

Step 1: use palmmprint acquisition instrument that the palmmprint of staff is acquired, then being collected Palmprint information passes to palmmprint and processes and identifier, and palmmprint processes and carries out pre-with identifier to palmprint information Process；

Step 2: use the Eigenvector grader of neighborhood that palmprint information carries out certification of classifying；Described The computational methods of the Eigenvector grader of neighborhood are as follows:

Assuming each class at least two samples, defining a recurrence characteristic point tolerance is:

$d_{NFLR}(y,y_i^c{y}_j^c)=||y-y_r^{ij,c}||---\left(1\right)$

In formula:WithRepresenting two master samples, y represents test sample,Represent and return feature Point, can be calculated as:

$y_r^{ij,c}={\mathrm{Aw}}_{ij}^c---\left(2\right)$

In formula: Being linear regression coeffficient, linear regression coeffficient can be by formula (3) solution square Difference obtains:

$w_{ij}^c={\left(A^{T}A\right)}^{-1}{A}^{T}y---\left(3\right)$

In formula: if A^TA is unusual, by A^TA+0.01I replaces A^TA, by formula (2) and formula (3) characteristic point can be returnedIt is expressed as:

$y_r^{ij,c}=A{\left(A^{T}A\right)}^{-1}{A}^{T}y=p_{ij}^{c}y---\left(4\right)$

Calculate the Eigenvector distance of neighborhood: first check two master samplesWithWhether in definition In hypersphere, then carrying out the radius calculation of the Eigenvector of neighborhood, radius is that test sample arrives all classes The mean value of all sample distances, it is expressed as:

$d_{avg}=\frac{{\Σ}_{c=1}^M{\Σ}_{i=1}^{N_c}||y-y_i^c||}{{\Σ}_{c=1}^M{N}_c}---\left(5\right)$

In formula: y represents test sample, also illustrate that the center of the hypersphere of definition；

The Eigenvector distance of neighborhood is expressed as:

$d_{NFLS}(y,\stackrel{\‾}{y_i^c{y}_j^c})=\left\{\begin{array}{cc}||y-y_r^{ij,c}||,& y\∈{\mathrm{\Θ}}_1^{II}\\ \frac{||y-y_r^{ij,c}||+\min (l_i^c,l_j^c)}{2},& y\∈{\mathrm{\Θ}}_2^{II}\\ \min (l_i^c,l_j^c),& y\∈{\mathrm{\Θ}}_3^{II}\end{array}\right.---\left(6\right)$

In formula: Refer toWithAll situations in hypersphere； Refer toWithOnly one of which situation in hypersphere；Refer toWithAll feelings outside hypersphere Condition.

As a further improvement on the present invention, two samples in described step 2WithAll super In ball, then the Eigenvector distance of neighborhoodEqual to returning defined in formula (1) Characteristic point is returned to measure.

As a further improvement on the present invention, two samples in described step 2WithOnly one Individual in hypersphere, then the Eigenvector distance of neighborhoodEqual to d₁And d₂Flat Average, wherein d₁Measure equal to the recurrence characteristic point defined in formula (1), d₂Refer to test sample Smaller that with the spacing of two master samples.

As a further improvement on the present invention, two samples in described step 2WithAll super Outside ball, then the Eigenvector distance of neighborhoodEqual to test sample and two standards Smaller that in the spacing of sample.

The invention has the beneficial effects as follows: 1. inside automobile, add an easy palmmprint acquisition instrument, Substantially without design original in affecting automobile, retain traditional attractive in appearance.2. present invention uses ratio Palmprint information is carried out differentiating certification by more popular nearest tagsort.3. because of everyone palmmprint The different volume in information room, so car owner has only to input the palmprint information of oneself and pass through certification Just can realize automobile starting, and other people can not start automobile.

Accompanying drawing explanation

Fig. 1 is that the Eigenvector of present invention automobile starting based on palmprint authentication method neighborhood measures two Master sample all situation schematic diagrames in hypersphere；

Fig. 2 is that the Eigenvector of present invention automobile starting based on palmprint authentication method neighborhood measures two Master sample only one of which situation schematic diagram in hypersphere；

Fig. 3 is that the Eigenvector of present invention automobile starting based on palmprint authentication method neighborhood measures two Master sample all situation schematic diagrames outside hypersphere.

Detailed description of the invention

The present invention is further described for explanation and detailed description of the invention below in conjunction with the accompanying drawings.

Present invention vehicle startup system based on palmprint authentication is by adding an easy palm inside automobile Line acquisition instrument, thus by being acquired the palmmprint of staff, this method is fairly simple will not give Driver makes troubles.After having gathered heart palmprint information, this new vehicle startup system can use The Eigenvector grader of the neighborhood proposed carries out palmprint information certification, to judge whether being that car owner exists Start automobile.

Driver uses an easy palmmprint acquisition instrument to be acquired the palmmprint of staff, then The palmprint information being collected passes to palmmprint and processes and identifier.Adopt when palmmprint processes to receive with identifier After the palmmprint signal that storage sheet transmits, it can carry out some pretreatment to palmmprint signal, followed by uses me The Eigenvector grader of neighborhood that proposes it is carried out certification of classifying.

As shown in Figure 1 to Figure 3, in order to improve extension inaccuracy and the computation complexity of nearest feature line Problem, it is proposed that the Eigenvector grader of neighborhood.First one new distance metric is suggested, It is recalled and returns characteristic point to measure.With nearest eigencenter grader, the nearest feature line grader of restriction, The shortest Eigenvector grader is similar to, and returns characteristic measure and assumes each class at least two samples.Return Characteristic point tolerance is returned to be defined as

$d_{NFLR}(y,y_i^c{y}_j^c)=||y-y_r^{ij,c}||---\left(1\right)$

In formula:WithRepresenting two master samples, y represents test sample,Represent and return feature Point, wherein returns characteristic pointCan by calculate by

$y_r^{ij,c}={\mathrm{Aw}}_{ij}^c---\left(2\right)$

Wherein Being linear regression coeffficient, it can be solved least square error by formula (3) and obtain,

$w_{ij}^c={\left(A^{T}A\right)}^{-1}{A}^{T}y---\left(3\right)$

If A^TA is unusual, A^TA will be by A^TA+0.01I replaces. by formula (2) and (3), I It is known that recurrence characteristic point

$y_r^{ij,c}=A{\left(A^{T}A\right)}^{-1}{A}^{T}y=p_{ij}^{c}y---\left(4\right)$

WhereinThe characteristic point returned can be obtained by the linear combination of two master samples , if it is by due to other characteristic point. we precalculate and storeSo return spy Levy amount of calculation a little and would is that ratio is relatively low.

The nearest feature line section of neighborhood the most directly calculates the distance between test sample and characteristic curve.Neighborhood Eigenvector first checks two master samplesWithWhether in the hypersphere of definition.As it is shown in figure 1, The center of hypersphere of definition is test sample y, radius be test sample arrive all samples of all classes away from From mean value, it can calculate by

$d_{avg}=\frac{{\Σ}_{c=1}^M{\Σ}_{i=1}^{N_c}||y-y_i^c||}{{\Σ}_{c=1}^M{N}_c}---\left(5\right)$

IfWithAll in hypersphere, as shown in Figure 1, so Eigenvector distance of neighborhood,Equal to formula (1) if defined in recurrence characteristic point tolerance.WithOnly Have one in hypersphere, as shown in Figure 2, so Eigenvector distance of neighborhood,Deng In d₁And d₂Mean value, wherein d₁Measure equal to the recurrence characteristic point defined in formula (1), d₂ If referring to smaller that in the spacing of test sample and two master samples.WithAll exist Outside hypersphere, as shown in Figure 3, so Eigenvector distance of neighborhood,Equal to test specimens This is smaller that with the spacing of two master samples. in summary, and the Eigenvector of neighborhood Distance can be expressed as

$d_{NFLS}(y,\stackrel{\‾}{y_i^c{y}_j^c})=\left\{\begin{array}{cc}||y-y_r^{ij,c}||,& y\∈{\mathrm{\Θ}}_1^{II}\\ \frac{||y-y_r^{ij,c}||+\min (l_i^c,l_j^c)}{2},& y\∈{\mathrm{\Θ}}_2^{II}\\ \min (l_i^c,l_j^c),& y\∈{\mathrm{\Θ}}_3^{II}\end{array}\right.---\left(6\right)$

Wherein: Refer toWithAll situations in hypersphere；Refer toWithOnly one of which situation in hypersphere；Refer toWithAll situations outside hypersphere.

Above content is that to combine concrete preferred embodiment made for the present invention the most specifically Bright, it is impossible to assert the present invention be embodied as be confined to these explanations.For technology belonging to the present invention For the those of ordinary skill in field, without departing from the inventive concept of the premise, it is also possible to if making Dry simple deduction or replace, all should be considered as belonging to protection scope of the present invention.

Claims (4)

1. an automobile starting method based on palmprint authentication, it is characterised in that: comprise the following steps

Step 1: use palmmprint acquisition instrument the palmmprint of staff to be acquired, then the palm being collected Line information passes to palmmprint and processes and identifier, and palmmprint processes and carries out pre-with identifier to palmprint information Process；

Step 2: use the Eigenvector grader of neighborhood that palmprint information carries out certification of classifying；Described neighbour The computational methods of the Eigenvector grader in territory are as follows:

Assuming each class at least two samples, defining a recurrence characteristic point tolerance is:

$d_{NFLR}(y,y_i^c{y}_j^c)=\left|\right|y-y_r^{ij,c}\left|\right|---\left(1\right)$

In formula:WithRepresenting two master samples, y represents test sample,Represent and return feature Point, can be calculated as:

$y_r^{ij,c}={\mathrm{Aw}}_{ij}^c---\left(2\right)$

In formula: Being linear regression coeffficient, linear regression coeffficient can be by formula (3) solution square Difference obtains:

$w_{ij}^c={\left(A^{T}A\right)}^{-1}{A}^{T}y---\left(3\right)$

In formula: if A^TA is unusual, by A^TA+0.01I replaces A^TA, by formula (2) and formula (3) characteristic point can be returnedIt is expressed as:

$y_r^{ij,c}=A{\left(A^{T}A\right)}^{-1}{A}^{T}y=p_{ij}^{c}y---\left(4\right)$

Calculate the Eigenvector distance of neighborhood: first check two master samplesWithWhether in definition In hypersphere, then carrying out the radius calculation of the Eigenvector of neighborhood, radius is that test sample arrives all classes The mean value of all sample distances, it is expressed as:

$d_{avg}=\frac{{\mathrm{\Σ}}_{c=1}^M{\mathrm{\Σ}}_{i=1}^{N_c}\left|\right|y-y_i^c\left|\right|}{{\mathrm{\Σ}}_{c=1}^M{N}_c}---\left(5\right)$

In formula: y represents test sample, also illustrate that the center of the hypersphere of definition；

The Eigenvector distance of neighborhood is expressed as:

$d_{NFLS}(y,\stackrel{\‾}{y_i^c{y}_j^c})=\left\{\begin{array}{cc}\left|\right|y-y_r^{ij,c}\left|\right|,& y\∈{\mathrm{\Θ}}_1^{II}\\ \frac{\left|\right|y-y_r^{ij,c}\left|\right|+\min (l_i^c,l_j^c)}{2},& y\∈{\mathrm{\Θ}}_2^{II}\\ \min (l_i^c,l_j^c),& y\∈{\mathrm{\Θ}}_3^{II}\end{array}\right.---\left(6\right)$

In formula: Refer toWithAll situations in hypersphere； Refer toWithOnly one of which situation in hypersphere；Refer toWithAll feelings outside hypersphere Condition.

Automobile starting method based on palmprint authentication the most according to claim 1, it is characterised in that: institute State two samples in step 2WithAll in hypersphere, then the Eigenvector distance of neighborhoodMeasure equal to the recurrence characteristic point defined in formula (1).

Automobile starting method based on palmprint authentication the most according to claim 1, it is characterised in that: institute State two samples in step 2WithOnly one of which is in hypersphere, then the Eigenvector of neighborhood DistanceEqual to d₁And d₂Mean value, wherein d₁Defined in formula (1) Recurrence characteristic point tolerance, d₂Refer to that test sample is smaller with the spacing of two master samples That.

Automobile starting method based on palmprint authentication the most according to claim 1, it is characterised in that: institute State two samples in step 2WithAll outside hypersphere, then the Eigenvector distance of neighborhoodSmaller that equal in the spacing of test sample and two master samples.