CN106529535A - License plate positioning method in complex lighting environment based on wavelet transform - Google Patents

License plate positioning method in complex lighting environment based on wavelet transform Download PDF

Info

Publication number
CN106529535A
CN106529535A CN201610990778.0A CN201610990778A CN106529535A CN 106529535 A CN106529535 A CN 106529535A CN 201610990778 A CN201610990778 A CN 201610990778A CN 106529535 A CN106529535 A CN 106529535A
Authority
CN
China
Prior art keywords
image
license plate
wavelet transformation
carried out
wavelet
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201610990778.0A
Other languages
Chinese (zh)
Inventor
韦树艺
许焱
刘伟
娄刚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Fujitsu Nanda Software Technology Co Ltd
Original Assignee
Nanjing Fujitsu Nanda Software Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing Fujitsu Nanda Software Technology Co Ltd filed Critical Nanjing Fujitsu Nanda Software Technology Co Ltd
Priority to CN201610990778.0A priority Critical patent/CN106529535A/en
Publication of CN106529535A publication Critical patent/CN106529535A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/30—Noise filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/625—License plates

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a license plate positioning method in a complex lighting environment based on wavelet transform. The method comprises the following steps: (1) acquiring an image including a license plate, and graying the image to obtain a gray scale image; (2) de-noising the gray scale image by using wavelet transform to obtain a de-noised image; (3) searching a license plate area, where the license plate is located, in the de-noised image; and (4) cutting the license plate area and keeping the license number information of the license plate. The license plate positioning method in a complex lighting environment based on wavelet transform can remove the influence of complex illumination on the license plate image and improve the system positioning accuracy.

Description

It is a kind of to be based on wavelet transformation license plate locating method in complex illumination environment
Technical field
The present invention relates to a kind of be based on wavelet transformation license plate locating method in complex illumination environment.
Background technology
With expanding economy, motor vehicles become increasingly popular, and highway communication cause is developed rapidly, traditional labor management side Formula can not meet the needs of real work, and license plate recognition technology is gradually received as an important directions of intelligent transportation system To the attention of people.Car license recognition mainly includes positioning, three portions of the segmentation of characters on license plate and the identification of characters on license plate of car plate Point, the positioning of wherein car plate is the most important link in three parts, and the quality of License Plate decides that follow-up work whether can Normally carry out.
The method of existing License Plate has at present：(1) license plate locating method (2) based on Gray Level Jump is based on colored empty Between license plate locating method (3) based on morphologic license plate locating method etc..But mostly have under complicated illumination certain Limitation, the particularly License Plate under bloom or the low light level.Method 1 is the saltus step letter of the character rule using license plate area Breath as identification judge feature, the method ideally realize in license plate image it is simple, quick, but complexity illumination Under, the suitable reduction of the texture information at the edge and character of car plate, therefore 1 locating effect of method is undesirable, does not reach reality When traffic application system needed for requirement.The RGB color of image is transformed under the color space of HSV by method 2, then right The information of input image colors is detected there is fixed color according to the background of priori car plate, therefore can be by car plate face Color as known conditions by the pixel cluster for meeting designated color for scanning, so as to obtain the region of car plate.But consider here To the positioning of the situation in bloom, the method credibility based on color characteristic is really little.Method 3 proposes a kind of based on Mathematical Morphology License plate locating method, carries out top cap operation to image with the method for mathematical morphology, can remove car plate bloom part, but It is while the texture of car plate is often destroyed serious, to be unfavorable for carrying out work down.
To sum up method in one's power, does not fully take into account License Plate situation of the image under complex illumination.In real-time intelligent In traffic system, illumination is the key factor for affecting Car license recognition accuracy rate, particularly the License Plate in bloom according under, the above The algorithm for referring to not can solve.
Therefore, in order to reduce impact of the illumination to car plate, it is necessary to propose that one kind can effectively remove illumination to car plate Impact based on wavelet transformation in complex illumination environment license plate locating method.
The content of the invention
It is an object of the invention to provide it is a kind of effectively can remove impact of the illumination to car plate based on wavelet transformation The license plate locating method in complex illumination environment.
Technical scheme is as follows：It is a kind of to be based on wavelet transformation license plate locating method in complex illumination environment, bag Include following steps：First, the image comprising car plate is obtained, and gray processing process is carried out to described image, obtain gray level image；2nd, Wavelet denoising is utilized to the gray level image, the image after denoising is obtained；3rd, search in the image after the denoising The license plate area that Suo Suoshu car plates are located；4th, cutting is carried out to the license plate area, retains the license plate number information of the car plate.
Preferably, in step one, in gray processing processing procedure, gray processing formula is as follows：
Gray=0.5 × R+0.4 × G+0.1 × B,
Wherein, R, G, B represent three components of color diagram respectively.
Preferably, the step 2 specifically includes following steps：The gray level image to being input into carries out 2-d wavelet change Operation is changed, and the gray level image after the two-dimensional wavelet transformation operation described in is divided into into 4 sub-regions of LL, HL, LH and HH； Carry out, in the image after wavelet transform function, LL, HL and HH region being done zero-setting operation；The gray scale after to zero-setting operation Image does the image after 2-d wavelet inverse transformation obtains denoising.
Preferably, comprise the steps in the operation for carrying out two-dimensional wavelet transformation：Extract every a line of the gray level image It is set to hi, the height of 1≤hi≤H, H for image；One-dimensional wavelet transformation is carried out to hi, average signal Lhi and detail signal is obtained Hhi；Matrix L to Lhi compositions, the matrix H of Hhi compositions, extracts the row vdj of the row vaj and H of L, respectively wherein 1≤j≤W, W For the width of image；Wavelet transformation is carried out to vaj, it is Laj to obtain average information and detailed information is Haj；Small echo change is carried out to vdj Change, the average information for obtaining is Ldj, and detailed information is Hdj；All Laj, Haj, Ldj, Hdj are constituted matrix, is referred to as LL, HL, LH and HH.
Preferably, following steps are specifically included carrying out two-dimensional wavelet transformation Transform operations：After extracting zero-setting operation Every string of the gray level image is set to Wi, the width of 1≤Wi≤W, W for image；One-dimensional wavelet inverse transformation is carried out to Wi, is reduced The average signal and detail signal of column direction；To the image after above-mentioned process, extract and ki is set to per a line, 1≤ki≤H, H are figure The height of picture；One-dimensional wavelet inverse transformation is carried out to ki, the average signal and detail signal of line direction is reduced；Floating type image is turned 8bit images are turned to, algorithm is completed.
Preferably, the step 3 specifically includes following steps：For the image after the denoising of input, using bilateral filtering The characteristics of device, also protects marginal information not to be destroyed to its Filtering Processing, while smoothed image, then calculates gradient seal For Δ G；Binary conversion treatment is done to Δ G and obtains binary edge figure T, denoising is carried out to the binary edge figure T；Define one Cores of the structural element elem as closing operation of mathematical morphology, and closed operation operation is done to T so that texture-rich is distributed compact region A connected domain is fused into, wherein elem sizes are set as 16 × 1；The connected domain is screened, the two-value side is found out All of connected domain in edge figure T, filters out according to the analysis of car plate wide high proportion, area and region background color and does not meet car plate Region；According to the anglec of rotation, the image of the license plate area is corrected.
Preferably, the filtercondition of the connected domain is as follows：Connected domain area is set as S, the width and height of boundary rectangle Degree is respectively A and B, and external rotation rectangle is R and the anglec of rotation is d, then：If S<1000, then filter the ungratified connection of area Domain；If A.>Picture traverse × 0.3, filters the longer connected domain of width；If B<30, filter the less region of region height；If R The ratio of width to height be unsatisfactory for 3.14：1, filter pseudo- license plate area；If | d |>15 °, filter the larger region of gradient.
Preferably, in step 4, trimming operation is carried out to the license plate area and is comprised the steps：Using wavelet transformation The up-and-down boundary for removing the car plate and the right boundary that the car plate is removed using projection.
The beneficial effects of the present invention is：It is described that based on wavelet transformation, in complex illumination environment, license plate locating method is utilized Wavelet transformation has the advantages that low entropy, multiresolution and decorrelation, can carry out time domain, frequency domain point simultaneously to license plate image Analysis, can be very good to remove impact of the complex illumination to license plate image, and the effectively accuracy rate of lift system positioning.
And, the pretreatment of image is carried out according to the result figure after wavelet transformation, then which is asked for pretreatment output image Gradient image simultaneously extracts marginal information, combining form operation, and the Edge texture of license plate area is fused into the company of a rectangle Logical domain, obtains the approximate location of car plate through suitable screening rule, therefore, it is possible to effectively remove impact of the illumination to car plate, Anti-interference is preferable.
Description of the drawings
Fig. 1 is the flow process based on wavelet transformation license plate locating method in complex illumination environment provided in an embodiment of the present invention Schematic diagram；
Fig. 2 was decomposed based on wavelet transformation wavelet transformation in license plate locating method in complex illumination environment shown in Fig. 1 Journey schematic diagram；
Fig. 3 be after the image gray processing containing car plate using Wavelet denoising, obtain the image after denoising with it is former The contrast schematic diagram of figure.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, it is below in conjunction with drawings and Examples, right The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only to explain the present invention, and It is not used in the restriction present invention.
The description of specific distinct unless the context otherwise, the element and component in the present invention, quantity both can be with single shape Formula is present, it is also possible in the form of multiple, and the present invention is not defined to this.Although the step in the present invention is entered with label Arrangement is gone, but is not used to limit the precedence of step, unless expressly stated the order of step or holding for certain step Based on row needs other steps, the relative rank of otherwise step is adjustable.It is appreciated that used herein Term "and/or" is related to and covers one of associated Listed Items or one or more of any and all possible group Close.
Refer to Fig. 1, be it is provided in an embodiment of the present invention based on wavelet transformation in complex illumination environment License Plate side The schematic flow sheet of method.It is described that based on wavelet transformation, in complex illumination environment, license plate locating method specifically includes following steps：
First, the image comprising car plate is obtained, and gray processing process is carried out to described image, obtain gray level image.
As, in actual license plate system, the image that Jing photographic head is collected is typically all the color diagram based on RGB, but in car In order to avoid some unnecessary calculating in the flow process of board positioning, it usually needs color diagram is converted to single pass gray-scale maps. The formula of conversion is as follows:
Gray=α × R+ β × G+ γ × B
Wherein, R, G, B represent three components of color diagram respectively, and α, β, γ are weight, and alpha+beta+γ=1.It is modal Gray processing method is that value is respectively determining the value of α, β, γ according to human eye to the varying sensitivity of RGB：0.3、0.59、 0.11。
Various informative, the not first-class factor of character color, using the method pair of general gray processing in view of the car plate of China Car plate red glyphs sensitivity is relatively low, therefore the weights of tri- components of R, G, B need to be adjusted correspondingly.By experimental verification It is that effect is best when α, β, γ value is respectively 0.5,0.4,0.1, i.e., in the gray processing processing procedure of step one, after adjustment Gray processing formula it is as follows：
Gray=0.5 × R+0.4 × G+0.1 × B,
Wherein, R, G, B represent three components of color diagram respectively.
2nd, Wavelet denoising is utilized to the gray level image, the image after denoising is obtained.
Wavelet transformation is the sharp weapon of Digital Signal Processing, famous with " school microscop ".Wavelet transformation is in adding window Fu Develop on the basis of leaf transformation, absorb adding window Fourier transformation and can realize what the Time-Frequency Localization to signal was analyzed Function, while but also with the ability of self-adaptative adjustment window size, therefore, it is possible to good observation signal.Wavelet transformation why may be used For image procossing, it is because from from the viewpoint of mathematics, signal can be unified to regard signal processing as with image procossing, i.e., Image can be understood as a two-dimentional signal.For a single pass gray-scale maps, it is a binary function；For one Coloured image, it is a binary vector value function.As picture signal is discrete, therefore only focus in the present embodiment Application of the 2-d discrete wavelet in image procossing.
2-d discrete wavelet needs a two-dimentional scaling functionThe direction small echo ψ two-dimentional with threeH(x, y), ψV(x, y) and ψD(x, y).Each of which is unidimensional scale functionWith the product of the corresponding wavelet functions of ψ, i.e.,：
ψD(x, y)=ψ (x) ψ (y).
ψHIt is to measure the conversion along row, i.e. level of response edge, ψVThe change of tolerance row, i.e., corresponding vertical edge, ψD Corresponding to diagonally opposed change.Define a two-dimentional yardstick and translation basic function is as follows：
Then for image f (x, y) that size is M × N, its two-dimensional discrete wavelet conversion formula is：
And its 2-d discrete wavelet inverse transformation formula is：
And, the subimage of 4 a quarter sizes can be obtained after wavelet transformation, respectively WithCatabolic process is as shown in Figure 2.
Wherein,The information of the level detail, vertical detail and diagonal detail of image is represented respectively. In the image containing car plate, as the character of car plate is evenly distributed, details in vertical direction is abundant and image other make an uproar The interference that sound is caused is less, therefore the component in other directions after wavelet transformation is all filtered out, and only retains the little of vertical direction Wave conversion result.
At present, wavelet basiss have a lot such as haar small echos, symN wavelets, dbN small echos, Morlet small echos, Meyer small echos Etc..The effect for choosing different wavelet basiss process is not quite similar, and in several wavelet basiss, haar small echos have some protrusions Advantage, such as symmetry, high efficiency etc..Therefore in the present embodiment, selection uses haar wavelet basiss as the basis of conversion.
Specifically, the step 2 specifically includes following steps：
The gray level image to being input into carries out two-dimensional wavelet transformation operation, and by after the two-dimensional wavelet transformation operation described in Gray level image be divided into 4 sub-regions of LL, HL, LH and HH；
In image after wavelet transform function is carried out, LL, HL and HH region is carried out into zero-setting operation；
The gray level image after to zero-setting operation carries out the image after 2-d wavelet inverse transformation obtains denoising.
It should be noted that comprising the steps in the operation for carrying out two-dimensional wavelet transformation：
The every a line for extracting the gray level image is set to hi, the height of 1≤hi≤H, H for image；
One-dimensional wavelet transformation is carried out to hi, average signal Lhi and detail signal Hhi is obtained；
Matrix L to Lhi compositions, the matrix H of Hhi compositions, extracts the row vdj of the row vaj and H of L respectively, wherein 1≤j≤ The width of W, W for image；
Wavelet transformation is carried out to vaj, it is Laj to obtain average information and detailed information is Haj；Wavelet transformation is carried out to vdj, The average information for obtaining is Ldj, and detailed information is Hdj；
All Laj, Haj, Ldj, Hdj are constituted matrix, LL, HL, LH and HH is referred to as.
Wherein, LL is the average signal in average signal；HL is average signal in the horizontal direction, and in the vertical direction is Detail signal, highlights high-frequency characteristic straight up on the whole；LH and HL is just conversely, in the vertical direction is average letter Number, it is detail signal in the horizontal direction, highlights the high-frequency characteristic in horizontal direction on the whole；HH is the details in details Signal, what is highlighted are the high-frequency characteristics on diagonally opposed.
It is additionally, since horizontal, the vertical and diagonally opposed texture of characters on license plate relatively to enrich, generally, automobile figure The horizontal band of picture is more, and the interference of vertical direction is less.Accordingly, it would be desirable in Wavelet image after the conversion, LL, HL and HH regions filter, and only retain the band information of LH, i.e., only retain the frequency range of response vertical direction.To the gray scale on the basis of this Image carries out Transform operations.
Further, following steps are specifically included carrying out two-dimensional wavelet transformation Transform operations：
The every string for extracting the gray level image after zero-setting operation is set to Wi, the width of 1≤Wi≤W, W for image；
One-dimensional wavelet inverse transformation is carried out to Wi, the average signal and detail signal of column direction is reduced；
To the image after above-mentioned process, extract and ki is set to per a line, the height of 1≤ki≤H, H for image；
One-dimensional wavelet inverse transformation is carried out to ki, the average signal and detail signal of line direction is reduced；
Floating type image is converted into into 8bit images, algorithm is completed.
For example, as shown in figure 3, Fig. 3 is that small echo is carried out after the image gray processing containing car plate using at Noise Elimination from Wavelet Transform Reason, obtains the contrast schematic diagram of the image after denoising and artwork.
3rd, the license plate area at the car plate place is searched in the image after the denoising.
For the image after denoising is obtained after wavelet transformation in step 2, the Texture features edge of car plate is more clear, and And effectively filtered out the interference that horizontal texture may be brought.In order to be able to the region for obtaining car plate also needs further to be located Reason, the region for causing vertical edge abundant using gradient map and morphological transformation merge to form connected domain.With reference to China's car plate Some intrinsic features and priori, filter pseudo- license plate area so as to obtain real license plate area.
Specifically, the step 3 specifically includes following steps：
For the image after the denoising of input, using the characteristics of two-sided filter to its Filtering Processing, smoothed image it is same When also protect marginal information not to be destroyed, then calculate gradient map be designated as Δ G；
Binary conversion treatment is done to Δ G and obtains binary edge figure T, denoising is carried out to the binary edge figure T；
A structural element elem is defined as the core of closing operation of mathematical morphology, and closed operation operation is done to T so that texture is rich The compact region of rich distribution is fused into a connected domain, wherein setting elem sizes as 16 × 1；
The connected domain is screened, all of connected domain in the binary edge figure T is found out, according to car plate the ratio of width to height The analysis of example, area and region background color filters out the region for not meeting car plate；
According to the anglec of rotation, the image of the license plate area is corrected.
Specifically include it should be noted that binary conversion treatment being done to Δ G and obtaining the operation of binary edge figure T：Search image All of contour edge is designated as C, is denoted as Area (Ci) to the area of arbitrary Ci, if Area (Ci)<Ma or Area (Ci)>Ma is then The profile is filled using black picture element, is otherwise retained.Wherein ma and Ma represent the car plate minimum of permission, maximum area.
Further, the filtercondition of the connected domain is as follows：
Connected domain area is set as S, the width and height of boundary rectangle are respectively A and B, and external rotation rectangle is R and rotation Gyration is d, then：
If S<1000, then filter the ungratified connected domain of area；
If A.>Picture traverse × 0.3, filters the longer connected domain of width；
If B<30, filter the less region of region height；
If the ratio of width to height of R is unsatisfactory for 3.14：1, filter pseudo- license plate area；
If | d |>15 °, filter the larger region of gradient.
4th, cutting is carried out to the license plate area, retains the license plate number information of the car plate.
Through the result obtained by step 3, the approximate region of car plate is obtained, but still has remained the interference such as frame of car plate, It is unfavorable for the segmentation of follow-up car plate, it is therefore desirable to which cutting is carried out to the frame of car plate.In step 4, the license plate area is entered Row trimming operation comprises the steps：The up-and-down boundary of the car plate is removed and using the projection removal car using wavelet transformation The right boundary of board.
For the up-and-down boundary of the car plate is removed using wavelet transformation, as wavelet transformation can be with separation of level texture With vertical texture, therefore wavelet transformation is carried out to license plate area obtained above, and extract the skirt response of vertical direction, so as to The horizontal texture of car plate frame can be destroyed, so as to reach the effect of the upper and lower side frame for removing license plate area.Comprise the following steps that： The image after the wavelet transformation of car plate vertical direction is extracted, smoothing processing is done to image after the wavelet transformation and is filtered isolated making an uproar Point edge, then asks for the gradient map Δ T of image after the wavelet transformation, using a suitable structural element fusion Δ T's Marginal area, intercepted connected domain out are exactly the license plate image after removing upper and lower side frame.
For utilizing projection to remove the right boundary of the car plate, above-mentioned steps eliminate the upper following of license plate area Boundary, eliminates horizontal noise jamming.For the method that the right boundary of car plate is processed includes：Using the projection of edge graph come really Determine right boundary, as car plate left and right side frame is two elongated vertical lines, line width is less than characters on license plate, therefore can basis The width of projection histogram is determining right boundary.
Compared to prior art, the present invention provide based on wavelet transformation in complex illumination environment license plate locating method profit There is low entropy, multiresolution and decorrelation with wavelet transformation, time domain, frequency domain can be carried out to license plate image simultaneously Analysis, can be very good to remove impact of the complex illumination to license plate image.And the accuracy rate that effectively lift system is positioned.
And, the pretreatment of image is carried out according to the result figure after wavelet transformation, then which is asked for pretreatment output image Gradient image simultaneously extracts marginal information, combining form operation, and the Edge texture of license plate area is fused into the company of a rectangle Logical domain, obtains the approximate location of car plate through suitable screening rule, therefore, it is possible to effectively remove impact of the illumination to car plate, Anti-interference is preferable.
It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie In the case of spirit or essential attributes without departing substantially from the present invention, the present invention can be realized in other specific forms.Therefore, no matter From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the present invention is by appended power Profit is required rather than described above is limited, it is intended that all in the implication and scope of the equivalency of claim by falling Change is included in the present invention.Any reference in claim should not be considered as and limit involved claim.
Moreover, it will be appreciated that although this specification is been described by according to embodiment, not each embodiment is only wrapped Containing an independent technical scheme, this narrating mode of description is only that those skilled in the art should for clarity Using description as an entirety, the technical scheme in each embodiment can also Jing it is appropriately combined, form those skilled in the art Understandable other embodiment.

Claims (8)

  1. It is 1. a kind of to be based on wavelet transformation license plate locating method in complex illumination environment, it is characterised in that：Comprise the steps：
    First, the image comprising car plate is obtained, and gray processing process is carried out to described image, obtain gray level image；
    2nd, Wavelet denoising is utilized to the gray level image, the image after denoising is obtained；
    3rd, the license plate area at the car plate place is searched in the image after the denoising；
    4th, cutting is carried out to the license plate area, retains the license plate number information of the car plate.
  2. 2. it is according to claim 1 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： In step one, in gray processing processing procedure, gray processing formula is as follows：
    Gray=0.5 × R+0.4 × G+0.1 × B,
    Wherein, R, G, B represent three components of color diagram respectively.
  3. 3. it is according to claim 1 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： The step 2 specifically includes following steps：
    The gray level image to being input into carries out two-dimensional wavelet transformation operation, and the ash after the two-dimensional wavelet transformation described in is operated Degree image division is 4 sub-regions of LL, HL, LH and HH；
    In image after wavelet transform function is carried out, zero-setting operation is done in LL, HL and HH region；
    The gray level image after to zero-setting operation does the image after 2-d wavelet inverse transformation obtains denoising.
  4. 4. it is according to claim 3 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： Comprise the steps in the operation for carrying out two-dimensional wavelet transformation：
    The every a line for extracting the gray level image is set to hi, the height of 1≤hi≤H, H for image；
    One-dimensional wavelet transformation is carried out to hi, average signal Lhi and detail signal Hhi is obtained；
    Matrix L to Lhi compositions, the matrix H of Hhi compositions, extracts the row vdj of the row vaj and H of L, respectively wherein 1≤j≤W, W For the width of image；
    Wavelet transformation is carried out to vaj, it is Laj to obtain average information and detailed information is Haj；Wavelet transformation is carried out to vdj, is obtained Average information be Ldj, detailed information is Hdj；
    All Laj, Haj, Ldj, Hdj are constituted matrix, LL, HL, LH and HH is referred to as.
  5. 5. it is according to claim 3 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： Following steps are specifically included two-dimensional wavelet transformation Transform operations are carried out：
    The every string for extracting the gray level image after zero-setting operation is set to Wi, the width of 1≤Wi≤W, W for image；
    One-dimensional wavelet inverse transformation is carried out to Wi, the average signal and detail signal of column direction is reduced；
    To the image after above-mentioned process, extract and ki is set to per a line, the height of 1≤ki≤H, H for image；
    One-dimensional wavelet inverse transformation is carried out to ki, the average signal and detail signal of line direction is reduced；
    Floating type image is converted into into 8bit images, algorithm is completed.
  6. 6. it is according to claim 1 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： The step 3 specifically includes following steps：
    For input denoising after image, using the characteristics of two-sided filter to its Filtering Processing, while smoothed image Protect marginal information not to be destroyed, then calculate gradient map and be designated as Δ G；
    Binary conversion treatment is done to Δ G and obtains binary edge figure T, denoising is carried out to the binary edge figure T；
    A structural element elem is defined as the core of closing operation of mathematical morphology, and closed operation operation is done to T so that texture-rich point The compact region of cloth is fused into a connected domain；
    The connected domain is screened, all of connected domain in the binary edge figure T is found out, according to car plate wide high proportion, The analysis of area and region background color filters out the region for not meeting car plate；
    According to the anglec of rotation, the image of the license plate area is corrected.
  7. 7. it is according to claim 6 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： The filtercondition of the connected domain is as follows：
    Connected domain area is set as S, the width and height of boundary rectangle are respectively A and B, and external rotation rectangle is R and the anglec of rotation Spend for d, then：
    If S<1000, then filter the ungratified connected domain of area；
    If A.>Picture traverse × 0.3, filters the longer connected domain of width；
    If B<30, filter the less region of region height；
    If the ratio of width to height of R is unsatisfactory for 3.14：1, filter pseudo- license plate area；
    If | d |>15 °, filter the larger region of gradient.
  8. 8. it is according to claim 1 based on wavelet transformation in complex illumination environment license plate locating method, it is characterised in that： In step 4, trimming operation is carried out to the license plate area and is comprised the steps：The car plate is removed using wavelet transformation Up-and-down boundary and the right boundary using the projection removal car plate.
CN201610990778.0A 2016-11-10 2016-11-10 License plate positioning method in complex lighting environment based on wavelet transform Pending CN106529535A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610990778.0A CN106529535A (en) 2016-11-10 2016-11-10 License plate positioning method in complex lighting environment based on wavelet transform

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610990778.0A CN106529535A (en) 2016-11-10 2016-11-10 License plate positioning method in complex lighting environment based on wavelet transform

Publications (1)

Publication Number Publication Date
CN106529535A true CN106529535A (en) 2017-03-22

Family

ID=58350924

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610990778.0A Pending CN106529535A (en) 2016-11-10 2016-11-10 License plate positioning method in complex lighting environment based on wavelet transform

Country Status (1)

Country Link
CN (1) CN106529535A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107578042A (en) * 2017-05-08 2018-01-12 浙江工业大学 A kind of license plate locating method calculated based on R passage horizontal neighbors variance
CN107909071A (en) * 2017-12-19 2018-04-13 安徽理工大学 A kind of license plate locating method based on mathematical morphology
CN108197521A (en) * 2017-12-31 2018-06-22 南京理工大学 A kind of leggy Quick Response Code obtains identification device and method
CN108920420A (en) * 2018-03-23 2018-11-30 同济大学 A kind of Wavelet noise-eliminating method suitable for driving evaluation test data processing
CN110298339A (en) * 2019-06-27 2019-10-01 北京史河科技有限公司 A kind of instrument disk discrimination method, device and computer storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1932837A (en) * 2005-09-12 2007-03-21 电子科技大学 Vehicle plate extracting method based on small wave conversion and Redon transform
CN101064011A (en) * 2006-04-26 2007-10-31 电子科技大学 Vehicle registration extract method in complicated background based on wavelet transforming
CN101931792A (en) * 2010-08-04 2010-12-29 丁天 Multi-plate number locating method in high definition video steaming
CN104036262A (en) * 2014-06-30 2014-09-10 南京富士通南大软件技术有限公司 Method and system for screening and recognizing LPR license plate

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1932837A (en) * 2005-09-12 2007-03-21 电子科技大学 Vehicle plate extracting method based on small wave conversion and Redon transform
CN101064011A (en) * 2006-04-26 2007-10-31 电子科技大学 Vehicle registration extract method in complicated background based on wavelet transforming
CN101931792A (en) * 2010-08-04 2010-12-29 丁天 Multi-plate number locating method in high definition video steaming
CN104036262A (en) * 2014-06-30 2014-09-10 南京富士通南大软件技术有限公司 Method and system for screening and recognizing LPR license plate

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DONGXUEGO: "小波变换及其在图像处理中的典型应用：基于小波的去噪方法", 《豆丁文库：WWW.DOCIN.COM/P-465890427.HTML》 *
周月元 等: "基于小波变换的汽车牌照定位算法", 《科技资讯》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107578042A (en) * 2017-05-08 2018-01-12 浙江工业大学 A kind of license plate locating method calculated based on R passage horizontal neighbors variance
CN107909071A (en) * 2017-12-19 2018-04-13 安徽理工大学 A kind of license plate locating method based on mathematical morphology
CN108197521A (en) * 2017-12-31 2018-06-22 南京理工大学 A kind of leggy Quick Response Code obtains identification device and method
CN108920420A (en) * 2018-03-23 2018-11-30 同济大学 A kind of Wavelet noise-eliminating method suitable for driving evaluation test data processing
CN110298339A (en) * 2019-06-27 2019-10-01 北京史河科技有限公司 A kind of instrument disk discrimination method, device and computer storage medium

Similar Documents

Publication Publication Date Title
CN106529535A (en) License plate positioning method in complex lighting environment based on wavelet transform
CN106651872B (en) Pavement crack identification method and system based on Prewitt operator
CN106056155B (en) Superpixel segmentation method based on boundary information fusion
CN104299008B (en) Vehicle type classification method based on multi-feature fusion
WO2018145470A1 (en) Image detection method and device
CN104200437B (en) A kind of image defogging processing method
CN103942797B (en) Scene image text detection method and system based on histogram and super-pixels
CN104809461B (en) The licence plate recognition method and system of binding sequence image super-resolution rebuilding
CN101551853A (en) Human ear detection method under complex static color background
CN101615292B (en) Accurate positioning method for human eye on the basis of gray gradation information
CN107066972B (en) Natural scene Method for text detection based on multichannel extremal region
CN103927741A (en) SAR image synthesis method for enhancing target characteristics
CN103186904A (en) Method and device for extracting picture contours
CN104299009A (en) Plate number character recognition method based on multi-feature fusion
CN105893960A (en) Road traffic sign detecting method based on phase symmetry
CN111223110B (en) Microscopic image enhancement method and device and computer equipment
CN111832461A (en) Non-motor vehicle riding personnel helmet wearing detection method based on video stream
CN106157266A (en) A kind of orchard fruit image acquiring method
CN113537211B (en) Asymmetric IOU-based deep learning license plate frame positioning method
CN107784639A (en) A kind of polygon filtering and noise reduction method of unmanned aerial vehicle remote sensing image improvement
CN103279755A (en) Vehicle bottom shadow characteristics-based rapid license plate positioning method
CN110097617A (en) Image interfusion method based on convolutional neural networks Yu conspicuousness weight
CN104217430B (en) Image significance detection method based on L1 regularization
CN106599891A (en) Remote sensing image region-of-interest rapid extraction method based on scale phase spectrum saliency
CN102496144A (en) NSCT (nonsubsampled contourlet transform) sonar image enhancement method based on HSV (hue, saturation and value) color space

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20170322

RJ01 Rejection of invention patent application after publication