CN104463105A - Guide board recognizing method and device - Google Patents
Guide board recognizing method and device Download PDFInfo
- Publication number
- CN104463105A CN104463105A CN201410663502.2A CN201410663502A CN104463105A CN 104463105 A CN104463105 A CN 104463105A CN 201410663502 A CN201410663502 A CN 201410663502A CN 104463105 A CN104463105 A CN 104463105A
- Authority
- CN
- China
- Prior art keywords
- guideboard
- image
- color
- pixel
- connected region
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
Abstract
The invention discloses a guide board recognizing method and device and belongs to the technical field of computers. The guide board recognizing method comprises the steps that color detection is conducted on each pixel point in a street scene image to be processed, so that the pixel points corresponding to the color of a guide board are obtained; the pixel points corresponding to the color of the guide board are connected, so that at least one first connection area is obtained; area filtration is conducted on the at least one first connection area according to the image characteristics of the guide board, so that at least one second connection area is obtained; the image characteristic of each second connection area is extracted; the image characteristics of the second connection areas are filtered by means of a guide board reorganization support vector machine, and a guide board area is screened out from the at least one second connection area, wherein the guide board reorganization support vector machine is obtained through guide board image training; character recognition is conducted on the guide board area, so that guide board information of the guide board area is obtained. According to the guide board recognizing method and device, due to the fact that manual intervention is not needed in the guide board reorganization process, guide board reorganization is reduced, and guide board reorganization efficiency is improved.
Description
Technical field
The present invention relates to field of computer technology, particularly a kind of guideboard recognition methods and device.
Background technology
Grasp the periphery situation of POI (Point of Interest, information point) for the ease of user, occur increasing streetscape map.Wherein, in the street view image of streetscape map, a lot of guideboard may be comprised, in order to represent that this streetscape belongs to the information such as which Huo Natiao street, bar road.By identifying the guideboard in street view image, the information that can comprise streetscape map is supplemented, mark and calibration etc., thus is convenient to user and grasps geographic position belonging to streetscape.
Correlation technique, when identifying the guideboard in street view image, usually needs by manually realizing, and namely by the guideboard in artificial cognition street view image, and extracts the guideboard information that guideboard comprises.
Realizing in process of the present invention, inventor finds that correlation technique at least exists following problem:
Because the data scale of street view image is very large, therefore, need to consume a large amount of manpower by manual type identification guideboard, cost is higher.In addition, artificial cognition guideboard is often more consuming time, thus makes guideboard recognition efficiency not high.
Summary of the invention
In order to solve the problem of correlation technique, embodiments provide a kind of guideboard recognition methods and device.Described technical scheme is as follows:
First aspect, provide a kind of guideboard recognition methods, described method comprises:
Color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color;
Be communicated with the described pixel meeting guideboard color, obtain at least one first connected region;
According to guideboard characteristics of image, area filter is carried out at least one first connected region described, obtains at least one second connected region;
Extract the characteristics of image of each second connected region;
Use guideboard identification SVM (Support Vector Machine, support vector machine) characteristics of image of each second connected region is filtered, from at least one second connected region described, filter out guideboard region, described guideboard identification support vector machine obtains according to the training of guideboard image;
Text region is carried out to described guideboard region, obtains the guideboard information in described guideboard region.
Second aspect, provide a kind of guideboard recognition device, described device comprises:
Detection module, for carrying out color detection to each pixel in pending street view image, is met the pixel of guideboard color;
Connectivity module, for being communicated with the described pixel meeting guideboard color, obtains at least one first connected region;
First filtering module, for according to guideboard characteristics of image, carries out area filter at least one first connected region described, obtains at least one second connected region;
First extraction module, for extracting the characteristics of image of each second connected region;
Second filtering module, filter for using the characteristics of image of guideboard identification support vector machine to each second connected region, from at least one second connected region described, filter out guideboard region, described guideboard identification support vector machine obtains according to the training of guideboard image;
Identification module, for carrying out Text region to described guideboard region, obtains the guideboard information in described guideboard region.
The beneficial effect that the technical scheme that the embodiment of the present invention provides is brought is:
By carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then completes the identifying of guideboard by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
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 is the process flow diagram of a kind of guideboard recognition methods that one embodiment of the invention provides;
Fig. 2 is the process flow diagram of a kind of guideboard recognition methods that another embodiment of the present invention provides;
Fig. 3 is the structural representation of a kind of guideboard recognition device that another embodiment of the present invention provides;
Fig. 4 is the structural representation of a kind of guideboard recognition device that another embodiment of the present invention provides;
Fig. 5 is the structural representation of a kind of detection module that another embodiment of the present invention provides;
Fig. 6 is the structural representation of a kind of first filtering module that another embodiment of the present invention provides;
Fig. 7 is the structural representation of a kind of guideboard recognition device that another embodiment of the present invention provides;
Fig. 8 is the structural representation of a kind of second extraction module that another embodiment of the present invention provides;
Fig. 9 is the structural representation of a kind of terminal that another 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.
Along with developing rapidly of Cartographic Technique, there is increasing streetscape map, it can provide 360 degree of panoramic pictures of city, street or other environment for user, and user can be obtained map view as if on the spot in person and experience by the street view image in streetscape map.In order to identify the geographic position at streetscape place, in a lot of street view image, include guideboard.Therefore, by identifying the guideboard region in street view image, and extracting the cartographic information in guideboard, the cartographic information of streetscape map being supplemented, classifies, marking and calibration has great role.Therefore, often need to identify the guideboard in street view image.In order to realize the efficient guideboard identified rapidly in street view image, embodiments provide a kind of guideboard recognition methods.
In conjunction with foregoing, Fig. 1 is the process flow diagram of a kind of guideboard recognition methods provided according to an exemplary embodiment.As shown in Figure 1, the method flow that the embodiment of the present invention provides comprises:
101, color detection is carried out to each pixel in pending street view image, be met the pixel of guideboard color.
102, be communicated with the pixel meeting guideboard color, obtain at least one first connected region.
103, according to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region.
104, the characteristics of image of each second connected region is extracted.
105, use the characteristics of image of guideboard identification SVM to each second connected region to filter, from least one second connected region, filter out guideboard region, wherein, guideboard identification SVM obtains according to the training of guideboard image.
106, Text region is carried out to guideboard region, obtain the guideboard information in guideboard region.
The method that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
Alternatively, before color detection is carried out to each pixel in pending street view image, also comprise:
Obtain initial street view image;
According to the conventional distributed intelligence in the guideboard region in initial street view image, cutting is carried out to initial street view image, obtains pending street view image.
Alternatively, color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color, comprises:
Under HSV (Hue-Saturation-Value, the brightness of tone saturation degree) pattern, HSV threshold filtering is carried out to the color of pixel each in pending street view image;
If the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, then using pixel as the pixel meeting guideboard color.
Alternatively, according to guideboard characteristics of image, area filter is carried out at least one first connected region, obtains at least one second connected region, comprising:
According to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region is filtered;
If arbitrary first connected region meets size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features, then using the first connected region as second connected region.
Alternatively, when the color convergence of the first connected region is for specifying numerical value color class, determine that the first connected region meets Color Distribution Features.
Alternatively, characteristics of image is HOG (Histogram of Gradients, histograms of oriented gradients) feature, before the characteristics of image of use guideboard identification support vector machine to each second connected region filters, also comprises:
Obtain the guideboard image of default value;
Extract the HOG feature of each guideboard image;
Trained the HOG feature of each guideboard image by SVM mode, obtain guideboard identification SVM.
Alternatively, extract the HOG feature of each guideboard image, comprising:
Determine the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises;
Each guideboard image is divided into the block of pixels of specified pixel size;
According to the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises, determine the HOG value of each block of pixels;
According to the HOG value of all block of pixels that each guideboard image comprises, determine the HOG feature of each guideboard image.
Above-mentioned all alternatives, can adopt and combine arbitrarily formation optional embodiment of the present invention, this is no longer going to repeat them.
The content of the corresponding embodiment of composition graphs 1, Fig. 2 is the process flow diagram of a kind of guideboard recognition methods provided according to an exemplary embodiment.As shown in Figure 2, the method flow that the embodiment of the present invention provides comprises:
201, obtain the guideboard image of default value, extract the characteristics of image of each guideboard image, trained the characteristics of image of each guideboard image by SVM mode, obtain guideboard identification SVM.
This step is the process of training guideboard identification SVM.Wherein, when obtaining guideboard image, can by the guideboard image manually chosen in street view image, thus using the guideboard image manually chosen as the guideboard image got.About the concrete quantity of default value, the embodiment of the present invention does not do concrete restriction.Such as, the guideboard image in 100 street view image can be selected, also can select the guideboard image etc. in 1000 street view image.But, in order to make the guideboard identification SVM arrived trained relatively more accurate, thus can as the foundation of the guideboard in follow-up identification street view image, the quantity of this default value is the bigger the better.
In addition, due to when training guideboard identification SVM, realizing according to the characteristics of image of each guideboard image often, therefore, needing the characteristics of image extracting each guideboard image.Wherein, characteristics of image includes but not limited to the HOG feature into guideboard image.
Particularly, when characteristics of image is HOG feature, when extracting the characteristics of image of each guideboard image, be specially the HOG feature extracting each guideboard image.About the mode of the HOG feature of each guideboard image of extraction, include but not limited to that 201a to step 201c realizes as follows:
201a, determine the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises.
Wherein, designated space coordinate system is used for when determining the spatial value of each pixel that each guideboard image comprises, as reference coordinate system.
Because the HOG feature of each guideboard image is relevant to the coordinate figure of the pixel that each guideboard image comprises, therefore, when extracting the HOG feature of each guideboard image, the coordinate figure of each pixel under space coordinates determining that each guideboard image comprises is needed.
When determining coordinate figure under designated space coordinate system of each pixel that each guideboard image comprises, needing with designated space coordinate is reference, determine the value of each pixel in each coordinate axis of designated space coordinate system, according to the value of each pixel in each coordinate axis of designated space coordinate system, determine the coordinate figure of each pixel under designated space coordinate system.
Alternatively, in order to regulate the contrast of each guideboard image, thus reduce the impact that the shade of each guideboard image local and illumination variation cause, the interference of noise can be suppressed simultaneously, gray scale correction method can be adopted to carry out the normalization of color space to each guideboard image.
201b, each guideboard image is divided into the block of pixels of specified pixel size, according to the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises, determines the HOG value of each block of pixels.
Wherein, specified pixel size is for describing the size of each block of pixels.About the number of the pixel that each block of pixels comprises, the embodiment of the present invention does not do concrete restriction.Such as, each block of pixels can comprise 2*2 pixel, also can comprise 4*4 pixel etc.
Particularly, when determining the HOG value of each block of pixels, include but not limited to: the gradient calculating all pixels that each block of pixels comprises, wherein, the gradient of each pixel comprises size and Orientation; Add up the number of different gradient in the gradient of all pixels that each block of pixels comprises, using the HOG value of this number as each block of pixels.
201c, the HOG value of all block of pixels comprised according to each guideboard image, determine the HOG feature of each guideboard image.
In the HOG value of all block of pixels comprised according to each guideboard image, when determining the HOG feature of each guideboard image, the HOG value of all block of pixels that each guideboard image can be comprised is connected in series, thus obtains the HOG feature of each guideboard image.
Further, when after the characteristics of image obtaining each image, just can be trained the characteristics of image of each guideboard image by SVM mode, obtain guideboard identification SVM.Wherein, when being trained the characteristics of image of each guideboard image by SVM mode, the mapping relations setting up guideboard image and characteristics of image are included but not limited to.After training obtains guideboard identification SVM, follow-up when carrying out guideboard identification, according to the HOG feature of the image in region a certain in street view image, can determine whether this region is guideboard region.
It should be noted that, this step is the step before guideboard identification, all will perform this step at every turn, ensureing when carrying out guideboard identification, having trained and having obtained guideboard identification SVM when not carrying out guideboard identification.
202, obtain initial street view image, according to the conventional distributed intelligence in the guideboard region in initial street view image, cutting is carried out to initial street view image, obtains pending street view image.
Wherein, initial street view image is the street view image of showing in streetscape map.Because initial street view image often comprises the full detail of this position, street, such as, the content that street view image comprises could be up to the sky condition in this street, down to the surface state in this street.But be no matter the sky areas in street view image or ground region, all can not comprise guideboard region, namely initial street view image comprises the inactive area much not comprising guideboard region.Therefore, the inactive area in this initial street view image can first be cropped.Particularly, when the inactive area of the initial street view image of cutting, according to the conventional distributed intelligence in the guideboard region in initial street view image, cutting can be carried out to initial street view image.After cropping the inactive area in initial street view image, obtain pending street view image.Wherein, the conventional distributed intelligence in guideboard region is the distribution situation of signpost region usually in initial street view image.
Particularly, because guideboard region is often positioned in the middle part of initial street view image, therefore, the altitude range corresponding to this image medium position can be set to preset height scope, and using the conventional distributed intelligence as guideboard region of this preset height scope.Such as, can using the part between the first preset height to the second preset height from top to bottom in initial street view image as preset height scope.On this basis, when the initial street view image of cutting, the region except this preset height scope in initial street view image can be cropped, thus obtain pending street view image.Wherein, the embodiment of the present invention does not limit the concrete numerical value of the first preset height and the second preset height.
Such as, can using in the middle part of initial street view image 40% region as preset height scope, and crop the image-region of 30% altitude range of initial street view image top and bottom, thus using the image-region of the altitude range of 40% in the middle part of initial street view image as pending street view image.Particularly, if the height of initial street view image is 10cm (centimetre), then can crop the image of initial street view image top 3cm altitude range and the image of bottom 3cm altitude range, using the image-region of middle part 4cm altitude range as pending street view image.
It should be noted that, this step is optional step.By this optional step, a lot of inactive area in initial street view image can be cropped, therefore, during guideboard region in follow-up identification street view image, region to be calculated can be simplified, thus guideboard region recognition efficiency can be improved.But, concrete identify guideboard time, also can not perform this step and directly be realized the identifying of guideboard by step below, referring to each step following.
203, color detection is carried out to each pixel in pending street view image, be met the pixel of guideboard color.
When roading sets up a signpost, often guideboard is set to certain color, such as, blue, green or white etc.Truly can reflect the guideboard color in real road due to the guideboard color in street view image, therefore, first can carry out color detection to pending street view image, may for first connected region in guideboard region to filter out from pending street view image.Particularly, when carrying out color detection, first can carry out color detection to each pixel in pending street view image, being met the pixel of guideboard color.
Wherein, color detection is being carried out to each pixel in pending street view image, when being met the pixel of guideboard color, is including but not limited to that 203a and step 203b realizes as follows:
203a, under HSV pattern, HSV threshold filtering is carried out to the color of pixel each in pending street view image.
Due to often kind of color all corresponding HSV threshold value under HSV pattern, therefore, under HSV pattern, HSV threshold filtering can be carried out to the color of pixel each in pending street view image, thus filters out the pixel belonging to guideboard color.
Particularly, under HSV pattern, when HSV threshold filtering is carried out to the color of pixel each in pending street view image, each pixel can be mapped to the color space at HSV place, obtain the H value of each pixel, S value and V value, then the H value of each pixel, S value and V value are met the H value of the pixel of guideboard color, S value and V value and compare and realize with presetting respectively.
Such as, with when current road construction planning, being set to by guideboard blue is example, when screening meets the pixel of guideboard color, under HSV pattern, blue HSV threshold filtering can be carried out to the color of pixel each in pending street view image, thus filter out blue pixel point.
Particularly, by adding up the relation between color and HSV threshold value, can obtain: under HSV pattern, blue HSV threshold value can comprise following several situation:
The first situation: 100<H<125, S>150, V>55;
The second situation: 90<H<125, S>95, V>55;
The third situation: 90<H<125, S>200, V>155;
4th kind of situation: 90<H<125, S>110, V>55.
On this basis, under HSV pattern, when blue HSV threshold filtering is carried out to the color of pixel each in pending street view image, the HSV value that the H value of each pixel, S value and V value are corresponding with the arbitrary situation in above-mentioned four kinds of situations can be compared and realize.
In like manner, when in current road construction planning, when guideboard color is set to green, by adding up the relation between color and HSV threshold value, can obtain: under HSV pattern, green HSV threshold value can be: 60<H<100, S>150, V>55.
On this basis, under HSV pattern, when green HSV threshold filtering is carried out to the color of pixel each in pending street view image, the H value of each pixel, S value and V value can be compared to realize with " 60<H<100; S>150, V>55 " respectively.
Further, when guideboard color is white, because white HSV threshold value setting is more difficult, therefore, when screening white pixel point, can be split pending street view image by meanshfit (average drifting) algorithm, and obtain the mean value splitting each pixel that the region that obtains comprises, using the color of this mean value as each pixel in this region.By above-mentioned processing procedure, white pixel point can be obtained.Particularly, by adding up the relation between color and HSV threshold value, can obtain: under HSV pattern, can using S<55 as white HSV threshold value.
If the color of the arbitrary pixel of 203b is in the HSV threshold range that guideboard Color pair is answered, then using pixel as the pixel meeting guideboard color.
In conjunction with foregoing, under HSV pattern, when carrying out HSV threshold filtering to the color of pixel each in pending street view image, if the color of a certain pixel meets 100<H<125, S>150, V>55; 90<H<125, S>95, V>55; 90<H<125, S>200, V>155; Any one situation in 90<H<125, S>110, V>55, then using the pixel of this pixel as the satisfied blue HSV threshold value filtered out.If the color of a certain pixel meets 60<H<100, S>150, V>55, then using the pixel of this pixel as the satisfied green HSV threshold value filtered out.If by meanshift algorithm to after pending Image Segmentation Using, the color average obtaining the pixel in a certain region meets S<55, then using the pixel of all pixels in this region as the satisfied white HSV threshold value filtered out.
204, be communicated with the pixel meeting guideboard color, obtain at least one first connected region.
Because guideboard is often a region in street view image, in order to obtain the guideboard region in street view image, the pixel meeting guideboard color can be communicated with, obtain at least one first connected region.Wherein, the first connected region may be guideboard region.About the quantity of the first connected region, the embodiment of the present invention does not do concrete restriction.During concrete enforcement, combination is needed to meet the distribution situation of the pixel of guideboard color and determine.
205, according to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region.
Particularly, due to when setting up a signpost, the size, wide high proportion, the anglec of rotation, color distribution etc. of guideboard have certain standard, the size of the guideboard in street view image, wide high proportion, the anglec of rotation, color distribution also all meet some requirements, therefore, according to guideboard characteristics of image, area filter can be carried out at least one first connected region, obtain at least one second connected region.Wherein, the second connected region be filter out from the first connected region do not belong to guideboard region region after, screen the region that may belong to guideboard region that obtains.
Wherein, area filter is being carried out at least one first connected region, when obtaining at least one second connected region, is including but not limited to that 205a and step 205b realizes as follows:
205a, according to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region to be filtered.
Particularly, the size of the guideboard image in street view image, wide high proportion, the anglec of rotation, color distribution have certain threshold range.Therefore, the size characteristic threshold range of guideboard image, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features can be added up in advance, make when carrying out guideboard identification, according to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region can be filtered.
Such as, by analyzing a large amount of guideboard images, the threshold range of the size characteristic in the guideboard region counted is: 23< height <165,40< width <600; Wide high proportion threshold range is: 1.6< the ratio of width to height <7.5; Anglec of rotation threshold range is: anglec of rotation <40 degree.
In addition, when determining whether some first UNICOM regions meet Color Distribution Features, because guideboard is made up of appointment numerical value kind color usually, such as, guideboard comprises the color of plate and the color of word, and namely guideboard is made up of two kinds of colors.Therefore, when determining whether this first connected region meets the Color Distribution Features of guideboard, can by determining whether this first connected region realizes by specifying a numerical value color class to form.Particularly, when determining whether the first connected region is made up of an appointment numerical value color class, whether the color can passing through some first connected regions of K-means (K-average) algorithm detection can be polymerized to is specified a numerical value color class to realize.Such as, if guideboard is made up of two kinds of colors usually, then whether the color that can detect this first connected region by K-means algorithm can be polymerized to two color class realizes.If the color of this first connected region can be polymerized to specify a numerical value color class, such as, the color of this first connected region can be polymerized to two color class, then determine that this first connected region meets the Color Distribution Features of guideboard.
About according to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, the order during mode of filtering some first connected regions, the embodiment of the present invention does not do concrete restriction.During concrete enforcement, first can carry out size characteristic filtration to this first connected region, then carry out wide high proportion characteristic filter, then carry out anglec of rotation characteristic filter, finally carry out color distribution detection.Certainly, other order can also be adopted to realize, do not repeat them here.
If arbitrary first connected region of 205b meets size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features, then using the first connected region as second connected region.
In conjunction with foregoing, if some first connected regions can not meet above-mentioned size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features simultaneously, then this first connected region is filtered out; If some first connected regions meet above-mentioned size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features simultaneously, then using this first connected region as second connected region.
206, the characteristics of image of each second connected region is extracted, the characteristics of image of guideboard identification SVM to each second connected region is used to filter, from at least one second connected region, filter out guideboard region, wherein, guideboard identification SVM obtains according to the training of guideboard image.
In order to filter out guideboard region from the second connected region, the guideboard identification SVM that can also obtain in conjunction with the characteristics of image of the second connected region and training in advance filters further.Wherein, the characteristics of image of the second connected region is consistent with the characteristics of image used during training guideboard identification SVM.Particularly, if the characteristics of image used during training guideboard identification SVM is HOG feature, then the HOG feature of each second connected region should be extracted herein.
About the mode using the characteristics of image of guideboard identification support vector machine to each second connected region to filter, include but not limited to: the guideboard characteristics of image of each second connected region extracted is inputted guideboard identification SVM, determines whether this second connected region is guideboard region according to the Output rusults of guideboard identification SVM.If it is not guideboard region that the Output rusults of guideboard identification SVM identifies this second connected region, then this second connected region is abandoned; If it is guideboard region that the Output rusults of guideboard identification SVM identifies this second connected region, then using this second connected region as guideboard region.
207, Text region is carried out to guideboard region, obtain the guideboard information in guideboard region.
Wherein, guideboard information comprises the street information, guideboard indicated direction information etc. that guideboard identifies.About the mode of guideboard region being carried out to Text region, include but not limited to: first binary conversion treatment is carried out to guideboard region, then use OCR (Optical Character Recognition, optical character identification) mode determination guideboard information.
By guideboard information, the geographic position of guideboard place street view image can be determined, thus realize the location to the street view image comprising guideboard region, be convenient to the information expanding streetscape map label.
Determine through statistics, by adopting said method identification guideboard, the recognition success rate of blue guideboard is 95%, and the recognition success rate of green guideboard and white guideboard is 87%.Therefore, the method provided by the embodiment of the present invention, can realize the automatic identifying of guideboard.
The method that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
Fig. 3 is the structural representation of a kind of guideboard recognition device provided according to an exemplary embodiment, the guideboard recognition methods that this guideboard recognition device provides for performing above-mentioned Fig. 1 or 2 illustrated embodiments.See Fig. 3, this device comprises:
Detection module 301, for carrying out color detection to each pixel in pending street view image, is met the pixel of guideboard color;
Connectivity module 302, for being communicated with the pixel meeting guideboard color, obtains at least one first connected region;
First filtering module 303, for according to guideboard characteristics of image, carries out area filter at least one first connected region, obtains at least one second connected region;
First extraction module 304, for extracting the characteristics of image of each second connected region;
Second filtering module 305, filter for using the characteristics of image of guideboard identification support vector machine to each second connected region, from at least one second connected region, filter out guideboard region, wherein, guideboard identification support vector machine obtains according to the training of guideboard image;
Identification module 306, for carrying out Text region to guideboard region, obtains the guideboard information in guideboard region.
Alternatively, see Fig. 4, device, also comprises:
First acquisition module 307, for obtaining initial street view image;
Cutting module 308, for according to the conventional distributed intelligence in the guideboard region in initial street view image, carries out cutting to initial street view image, obtains pending street view image.
Alternatively, see Fig. 5, detection module 301, comprising:
Filter element 3011, under HSV pattern, carries out HSV threshold filtering to the color of pixel each in pending street view image;
First determining unit 3012, for when the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, using pixel as the pixel meeting guideboard color.
Alternatively, see Fig. 6, the first filtering module 303, comprising:
Filter element 3031, for according to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, filters each first connected region;
Second determining unit 3032, during for meeting size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features when arbitrary first connected region, using the first connected region as second connected region.
Alternatively, the second determining unit 3032 when determining whether the first UNICOM region meets Color Distribution Features, for:
When the color convergence of the first connected region is for specifying numerical value color class, determine that the first connected region meets Color Distribution Features.
Alternatively, characteristics of image is HOG feature, and see Fig. 7, device, also comprises:
Second acquisition module 309, for obtaining the guideboard image of default value;
Second extraction module 310, for extracting the HOG feature of each guideboard image;
Training module 311, for being trained the HOG feature of each guideboard image by support vector machine mode, obtains guideboard identification support vector machine.
Alternatively, see Fig. 8, the second extraction module 310 comprises:
3rd determining unit 3101, for determining the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises;
Division unit 3102, for being divided into the block of pixels of specified pixel size by each guideboard image;
4th determining unit 3103, for the coordinate figure of each pixel under designated space coordinate system comprised according to each guideboard image, determines the HOG value of each block of pixels;
5th determining unit 3104, for the HOG value of all block of pixels comprised according to each guideboard image, determines the HOG feature of each guideboard image.
Above-mentioned all alternatives, can adopt and combine arbitrarily formation optional embodiment of the present invention, this is no longer going to repeat them.
The device that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
Please refer to Fig. 9, it illustrates the structural representation of the terminal involved by the embodiment of the present invention, this terminal may be used for implementing the guideboard recognition methods that above-mentioned Fig. 1 or Fig. 2 embodiment provides.Specifically:
Terminal 900 can comprise RF (Radio Frequency, radio frequency) circuit 110, the storer 120 including one or more computer-readable recording mediums, input block 130, display unit 140, sensor 150, voicefrequency circuit 160, WiFi (Wireless Fidelity, Wireless Fidelity) module 170, include the parts such as processor 180 and power supply 190 that more than or processes core.It will be understood by those skilled in the art that the restriction of the not structure paired terminal of the terminal structure shown in Fig. 9, the parts more more or less than diagram can be comprised, or combine some parts, or different parts are arranged.Wherein:
RF circuit 110 can be used for receiving and sending messages or in communication process, the reception of signal and transmission, especially, after being received by the downlink information of base station, transfer to more than one or one processor 180 to process; In addition, base station is sent to by relating to up data.Usually, RF circuit 110 includes but not limited to antenna, at least one amplifier, tuner, one or more oscillator, subscriber identity module (SIM) card, transceiver, coupling mechanism, LNA (Low Noise Amplifier, low noise amplifier), diplexer etc.In addition, RF circuit 110 can also by radio communication and network and other devices communicatings.Described radio communication can use arbitrary communication standard or agreement, include but not limited to GSM (Global System of Mobile communication, global system for mobile communications), GPRS (General Packet Radio Service, general packet radio service), CDMA (Code Division Multiple Access, CDMA), WCDMA (Wideband CodeDivision Multiple Access, Wideband Code Division Multiple Access (WCDMA)), LTE (Long Term Evolution, Long Term Evolution), Email, SMS (Short Messaging Service, Short Message Service) etc.
Storer 120 can be used for storing software program and module, and processor 180 is stored in software program and the module of storer 120 by running, thus performs the application of various function and data processing.Storer 120 mainly can comprise storage program district and store data field, and wherein, storage program district can store operating system, application program (such as sound-playing function, image player function etc.) etc. needed at least one function; Store data field and can store the data (such as voice data, phone directory etc.) etc. created according to the use of terminal 900.In addition, storer 120 can comprise high-speed random access memory, can also comprise nonvolatile memory, such as at least one disk memory, flush memory device or other volatile solid-state parts.Correspondingly, storer 120 can also comprise Memory Controller, to provide the access of processor 180 and input block 130 pairs of storeies 120.
Input block 130 can be used for the numeral or the character information that receive input, and produces and to arrange with user and function controls relevant keyboard, mouse, control lever, optics or trace ball signal and inputs.Particularly, input block 130 can comprise Touch sensitive surface 131 and other input equipments 132.Touch sensitive surface 131, also referred to as touch display screen or Trackpad, user can be collected or neighbouring touch operation (such as user uses any applicable object or the operations of annex on Touch sensitive surface 131 or near Touch sensitive surface 131 such as finger, stylus) thereon, and drive corresponding coupling arrangement according to the formula preset.Optionally, Touch sensitive surface 131 can comprise touch detecting apparatus and touch controller two parts.Wherein, touch detecting apparatus detects the touch orientation of user, and detects the signal that touch operation brings, and sends signal to touch controller; Touch controller receives touch information from touch detecting apparatus, and converts it to contact coordinate, then gives processor 180, and the order that energy receiving processor 180 is sent also is performed.In addition, the polytypes such as resistance-type, condenser type, infrared ray and surface acoustic wave can be adopted to realize Touch sensitive surface 131.Except Touch sensitive surface 131, input block 130 can also comprise other input equipments 132.Particularly, other input equipments 132 can include but not limited to one or more in physical keyboard, function key (such as volume control button, switch key etc.), trace ball, mouse, control lever etc.
Display unit 140 can be used for the various graphical user interface showing information or the information being supplied to user and the terminal 900 inputted by user, and these graphical user interface can be made up of figure, text, icon, video and its combination in any.Display unit 140 can comprise display panel 141, optionally, the form such as LCD (Liquid Crystal Display, liquid crystal display), OLED (Organic Light-Emitting Diode, Organic Light Emitting Diode) can be adopted to configure display panel 141.Further, Touch sensitive surface 131 can cover display panel 141, when Touch sensitive surface 131 detects thereon or after neighbouring touch operation, send processor 180 to determine the type of touch event, on display panel 141, provide corresponding vision to export with preprocessor 180 according to the type of touch event.Although in fig .9, Touch sensitive surface 131 and display panel 141 be as two independently parts realize input and input function, in certain embodiments, can by Touch sensitive surface 131 and display panel 141 integrated and realize input and output function.
Terminal 900 also can comprise at least one sensor 150, such as optical sensor, motion sensor and other sensors.Particularly, optical sensor can comprise ambient light sensor and proximity transducer, and wherein, ambient light sensor the light and shade of environmentally light can regulate the brightness of display panel 141, proximity transducer when terminal 900 moves in one's ear, can cut out display panel 141 and/or backlight.As the one of motion sensor, Gravity accelerometer can detect the size of all directions (are generally three axles) acceleration, size and the direction of gravity can be detected time static, can be used for the application (such as horizontal/vertical screen switching, dependent game, magnetometer pose calibrating) of identification terminal attitude, Vibration identification correlation function (such as passometer, knock) etc.; As for terminal 900 also other sensors such as configurable gyroscope, barometer, hygrometer, thermometer, infrared ray sensor, do not repeat them here.
Voicefrequency circuit 160, loudspeaker 161, microphone 162 can provide the audio interface between user and terminal 900.Voicefrequency circuit 160 can by receive voice data conversion after electric signal, be transferred to loudspeaker 161, by loudspeaker 161 be converted to voice signal export; On the other hand, the voice signal of collection is converted to electric signal by microphone 162, voice data is converted to after being received by voicefrequency circuit 160, after again voice data output processor 180 being processed, through RF circuit 110 to send to such as another terminal, or export voice data to storer 120 to process further.Voicefrequency circuit 160 also may comprise earphone jack, to provide the communication of peripheral hardware earphone and terminal 900.
WiFi belongs to short range wireless transmission technology, and terminal 900 can help user to send and receive e-mail by WiFi module 170, browse webpage and access streaming video etc., and its broadband internet wireless for user provides is accessed.Although Fig. 9 shows WiFi module 170, be understandable that, it does not belong to must forming of terminal 900, can omit in the scope of essence not changing invention as required completely.
Processor 180 is control centers of terminal 900, utilize the various piece of various interface and the whole terminal of connection, software program in storer 120 and/or module is stored in by running or performing, and call the data be stored in storer 120, perform various function and the process data of terminal 900, thus integral monitoring is carried out to terminal.Optionally, processor 180 can comprise one or more process core; Preferably, processor 180 accessible site application processor and modem processor, wherein, application processor mainly processes operating system, user interface and application program etc., and modem processor mainly processes radio communication.Be understandable that, above-mentioned modem processor also can not be integrated in processor 180.
Terminal 900 also comprises the power supply 190 (such as battery) of powering to all parts, preferably, power supply can be connected with processor 180 logic by power-supply management system, thus realizes the functions such as management charging, electric discharge and power managed by power-supply management system.Power supply 190 can also comprise one or more direct current or AC power, recharging system, power failure detection circuit, power supply changeover device or the random component such as inverter, power supply status indicator.
Although not shown, terminal 900 can also comprise camera, bluetooth module etc., does not repeat them here.Specifically in the present embodiment, the display unit of terminal is touch-screen display, and terminal also includes storer, and one or more than one program, one of them or more than one program are stored in storer, and are configured to be performed by more than one or one processor.Described more than one or one routine package is containing the instruction for performing following operation:
Color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color;
Be communicated with the pixel meeting guideboard color, obtain at least one first connected region;
According to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region;
Extract the characteristics of image of each second connected region;
Use the characteristics of image of guideboard identification support vector machine to each second connected region to filter, from least one second connected region, filter out guideboard region, wherein, guideboard identification support vector machine obtains according to the training of guideboard image;
Text region is carried out to guideboard region, obtains the guideboard information in guideboard region.
Suppose that above-mentioned is the first possible embodiment, in the embodiment that the second then provided based on the embodiment that the first is possible is possible, in the storer of terminal, also comprise the instruction for performing following operation: before color detection is carried out to each pixel in pending street view image, also comprise:
Obtain initial street view image;
According to the conventional distributed intelligence in the guideboard region in initial street view image, cutting is carried out to initial street view image, obtains pending street view image.
In the third the possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: color detection is carried out to each pixel in pending street view image, be met the pixel of guideboard color, comprise:
Under HSV pattern, HSV threshold filtering is carried out to the color of pixel each in pending street view image;
If the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, then using pixel as the pixel meeting guideboard color.
In the 4th kind of possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: according to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region, comprising:
According to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region is filtered;
If arbitrary first connected region meets size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features, then using the first connected region as second connected region.
In the 5th kind of possible embodiment provided based on the 4th kind of possible embodiment, in the storer of terminal, also comprise the instruction for performing following operation:
When the color convergence of the first connected region is for specifying numerical value color class, determine that the first connected region meets Color Distribution Features.
In the 6th kind of possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: characteristics of image is HOG feature, before the characteristics of image of use guideboard identification support vector machine to each second connected region filters, also comprise:
Obtain the guideboard image of default value;
Extract the HOG feature of each guideboard image;
Trained the HOG feature of each guideboard image by support vector machine mode, obtain guideboard identification support vector machine.
In the 7th kind of possible embodiment provided based on the 6th kind of possible embodiment, in the storer of terminal, also comprising the instruction for performing following operation: the HOG feature extracting each guideboard image, comprising:
Determine the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises;
Each guideboard image is divided into the block of pixels of specified pixel size;
According to the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises, determine the HOG value of each block of pixels;
According to the HOG value of all block of pixels that each guideboard image comprises, determine the HOG feature of each guideboard image.
The terminal that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
Embodiments provide a kind of computer-readable recording medium, this computer-readable recording medium can be the computer-readable recording medium comprised in the storer in above-described embodiment; Also can be individualism, be unkitted the computer-readable recording medium allocated in terminal.This computer-readable recording medium stores more than one or one program, and this more than one or one program is used for performing guideboard recognition methods by one or more than one processor, and the method comprises:
Color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color;
Be communicated with the pixel meeting guideboard color, obtain at least one first connected region;
According to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region;
Extract the characteristics of image of each second connected region;
Use the characteristics of image of guideboard identification support vector machine to each second connected region to filter, from least one second connected region, filter out guideboard region, wherein, guideboard identification support vector machine obtains according to the training of guideboard image;
Text region is carried out to guideboard region, obtains the guideboard information in guideboard region.
Suppose that above-mentioned is the first possible embodiment, in the embodiment that the second then provided based on the embodiment that the first is possible is possible, in the storer of terminal, also comprise the instruction for performing following operation: before color detection is carried out to each pixel in pending street view image, also comprise:
Obtain initial street view image;
According to the conventional distributed intelligence in the guideboard region in initial street view image, cutting is carried out to initial street view image, obtains pending street view image.
In the third the possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: color detection is carried out to each pixel in pending street view image, be met the pixel of guideboard color, comprise:
Under HSV pattern, HSV threshold filtering is carried out to the color of pixel each in pending street view image;
If the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, then using pixel as the pixel meeting guideboard color.
In the 4th kind of possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: according to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region, comprising:
According to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region is filtered;
If arbitrary first connected region meets size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features, then using the first connected region as second connected region.
In the 5th kind of possible embodiment provided based on the 4th kind of possible embodiment, in the storer of terminal, also comprise the instruction for performing following operation:
When the color convergence of the first connected region is for specifying numerical value color class, determine that the first connected region meets described Color Distribution Features.
In the 6th kind of possible embodiment provided based on the embodiment that the first is possible, in the storer of terminal, also comprise the instruction for performing following operation: characteristics of image is HOG feature, before the characteristics of image of use guideboard identification support vector machine to each second connected region filters, also comprise:
Obtain the guideboard image of default value;
Extract the HOG feature of each guideboard image;
Trained the HOG feature of each guideboard image by support vector machine mode, obtain guideboard identification support vector machine.
In the 7th kind of possible embodiment provided based on the 6th kind of possible embodiment, in the storer of terminal, also comprising the instruction for performing following operation: the HOG feature extracting each guideboard image, comprising:
Determine the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises;
Each guideboard image is divided into the block of pixels of specified pixel size;
According to the coordinate figure of each pixel under designated space coordinate system that each guideboard image comprises, determine the HOG value of each block of pixels;
According to the HOG value of all block of pixels that each guideboard image comprises, determine the HOG feature of each guideboard image.
The computer-readable recording medium that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
Provide a kind of graphical user interface in the embodiment of the present invention, this graphical user interface is used in terminal, and this terminal comprises touch-screen display, storer and one or more than one processor for performing one or more than one program; This graphical user interface comprises:
Color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color;
Be communicated with the pixel meeting guideboard color, obtain at least one first connected region;
According to guideboard characteristics of image, area filter is carried out at least one first connected region, obtain at least one second connected region;
Extract the characteristics of image of each second connected region;
Use the characteristics of image of guideboard identification support vector machine to each second connected region to filter, from least one second connected region, filter out guideboard region, wherein, guideboard identification support vector machine obtains according to the training of guideboard image;
Text region is carried out to guideboard region, obtains the guideboard information in guideboard region.
The graphical user interface that the embodiment of the present invention provides, by carrying out color detection to each pixel of pending street view image, be met the pixel of guideboard color, and by after being communicated with the pixel meeting guideboard color and obtaining the first connected region, area filter is carried out again according to guideboard characteristics of image, thus from the first connected region, filter out the second connected region meeting guideboard characteristics of image, again further according to the guideboard identification SVM that characteristics of image and the training in advance of the second connected region obtain, filtration obtains guideboard region, and then the identifying of guideboard is completed by the guideboard information extracting guideboard region.Because guideboard identifying is without the need to artificial participation, therefore, not only save guideboard identification cost, and improve guideboard recognition efficiency.
It should be noted that: the guideboard recognition device that above-described embodiment provides is when identifying guideboard, only be illustrated with the division of above-mentioned each functional module, in practical application, can distribute as required and by above-mentioned functions and be completed by different functional modules, inner structure by device is divided into different functional modules, to complete all or part of function described above.In addition, the guideboard recognition device that above-described embodiment provides and guideboard recognition methods embodiment belong to same design, and its specific implementation process refers to embodiment of the method, repeats no more here.
One of ordinary skill in the art will appreciate that all or part of step realizing above-described embodiment can have been come by hardware, the hardware that also can carry out instruction relevant by program completes, described program can be stored in a kind of computer-readable recording medium, the above-mentioned storage medium mentioned can be ROM (read-only memory), disk or CD etc.
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 (14)
1. a guideboard recognition methods, is characterized in that, described method comprises:
Color detection is carried out to each pixel in pending street view image, is met the pixel of guideboard color;
Be communicated with the described pixel meeting guideboard color, obtain at least one first connected region;
According to guideboard characteristics of image, area filter is carried out at least one first connected region described, obtains at least one second connected region;
Extract the characteristics of image of each second connected region;
Use the characteristics of image of guideboard identification support vector machine to each second connected region to filter, from least one second connected region described, filter out guideboard region, described guideboard identification support vector machine obtains according to the training of guideboard image;
Text region is carried out to described guideboard region, obtains the guideboard information in described guideboard region.
2. method according to claim 1, is characterized in that, described color detection is carried out to each pixel in pending street view image before, also comprise:
Obtain initial street view image;
According to the conventional distributed intelligence in the guideboard region in described initial street view image, cutting is carried out to described initial street view image, obtains described pending street view image.
3. method according to claim 1, is characterized in that, describedly carries out color detection to each pixel in pending street view image, is met the pixel of guideboard color, comprises:
Under tone saturation degree brightness HSV pattern, HSV threshold filtering is carried out to the color of pixel each in pending street view image;
If the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, then using described pixel as the pixel meeting guideboard color.
4. method according to claim 1, is characterized in that, described according to guideboard characteristics of image, carries out area filter, obtain at least one second connected region, comprising at least one first connected region described:
According to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, each first connected region is filtered;
If arbitrary first connected region meets size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features, then using described first connected region as second connected region.
5. method according to claim 4, is characterized in that,
When the color convergence of described first connected region is for specifying numerical value color class, determine that described first connected region meets described Color Distribution Features.
6. method according to claim 1, is characterized in that, described characteristics of image is histograms of oriented gradients HOG feature, before the characteristics of image of described use guideboard identification support vector machine to each second connected region filters, also comprises:
Obtain the guideboard image of default value;
Extract the HOG feature of each guideboard image;
Trained the HOG feature of each guideboard image by support vector machine mode, obtain guideboard identification support vector machine.
7. method according to claim 6, is characterized in that, the HOG feature of each guideboard image of described extraction, comprising:
Determine the coordinate figure of each pixel under designated space coordinate system that described each guideboard image comprises;
Described each guideboard image is divided into the block of pixels of specified pixel size;
According to the coordinate figure of each pixel under described designated space coordinate system that described each guideboard image comprises, determine the HOG value of each block of pixels;
According to the HOG value of all block of pixels that described each guideboard image comprises, determine the HOG feature of described each guideboard image.
8. a guideboard recognition device, is characterized in that, described device comprises:
Detection module, for carrying out color detection to each pixel in pending street view image, is met the pixel of guideboard color;
Connectivity module, for being communicated with the described pixel meeting guideboard color, obtains at least one first connected region;
First filtering module, for according to guideboard characteristics of image, carries out area filter at least one first connected region described, obtains at least one second connected region;
First extraction module, for extracting the characteristics of image of each second connected region;
Second filtering module, filter for using the characteristics of image of guideboard identification support vector machine to each second connected region, from at least one second connected region described, filter out guideboard region, described guideboard identification support vector machine obtains according to the training of guideboard image;
Identification module, for carrying out Text region to described guideboard region, obtains the guideboard information in described guideboard region.
9. device according to claim 8, is characterized in that, described device, also comprises:
First acquisition module, for obtaining initial street view image;
Cutting module, for according to the conventional distributed intelligence in the guideboard region in described initial street view image, carries out cutting to described initial street view image, obtains described pending street view image.
10. device according to claim 8, is characterized in that, described detection module, comprising:
Filter element, under tone saturation degree brightness HSV pattern, carries out HSV threshold filtering to the color of pixel each in pending street view image;
First determining unit, for when the color of arbitrary pixel is in the HSV threshold range that guideboard Color pair is answered, using described pixel as the pixel meeting guideboard color.
11. devices according to claim 8, is characterized in that, described first filtering module, comprising:
Filter element, for according to guideboard size characteristic, wide high proportion feature, anglec of rotation feature and Color Distribution Features, filters each first connected region;
Second determining unit, during for meeting size characteristic threshold range, wide high proportion characteristic threshold value scope, anglec of rotation characteristic threshold value scope and Color Distribution Features when arbitrary first connected region, using described first connected region as second connected region.
12. devices according to claim 11, is characterized in that, described second determining unit when determining whether described first UNICOM region meets described Color Distribution Features, for:
When the color convergence of described first connected region is for specifying numerical value color class, determine that described first connected region meets described Color Distribution Features.
13. devices according to claim 8, is characterized in that, described characteristics of image is histograms of oriented gradients HOG feature, and described device, also comprises:
Second acquisition module, for obtaining the guideboard image of default value;
Second extraction module, for extracting the HOG feature of each guideboard image;
Training module, for being trained the HOG feature of each guideboard image by support vector machine mode, obtains guideboard identification support vector machine.
14. devices according to claim 13, is characterized in that, described second extraction module, comprising:
3rd determining unit, for determining the coordinate figure of each pixel under designated space coordinate system that described each guideboard image comprises;
Division unit, for being divided into the block of pixels of specified pixel size by described each guideboard image;
4th determining unit, for the coordinate figure of each pixel under described designated space coordinate system comprised according to described each guideboard image, determines the HOG value of each block of pixels;
5th determining unit, for the HOG value of all block of pixels comprised according to described each guideboard image, determines the HOG feature of described each guideboard image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410663502.2A CN104463105B (en) | 2014-11-19 | 2014-11-19 | Guideboard recognition methods and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410663502.2A CN104463105B (en) | 2014-11-19 | 2014-11-19 | Guideboard recognition methods and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104463105A true CN104463105A (en) | 2015-03-25 |
| CN104463105B CN104463105B (en) | 2017-11-03 |
Family
ID=52909121
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410663502.2A Active CN104463105B (en) | 2014-11-19 | 2014-11-19 | Guideboard recognition methods and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104463105B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105279488A (en) * | 2015-10-16 | 2016-01-27 | 西南交通大学 | Mileage signboard automatic recognition method for road video routing inspection |
| CN105678318A (en) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术（北京）有限公司 | Traffic label matching method and apparatus |
| CN105825203A (en) * | 2016-03-30 | 2016-08-03 | 大连理工大学 | Ground arrowhead sign detection and identification method based on dotted pair matching and geometric structure matching |
| CN106503011A (en) * | 2015-09-07 | 2017-03-15 | 高德软件有限公司 | A kind of map data processing method and device |
| CN106560835A (en) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | Guideboard identification method and device |
| CN106709412A (en) * | 2015-11-17 | 2017-05-24 | 腾讯科技（深圳）有限公司 | Traffic sign detection method and apparatus |
| CN112528970A (en) * | 2021-02-07 | 2021-03-19 | 禾多科技（北京）有限公司 | Guideboard detection method, device, equipment and computer readable medium |
| CN112668675A (en) * | 2021-03-22 | 2021-04-16 | 腾讯科技（深圳）有限公司 | Image processing method and device, computer equipment and storage medium |
| WO2021212977A1 (en) * | 2020-04-24 | 2021-10-28 | 京东方科技集团股份有限公司 | Image color filtering method and apparatus, electronic device, and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101404117A (en) * | 2008-10-21 | 2009-04-08 | 东软集团股份有限公司 | Traffic sign recognition method and device |
| CN102542260A (en) * | 2011-12-30 | 2012-07-04 | 中南大学 | Method for recognizing road traffic sign for unmanned vehicle |
| CN102819728A (en) * | 2012-07-17 | 2012-12-12 | 中国航天科工集团第三研究院第八三五七研究所 | Traffic sign detection method based on classification template matching |
| CN103020623A (en) * | 2011-09-23 | 2013-04-03 | 株式会社理光 | Traffic sign detection method and equipment |
| CN103971126A (en) * | 2014-05-12 | 2014-08-06 | 百度在线网络技术（北京）有限公司 | Method and device for identifying traffic signs |
| US20140300741A1 (en) * | 2010-08-23 | 2014-10-09 | Ford Global Technologies, Llc | Arrangement and Method for Recognizing Road Signs |
-
2014
- 2014-11-19 CN CN201410663502.2A patent/CN104463105B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101404117A (en) * | 2008-10-21 | 2009-04-08 | 东软集团股份有限公司 | Traffic sign recognition method and device |
| US20140300741A1 (en) * | 2010-08-23 | 2014-10-09 | Ford Global Technologies, Llc | Arrangement and Method for Recognizing Road Signs |
| CN103020623A (en) * | 2011-09-23 | 2013-04-03 | 株式会社理光 | Traffic sign detection method and equipment |
| CN102542260A (en) * | 2011-12-30 | 2012-07-04 | 中南大学 | Method for recognizing road traffic sign for unmanned vehicle |
| CN102819728A (en) * | 2012-07-17 | 2012-12-12 | 中国航天科工集团第三研究院第八三五七研究所 | Traffic sign detection method based on classification template matching |
| CN103971126A (en) * | 2014-05-12 | 2014-08-06 | 百度在线网络技术（北京）有限公司 | Method and device for identifying traffic signs |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106503011A (en) * | 2015-09-07 | 2017-03-15 | 高德软件有限公司 | A kind of map data processing method and device |
| CN106560835B (en) * | 2015-09-30 | 2019-05-24 | 高德软件有限公司 | A kind of guideboard recognition methods and device |
| CN106560835A (en) * | 2015-09-30 | 2017-04-12 | 高德软件有限公司 | Guideboard identification method and device |
| CN105279488B (en) * | 2015-10-16 | 2018-12-25 | 西南交通大学 | A kind of mileage Sign Board automatic identifying method of highway video routing inspection |
| CN105279488A (en) * | 2015-10-16 | 2016-01-27 | 西南交通大学 | Mileage signboard automatic recognition method for road video routing inspection |
| CN106709412B (en) * | 2015-11-17 | 2021-05-11 | 腾讯科技（深圳）有限公司 | Traffic sign detection method and device |
| CN106709412A (en) * | 2015-11-17 | 2017-05-24 | 腾讯科技（深圳）有限公司 | Traffic sign detection method and apparatus |
| CN105678318B (en) * | 2015-12-31 | 2019-03-08 | 百度在线网络技术（北京）有限公司 | The matching process and device of traffic sign |
| CN105678318A (en) * | 2015-12-31 | 2016-06-15 | 百度在线网络技术（北京）有限公司 | Traffic label matching method and apparatus |
| CN105825203A (en) * | 2016-03-30 | 2016-08-03 | 大连理工大学 | Ground arrowhead sign detection and identification method based on dotted pair matching and geometric structure matching |
| CN105825203B (en) * | 2016-03-30 | 2018-12-18 | 大连理工大学 | Based on point to matching and the matched ground arrow mark detection of geometry and recognition methods |
| WO2021212977A1 (en) * | 2020-04-24 | 2021-10-28 | 京东方科技集团股份有限公司 | Image color filtering method and apparatus, electronic device, and storage medium |
| US11830107B2 (en) | 2020-04-24 | 2023-11-28 | Boe Technology Group Co., Ltd. | Method and apparatus for filtering image colors, electronic device and storage medium |
| CN112528970A (en) * | 2021-02-07 | 2021-03-19 | 禾多科技（北京）有限公司 | Guideboard detection method, device, equipment and computer readable medium |
| CN112668675A (en) * | 2021-03-22 | 2021-04-16 | 腾讯科技（深圳）有限公司 | Image processing method and device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104463105B (en) | 2017-11-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104463105A (en) | Guide board recognizing method and device | |
| CN105005457B (en) | Geographical location methods of exhibiting and device | |
| CN103871051B (en) | Image processing method, device and electronic equipment | |
| CN104135609B (en) | Auxiliary photo-taking method, apparatus and terminal | |
| CN106156711B (en) | Text line positioning method and device | |
| CN106296617B (en) | The processing method and processing device of facial image | |
| CN105809647A (en) | Automatic defogging photographing method, device and equipment | |
| CN109391762A (en) | A kind of method and apparatus of track up | |
| CN103729636A (en) | Method and device for cutting character and electronic device | |
| CN104679509A (en) | Graph rendering method and device | |
| CN103455969A (en) | Method and device for processing image | |
| CN104967790A (en) | Photo shooting method, photo shooting apparatus and mobile terminal | |
| CN103325107A (en) | Method, device and terminal device for processing image | |
| CN105989572B (en) | Picture processing method and device | |
| CN103714161A (en) | Image thumbnail generation method and device and terminal | |
| CN110022553A (en) | A kind of subscriber identification card management method and mobile terminal | |
| CN106204423A (en) | A kind of picture-adjusting method based on augmented reality, device and terminal | |
| CN105306802A (en) | Photographing mode switching method and device | |
| CN104751410A (en) | Image and two-dimensional code fusion method and device | |
| CN106203459A (en) | Image processing method and device | |
| CN106204552A (en) | The detection method of a kind of video source and device | |
| CN106296634A (en) | A kind of method and apparatus detecting similar image | |
| CN106961710A (en) | A kind of method for network access and terminal | |
| CN108427938A (en) | Image processing method, device, storage medium and electronic equipment | |
| CN110519503A (en) | A kind of acquisition methods and mobile terminal of scan image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20170830 Address after: 100084 Haidian District Tsinghua Yuan Beijing No. 1 Applicant after: Tsinghua University Applicant after: Shenzhen Tencent Computer System Co., Ltd. Address before: The South Road in Guangdong province Shenzhen city Fiyta building 518000 floor 5-10 Nanshan District high tech Zone Applicant before: Shenzhen Tencent Computer System Co., Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |