CN106709518A - Android platform-based blind way recognition system - Google Patents
Android platform-based blind way recognition system Download PDFInfo
- Publication number
- CN106709518A CN106709518A CN201611184275.0A CN201611184275A CN106709518A CN 106709518 A CN106709518 A CN 106709518A CN 201611184275 A CN201611184275 A CN 201611184275A CN 106709518 A CN106709518 A CN 106709518A
- Authority
- CN
- China
- Prior art keywords
- sidewalk
- visually impaired
- impaired people
- image
- module
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06V10/48—Extraction of image or video features by mapping characteristic values of the pattern into a parameter space, e.g. Hough transformation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an Android platform-based blind way recognition system. The system comprises an image acquisition module, an image processing module and a voice assistance module, wherein the image acquisition module is used for realizing real-time acquisition of a blind way image; the image processing module is used for performing classification processing on the detected blind way image and transmitting a processing result to the voice assistance module; and the voice assistance module broadcasts real-time condition information of a blind way and guides the blind to walk according to the image processing result. The system has the advantages that a blind way boundary, a corner and a barrier position can be efficiently detected in real time; and various blind way recognition algorithms and barrier extraction algorithms are applied to an Android intelligent mobile phone through OpenCV, and the blind way recognition system integrating image real-time acquisition and processing is established.
Description
Technical field
The present invention relates to digital image processing techniques field, and in particular to know to a kind of sidewalk for visually impaired people based on Android platform
Other system.
Background technology
Blind person (visually impaired people) is due to ablepsia, it is impossible to directly observe environment at one's side, and pole is often had in trip
Big not convenient property and danger.Visually impaired people must be by guide capital construction (sidewalk for visually impaired people, braille mark) and auxiliary equipment (guide hand
Cane, seeing-eye dog etc.) carry out outdoor activities.But the ability of traditional standbys acquisition of information limit blind person scope of activities and
The free degree, therefore people are increasingly obvious to the demand of new blind guiding system.With wireless location technology, embedded technology and sensing
Device technology is developed rapidly, and electronic communication auxiliary equipment (ETAs) research and development cause people and greatly pay close attention to.Based on different technologies
Platform, there has been proposed various navigation system for visually impaired people, including wearable navigation equipment, intelligent blind-guiding walking stick and anti-
Deviation system etc., and possess the functions such as sidewalk for visually impaired people identification, detection of obstacles, Objective extraction, path navigation, offset correction mostly.
In recent years, smart mobile phone be in rapid growth trend, and smart mobile phone additionally provide more stable running environment,
More outstanding operating system and more easily development environment.In addition, smart mobile phone is generally integrated with various perceptrons, such as shine
Camera, accelerometer, gyroscope, direction sensor and global positioning system etc. so that smart mobile phone possesses location navigation and ring
The function that border perceives.More it is essential that smart mobile phone price tends to popular, its cost advantage can break traditions guide equipment
Cost barrier, therefore, promote and had important practical significance based on the blind guiding system of smart mobile phone, with Digital Image Processing skill
Developing rapidly for art, the main way for realizing that sidewalk for visually impaired people recognizes is increasingly becoming by the essential information of image procossing acquisition sidewalk for visually impaired people, blind
Road recognizer is become for the focus of research, and vast scholar compares in-depth study.
The Chinese invention patent of Application No. 201010174012.8 is disclosed one kind and is carried out using computer vision technique
Sidewalk for visually impaired people and the method for crossing real-time detection, train grader to obtain sidewalk for visually impaired people and crossing using affine Transform Model
Sample set carries out the detection and treatment of target.The Chinese invention patent of Application No. 201110200597.0 discloses one kind and is based on
Computer vision blindman outdoor support system, image is gathered using binocular camera, and carrying out algorithm design by embedded platform comes
Assisting blind carries out the understanding of roadway scene.
Above-mentioned existing blind guiding system and sidewalk for visually impaired people image recognition algorithm there are problems that a certain degree of：(1) what is needed is auxiliary
Help excessive awareness apparatus, complex operation, expensive, development platform not intelligent enough；(2) blind person cannot be assisted effectively to understand week
Collarette border, such as accurate recognition of sidewalk for visually impaired people obstacle；(3) algorithm about image recognition is numerous, but various recognizers are in sidewalk for visually impaired people
Identification domain variability does not have complete recognition effect comparison and analysis.
The content of the invention
In order to solve the above problems, the present invention provides a kind of sidewalk for visually impaired people identifying system based on Android platform, solves guide
Complex operation, the auxiliary awareness apparatus for needing are excessive in system, and expensive, development platform is not intelligent enough, and sidewalk for visually impaired people detection is inadequate
Accurately, the numerous defect of algorithm.
The technical solution adopted by the present invention is：A kind of sidewalk for visually impaired people identifying system based on Android platform, including IMAQ
Module, image processing module and voice supplementary module, wherein image capture module are used to realize the Real-time Collection of sidewalk for visually impaired people image；Figure
As processing module is used to carry out classification treatment to the sidewalk for visually impaired people image for detecting, and result is sent to voice supplementary module；
Voice supplementary module reports real time status information and guiding blind person's walking of sidewalk for visually impaired people according to the result of image procossing.
Further, described image processing module includes aberration sidewalk for visually impaired people identification module, no color differnece sidewalk for visually impaired people identification module, blind
Road flex point detection module and detection of obstacles module；
It is described to have aberration sidewalk for visually impaired people identification module according to there is otsu Threshold segmentations under aberration sidewalk for visually impaired people image preprocessing → Lab space
The image processing flow of → Canny rim detections → accumulative Hough transform carries out the identification for having aberration sidewalk for visually impaired people；
No color differnece sidewalk for visually impaired people identification module is equal according to no color differnece sidewalk for visually impaired people image preprocessing → extraction LBP characteristic vectors → K-
The image processing flow of value cluster analysis → padding → Canny rim detections → accumulative Hough transform is carried out；
The sidewalk for visually impaired people flex point detection module is examined according on the basis of having aberration sidewalk for visually impaired people recognizer with reference to Harris angle points
Method of determining and calculating is realized；
The detection of obstacles module according to the analysis of sidewalk for visually impaired people image preprocessing (contain barrier) → K- averages color cluster →
The image processing flow of Canny rim detections → accumulative Hough transform carries out the detection of barrier.
It is described to there is aberration sidewalk for visually impaired people image preprocessing to change and extract b component maps including first carrying out RGB-Lab color spaces, its
It is secondary to carry out medium filtering and image expansion operation generation pretreatment image for the b component maps extracted, then scheme for pretreatment
As carrying out otsu Threshold segmentations generation image segmentation figure, finally carrying out Canny rim detections and accumulative Hough transformation carries out sidewalk for visually impaired people
The lookup of boundary straight line and mark.
No color differnece sidewalk for visually impaired people image preprocessing includes first carrying out image gray processing and gaussian filtering, is calculated secondly by LBP
Son extracts characteristic vector and realizes that pretreatment image is split then in conjunction with K- mean cluster analysises, then by corresponding padding
Eliminate erroneous judgement block, finally using Canny rim detections and accumulative Hough transformation carry out no color differnece sidewalk for visually impaired people boundary straight line lookup and
Mark.
Barrier sidewalk for visually impaired people image preprocessing includes that first carrying out RGB-Lab color spaces changes and extract b component maps, so
Medium filtering generation pretreatment image is carried out in the b component maps extracted afterwards, secondly K- average face is carried out for pretreatment photo
Color cluster generation image segmentation figure, finally carrying out Canny rim detections and accumulative Hough transformation carries out looking into for sidewalk for visually impaired people boundary straight line
Look for and identify.
The real time status information that the voice supplementary module reports sidewalk for visually impaired people includes：Front of advancing is whether obstacle, if partially
From sidewalk for visually impaired people and current direct of travel.
Described image acquisition module is the camera of smart mobile phone, and described image processing module is the control of smart mobile phone
Device, the voice supplementary module is the bluetooth earphone matched with smart mobile phone.
Using above-mentioned technical scheme, with advantages below：
A kind of sidewalk for visually impaired people identifying system based on Android platform proposed by the present invention, can real-time and efficiently detect sidewalk for visually impaired people
The position of border, turning and barrier, guiding blind safety is advanced；By OpenCV by all kinds of sidewalk for visually impaired people recognizers and obstacle
Thing extraction algorithm has been applied in Android intelligent, and the sidewalk for visually impaired people for having built an integrated image Real-time Collection and treatment is known
Other system, solves current blind guiding system complex operation, expensive, development platform not enough intelligent, sidewalk for visually impaired people and detects not accurate enough,
The numerous defect of algorithm.
Brief description of the drawings
Fig. 1 is system block diagram of the invention；
Fig. 2 is image procossing framework of the invention；
Fig. 3 is the algorithm flow chart for having aberration sidewalk for visually impaired people region recognition；
Fig. 4 is the result demonstration graph for having aberration sidewalk for visually impaired people region recognition；
Fig. 5 is the algorithm flow chart of no color differnece sidewalk for visually impaired people region recognition；
Fig. 6 is the result demonstration graph of no color differnece sidewalk for visually impaired people region recognition；
Fig. 7 is the algorithm flow chart of sidewalk for visually impaired people corner detection；
Fig. 8 is the result demonstration graph of sidewalk for visually impaired people corner detection；
Fig. 9 is the algorithm flow chart of detection of obstacles；
Figure 10 is the result demonstration graph of detection of obstacles；
Figure 11 is voice cue module schematic diagram；
Figure 12 is the schematic diagram of LBP operators；
Figure 13 is Hough transformation schematic diagram.
Specific embodiment
In order that the technical problem to be solved in the present invention, technical scheme and advantage are clearer, below in conjunction with accompanying drawing and
Specific embodiment is described in detail, and description here does not mean that all masters corresponding to the instantiation stated in embodiment
Topic all refer in the claims.
As shown in figure 1, a kind of sidewalk for visually impaired people identifying system based on Android platform, including image capture module, image procossing
Module and voice supplementary module, wherein image capture module are used to realize the Real-time Collection of sidewalk for visually impaired people image；Image processing module is used
Classification treatment is carried out in the sidewalk for visually impaired people image to detecting, and result is sent to voice supplementary module；Voice supplementary module
According to the result of image procossing, real time status information and guiding blind person's walking of sidewalk for visually impaired people are reported；
Wherein, described image processing module includes aberration sidewalk for visually impaired people identification module, no color differnece sidewalk for visually impaired people identification module, sidewalk for visually impaired people and turns
Point detection module and detection of obstacles module.
As shown in Fig. 2 the treatment of image is needed by then first input picture pre-processes, after pretreatment to image
Carry out image segmentation, and image to splitting carries out rim detection, and straight line lookup is carried out after edge detection process, and according to lookup
Result, judge whether to need to carry out Corner Detection, if it is not required, then directly output result figure；Need, carry out angle point inspection
Output result figure again after survey.
As shown in figure 3, the treatment to there is aberration sidewalk for visually impaired people image, has aberration sidewalk for visually impaired people identification module to have aberration sidewalk for visually impaired people image elder generation
Pre-processed, pretreatment includes image is transformed into Lab color spaces by RGB by cvtColor functions, then passes through
CvSpilt functions extract b component color passage figures, then carry out medium filtering and image expansion to b component color passage figures again
Operation generation pretreatment image；Then otsu Threshold segmentations generation image segmentation figure is carried out for pretreatment image, is then passed through
Canny functions realize that rim detection is detected and draw sidewalk for visually impaired people boundary straight line with accumulative Hough transformation is carried out, and have finally drawn aberration
Sidewalk for visually impaired people recognition result.
As shown in figure 4, Fig. 4 is the result demonstration graph for having aberration sidewalk for visually impaired people region recognition, there is aberration sidewalk for visually impaired people to recognize by performing
The algorithm flow of module, has the recognition result of the sidewalk for visually impaired people image of aberration and the profile for having aberration sidewalk for visually impaired people image of input to have well
Uniformity.
As shown in figure 5, the treatment to no color differnece sidewalk for visually impaired people image, no color differnece sidewalk for visually impaired people identification module is by no color differnece sidewalk for visually impaired people image elder generation
Pre-processed, pretreatment includes realizing that the gray processing and GaussianBlur functions of image realize height by cvtColor functions
This filtering, realizes that pretreatment image is split, so secondly by LBP operator extractions characteristic vector then in conjunction with K- mean cluster analysises
Erroneous judgement block, dilation erosion are eliminated by corresponding padding afterwards, is finally examined using Canny rim detections and accumulative Hough transformation
Sidewalk for visually impaired people boundary straight line is surveyed and drawn, no color differnece sidewalk for visually impaired people recognition result is drawn.
As shown in fig. 6, result demonstration graphs of the Fig. 6 for no color differnece sidewalk for visually impaired people region recognition, by no color differnece sidewalk for visually impaired people identification module
The algorithm flow of execution, the result for drawing can reflect the actual conditions of current sidewalk for visually impaired people.
As shown in fig. 7, the treatment to sidewalk for visually impaired people flex point image, sidewalk for visually impaired people flex point detection module is calculated according to there is aberration sidewalk for visually impaired people to recognize
Realized with reference to Harris Corner Detection Algorithms on the basis of method；Sidewalk for visually impaired people flex point detection module will first have aberration sidewalk for visually impaired people corner view
As being pre-processed, pretreatment includes image is transformed into Lab color spaces by RGB by cvtColor functions, secondly by
CvSpilt functions extract b component color passage figures, then realize medium filtering by GaussianBlur functions, and carry out 5 times
Corrosion and 5 expansive workings generate pretreatment image, then carry out otsu Threshold segmentations, Canny functions for pretreatment image
Realize that rim detection is detected and draw straight line with accumulative Hough transformation is carried out, finally realize that angle point is examined by Harris algorithms again
Survey, draw sidewalk for visually impaired people corner detection result.
As shown in figure 8, result demonstration graphs of the Fig. 8 for sidewalk for visually impaired people corner detection, by the base for having aberration sidewalk for visually impaired people recognizer
Realized with reference to Harris Corner Detection Algorithms on plinth, the accurate detection to turning can be realized.
As shown in figure 9, the treatment to sidewalk for visually impaired people obstructions chart picture, detection of obstacles module first carries out sidewalk for visually impaired people obstructions chart picture
Pretreatment, pretreatment includes image is transformed into Lab color spaces by RGB by cvtColor functions, then by cvSpilt
Function extracts b component color passage figures, then carries out medium filtering generation pretreatment image to b component color passage figures again, its
It is secondary to carry out K- averages color cluster generation image segmentation figure for pretreatment photo, then realize that edge is examined by Canny functions
Survey and carry out accumulative Hough transformation to detect and draw straight line, finally draw sidewalk for visually impaired people detection of obstacles result.
As shown in Figure 10, Figure 10 is the result demonstration graph of detection of obstacles, and the detection module that breaks the barriers is by sidewalk for visually impaired people obstacle
Object image carries out image procossing, is finally capable of detecting when barrier present on sidewalk for visually impaired people.
As shown in figure 11, Figure 11 is voice cue module schematic diagram, and voice cue module is recognized including sidewalk for visually impaired people, corner detection
And detection of obstacles, sidewalk for visually impaired people image is acquired by the camera of smart mobile phone is sent in image capture module, Ran Houjing
Cross image processing module the sidewalk for visually impaired people for gathering is processed, recognizes, draw whether there is turning, the testing result such as barrier, and will inspection
Survey result carries out voice broadcast prompting by voice supplementary module, and the information of report includes：Front of advancing is whether obstacle, if
Deviate sidewalk for visually impaired people and current direct of travel；
Further, described image acquisition module is the camera of smart mobile phone, and described image processing module is intelligent hand
The controller of machine, the voice supplementary module is the bluetooth earphone matched with smart mobile phone.
As shown in figure 12, Figure 12 is LBP operator schematic diagrams, and classical LBP operator definitions are the square window of 3*3, with window
Mouth center pixel is threshold value, and the threshold value in figure is 92, and its adjacent 8 neighborhood territory pixel gray scale is compared with center pixel, if surrounding
Pixel value is more than center pixel value, and the surrounding pixel position is marked as 1, otherwise for 0. by this rule, 3*3 square windows
Intraoral 8 neighborhood point is converted can to become 8 bits, and this binary digit is exactly the LBP values of center pixel, LBP values
There are 256 kinds, the LBP values of center pixel reflect the texture information of the pixel peripheral region；
The mathematic(al) representation for generating image window LBP operators is as follows：
Wherein window center pixel is that its strength definition is the brightness for being defined as its neighborhood territory pixel of rectangle 8.Centered on sentence
Determine function, it is defined as follows：
Algorithm flow is as follows：
1. image is split, is divided into grid_x*grid_y block (cell), grid_x, grid_y is defaulted as
32；
2. the histogram of each cell, i.e., the frequency that each digital (decimal number LBP values) occurs then are calculated；Then it is right
The histogram is normalized；
The statistic histogram of each cell that 3. will finally obtain is attached as a characteristic vector, that is, view picture
The LBP texture feature vectors of figure.
As shown in figure 13, Figure 13 is the principle of Hough transformation, the straight line in direct coordinate system, origin to the straight line
Vertical range be r, the angle of vertical line and x-axis is θ, then this straight line is unique, can write out its linear equation, and this
Straight line polar coordinate representation is then for a bit (r, θ), it is seen that in the straight line correspondence polar coordinates in rectangular coordinate system a bit, this
The conversion for planting line to point is exactly Hough transformation.
Further, color space change and extract color component figure algorithm principle it is as follows：
Above-mentioned matrix is exactly first gone to XYZ, then switched to by XYZ by RGB to the conversion of Lab space by RGB in OpenCV
Lab, RGB turn XYZ；
Assuming that r, g, b are three passages of pixel, span is [0,255], and conversion formula is as follows：
XYZ turns Lab, and the formula of conversion is as follows：
In both the above formula, L*,a*,b*It is three values of passage of final Lab color spaces.X, Y, Z are that RGB turns XYZ
The value calculated afterwards, Xn,Yn,ZnGeneral acquiescence is 95.047,100.0,108.883.
Final to extract b component maps, by b channel values according to location of pixels, correspondence is assigned to a single channel image.
Further, image segmentation is comprised the concrete steps that：Otsu threshold values are carried out in b component maps after by pretreatment
Segmentation, otsu is a kind of maximum between-cluster variance algorithm for finding image threshold, and its algorithm steps is as follows：
1. number of each pixel in entire image in gray level is counted；
2. probability distribution of each pixel in entire image is calculated；
3. traversal search is carried out to gray level, probability between prospect background class under calculating current grayvalue；
4. threshold value corresponding with inter-class variance in class is calculated by object function；
5. with the threshold binarization image for finding.
Further, Canny rim detections principle is：
Noise is eliminated, in image pre-processing phase, it is necessary to carry out noise reduction process.Generally, filtered using Gaussian smoothing
Ripple device convolution noise reduction.
Calculate gradient magnitude and direction：
The step of herein according to Sobel filter, operates, with a pair of convolution arrays (being respectively acting on x and y directions)：
Amplitude gradient and the direction of each pixel are calculated using following equation：
And gradient direction typically takes one of 4 possible angles：0 degree, 45 degree, 90 degree, 135 degree；
Non-maxima suppression：
The step can further exclude non-edge pixels, only retain some hachures as candidate edge；
Hysteresis threshold：
Canny has used hysteresis threshold, hysteresis threshold two threshold values of needs：High threshold and Low threshold；
If 1. the amplitude of a certain location of pixels exceedes high threshold, the pixel is left edge pixel；
If 2. the amplitude of a certain location of pixels is less than Low threshold, the pixel is excluded；
If 3. the amplitude of a certain location of pixels is between two thresholds, the pixel is only being connected to one higher than high threshold
Pixel when be retained；
A bianry image is finally given, indicate whether it is a marginal point at every.
Harris Corner Theories propose corresponding angle point receptance function, are shown below：
C (i, j)=det (M)-k (trace (M))2
Wherein M is structure matrix, and k is constant factor, and normal conditions value is 0.04 to 0.06, to data in image window
Summation weighting is carried out, window center characteristic can essentially be preferably portrayed, Harris Corner Detection Algorithms realize that step is as follows：
1. convolution operation is carried out to image using level and vertical difference operator, calculates corresponding gradient, according to real symmetrical
The composition of matrix M, calculates the value of homography element；
2. smooth operation is carried out to matrix M using Gaussian function, obtains new Metzler matrix, step 1 and 2 can change order,
Gaussian filtering can also be carried out to image, then seeks the gradient magnitude in respective direction；
3. to each pixel and given neighborhood window, the characteristic value and respective function of local feature matrix of consequence M are calculated；
4. the threshold value of receptance function C is chosen, according to non-maxima suppression principle, while meeting the office in threshold value and certain neighborhood
Portion's maximum is candidate angular.
Further, image segmentation is realized in K- averages color cluster analysis, and color cluster parser flow is as follows：
1. sample matrix is created, size is identical with original image, port number is 3；
2. category label matrix is created, size is identical with original image, and port number is 1；
3. the RGB triple channel values of each pixel of original image are obtained, and enters sample matrix in order；
4. K- mean clusters are carried out to sample matrix, result is stored in category label matrix；
5. corresponding bianry image, as cluster segmentation result are created according to category label matrix；
Note：Using color clustering method under Lab space, cluster classification is 3；Can be for sidewalk for visually impaired people inside and across sidewalk for visually impaired people
Barrier be identified, but on condition that there is different the color of barrier and its region.
Further, the basic thought of K- means clustering algorithms is：By the method for iteration, each cluster centre is gradually updated
Value, until obtaining best cluster result；
Assuming that sample set is divided into c classification, comprise the following steps that：
1. the c initial center of class is suitably selected；
2. in kth time iteration, to any one sample, the sample is grouped into distance most by the distance for asking it to arrive c center
In short class.Wherein using Euclidean distance as distance criterion function；
3. such central value is updated using average；
4. for all of c cluster centre, if after being updated using iterative method 2., 3., value keeps constant (target letter
Number convergence), then iteration terminates, and otherwise continues iteration.
In specific implementation process, k mean clusters are realized using cvKMeans2 functions in OpenCV.After cluster is completed
Image pixel is identified as 0 or 1, and used as two class results, compared with other clustering algorithms, K- mean clusters have K- mean algorithms
There is the features such as algorithm is simple, and cluster speed is fast.
Further, padding is for unit is clustered therefore unavoidable when being classified with image block (cell)
Can be wrong.Cell for cluster mistake is, it is necessary to carry out further amendment operation.Padding flow is as follows：
1. the image of grid_x*grid_y sizes is created, pixel value is 0, in new figure in a pixel correspondence original image
A cell block；
2. bianry image is generated, 1 cell is designated in correspondence original image, new figure is entered as 255 in respective pixel value；
3. fill up the gap, the new figure of full figure traversal, if equal in the presence of three adjacent pixel values around a pixel, just will
The pixel is set to corresponding value；
4. an etching operation, expansive working twice, so as to eliminate region deformation are successively carried out；
5. it is artwork size by image restoring；
6. the interference region filled up in binary map, if the contour area in binary map is less than certain value, is just carried out again
Fill up operation；
7. a gaussian filtering is carried out.
Due to taking the mode of Block Cluster, serrated boundary generation is had, but the range areas of sidewalk for visually impaired people positioning is relative
Compare accurate, current sidewalk for visually impaired people region can effectively be recognized according to this recognition result, the requirement of practicality can be met.
Operation principle of the invention is：The present invention builds a basic UI manipulation page, can click on screen or profit
Self-shooting bar bluetooth controller is used, the collection of sidewalk for visually impaired people image, and the image that will be gathered are carried out using the high-definition camera of smart mobile phone
Colour space transformation is carried out, is sent be further processed with corresponding graphics processing unit afterwards；In actual use, it is considered to
Directly mobile phone cannot be manipulated to blind person, so by the whole process for shooting and processing, completed all of background program,
Only need to capable of being opened by one key before the use；Most of electricity of smart mobile phone is consumed when lighting screen, so
For blind person, the present invention extinguishes screen completely, can so save most kwh loss；In addition, in the present invention, hand is allowed
Machine camera is in detecting state all the time, and when the situation of sidewalk for visually impaired people does not occur difference, image processing module is in interval
Property detection state, once having detected peculiar generation, such as front detects sidewalk for visually impaired people flex point, sidewalk for visually impaired people interrupt or sidewalk for visually impaired people on have barrier
Hinder, then do progressive treatment immediately, while the object information for the treatment of is sent into voice supplementary module in time.
Finally it should be noted that foregoing description is the preferred embodiments of the present invention, one of ordinary skill in the art exists
Under enlightenment of the invention, on the premise of without prejudice to present inventive concept and claim, expression as multiple types can be made, this
The conversion of sample is each fallen within protection scope of the present invention.
Claims (7)
1. a kind of sidewalk for visually impaired people identifying system based on Android platform, it is characterised in that including image capture module, image procossing
Module and voice supplementary module, wherein image capture module are used to realize the Real-time Collection of sidewalk for visually impaired people image；Image processing module is used
Classification treatment is carried out in the sidewalk for visually impaired people image to detecting, and result is sent to voice supplementary module；Voice supplementary module
According to the result of image procossing, real time status information and guiding blind person's walking of sidewalk for visually impaired people are reported.
2. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 1, it is characterised in that the figure
As processing module includes aberration sidewalk for visually impaired people identification module, no color differnece sidewalk for visually impaired people identification module, sidewalk for visually impaired people flex point detection module and barrier
Detection module；
It is described have aberration sidewalk for visually impaired people identification module according to have otsu Threshold segmentations under aberration sidewalk for visually impaired people image preprocessing → Lab space →
The image processing flow of Canny rim detections → accumulative Hough transform carries out the identification for having aberration sidewalk for visually impaired people；
No color differnece sidewalk for visually impaired people identification module is poly- according to no color differnece sidewalk for visually impaired people image preprocessing → extraction LBP characteristic vectors → K- averages
The image processing flow of alanysis → padding → Canny rim detections → accumulative Hough transform is carried out；
The sidewalk for visually impaired people flex point detection module is calculated according on the basis of having aberration sidewalk for visually impaired people recognizer with reference to Harris Corner Detections
Method is realized；
The detection of obstacles module according to sidewalk for visually impaired people image preprocessing → K- average color cluster analysis → Canny rim detections →
The image processing flow of accumulative Hough transform carries out the detection of barrier.
3. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 2, it is characterised in that described to have
Aberration sidewalk for visually impaired people image preprocessing includes that first carrying out RGB-Lab color spaces changes and extract b component maps, secondly for the b for extracting
Component map carries out medium filtering and image expansion operation generation pretreatment image, then carries out otsu threshold values for pretreatment image
Segmentation generation image segmentation figure, finally carrying out Canny rim detections and accumulative Hough transformation carries out the lookup of sidewalk for visually impaired people boundary straight line
And mark.
4. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 2, it is characterised in that the nothing
Aberration sidewalk for visually impaired people image preprocessing includes first carrying out image gray processing and gaussian filtering, secondly by LBP operator extraction characteristic vectors,
Realize that pretreatment image is split then in conjunction with K- mean cluster analysises, erroneous judgement block is then eliminated by corresponding padding, most
Lookup and the mark of no color differnece sidewalk for visually impaired people boundary straight line are carried out using Canny rim detections and accumulative Hough transformation afterwards.
5. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 2, it is characterised in that the barrier
Hindering thing sidewalk for visually impaired people image preprocessing includes that first carrying out RGB-Lab color spaces changes and extract b component maps, then in b points for extracting
Medium filtering generation pretreatment image is carried out in spirogram, secondly K- averages color cluster generation image is carried out for pretreatment photo
Segmentation figure, finally carrying out Canny rim detections and accumulative Hough transformation carries out lookup and the mark of sidewalk for visually impaired people boundary straight line.
6. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 1, it is characterised in that institute's predicate
The real time status information that sound supplementary module reports sidewalk for visually impaired people includes：Front of advancing is whether obstacle, if deviate sidewalk for visually impaired people and current
Direct of travel.
7. a kind of sidewalk for visually impaired people identifying system based on Android platform according to claim 1, it is characterised in that the figure
As acquisition module is the camera of smart mobile phone, described image processing module is the controller of smart mobile phone, the voice auxiliary
Module is the bluetooth earphone matched with smart mobile phone.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611184275.0A CN106709518A (en) | 2016-12-20 | 2016-12-20 | Android platform-based blind way recognition system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611184275.0A CN106709518A (en) | 2016-12-20 | 2016-12-20 | Android platform-based blind way recognition system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106709518A true CN106709518A (en) | 2017-05-24 |
Family
ID=58939349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611184275.0A Pending CN106709518A (en) | 2016-12-20 | 2016-12-20 | Android platform-based blind way recognition system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106709518A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107622502A (en) * | 2017-07-28 | 2018-01-23 | 南京航空航天大学 | The path extraction of robot vision leading system and recognition methods under the conditions of complex illumination |
| CN109602585A (en) * | 2018-11-30 | 2019-04-12 | 西安工程大学 | A kind of glasses for guiding blind and its anti-collision early warning method |
| CN109726681A (en) * | 2018-12-29 | 2019-05-07 | 北京航空航天大学 | It is a kind of that location algorithm is identified based on the blind way of machine learning identification and image segmentation |
| CN109766947A (en) * | 2019-01-16 | 2019-05-17 | 李�浩 | A kind of self-adapting intelligent image processing system |
| CN109931946A (en) * | 2019-04-10 | 2019-06-25 | 福州大学 | Blind visual range-finding navigation method based on Android intelligent |
| CN110070514A (en) * | 2019-04-01 | 2019-07-30 | 孙鑫宇 | A kind of FPGA embedded vision recognition methods of blind way state and portable device |
| CN110470308A (en) * | 2019-09-06 | 2019-11-19 | 北京云迹科技有限公司 | A kind of obstacle avoidance system and method |
| CN110659577A (en) * | 2019-08-26 | 2020-01-07 | 杭州电子科技大学 | Blind road obstacle detection method for smart phone platform |
| CN111783557A (en) * | 2020-06-11 | 2020-10-16 | 北京科技大学 | Wearable blind guiding equipment based on depth vision and server |
| CN113034403A (en) * | 2021-04-16 | 2021-06-25 | 重庆飞唐网景科技有限公司 | Working method for denoising medical image data through Gaussian filtering |
| CN113160167A (en) * | 2021-04-16 | 2021-07-23 | 重庆飞唐网景科技有限公司 | Medical image data extraction working method through deep learning network model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101336856A (en) * | 2008-08-08 | 2009-01-07 | 西安电子科技大学 | Information acquisition and transfer method of auxiliary vision system |
| CN102164344A (en) * | 2011-02-21 | 2011-08-24 | 中国华录集团有限公司 | Navigation mobile phone for the blind |
-
2016
- 2016-12-20 CN CN201611184275.0A patent/CN106709518A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101336856A (en) * | 2008-08-08 | 2009-01-07 | 西安电子科技大学 | Information acquisition and transfer method of auxiliary vision system |
| CN102164344A (en) * | 2011-02-21 | 2011-08-24 | 中国华录集团有限公司 | Navigation mobile phone for the blind |
Non-Patent Citations (1)
| Title |
|---|
| 汤智超等: "导盲机器人的交通标志视觉识别技术研究", 《计算机技术与发展》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107622502A (en) * | 2017-07-28 | 2018-01-23 | 南京航空航天大学 | The path extraction of robot vision leading system and recognition methods under the conditions of complex illumination |
| CN107622502B (en) * | 2017-07-28 | 2020-10-20 | 南京航空航天大学 | Path extraction and identification method of visual guidance system under complex illumination condition |
| CN109602585B (en) * | 2018-11-30 | 2022-02-22 | 西安工程大学 | Blind guiding glasses and anti-collision early warning method thereof |
| CN109602585A (en) * | 2018-11-30 | 2019-04-12 | 西安工程大学 | A kind of glasses for guiding blind and its anti-collision early warning method |
| CN109726681A (en) * | 2018-12-29 | 2019-05-07 | 北京航空航天大学 | It is a kind of that location algorithm is identified based on the blind way of machine learning identification and image segmentation |
| CN109766947A (en) * | 2019-01-16 | 2019-05-17 | 李�浩 | A kind of self-adapting intelligent image processing system |
| CN109766947B (en) * | 2019-01-16 | 2023-07-18 | 李�浩 | Self-adaptive intelligent image processing system |
| CN110070514A (en) * | 2019-04-01 | 2019-07-30 | 孙鑫宇 | A kind of FPGA embedded vision recognition methods of blind way state and portable device |
| CN109931946A (en) * | 2019-04-10 | 2019-06-25 | 福州大学 | Blind visual range-finding navigation method based on Android intelligent |
| CN110659577A (en) * | 2019-08-26 | 2020-01-07 | 杭州电子科技大学 | Blind road obstacle detection method for smart phone platform |
| CN110470308A (en) * | 2019-09-06 | 2019-11-19 | 北京云迹科技有限公司 | A kind of obstacle avoidance system and method |
| CN111783557A (en) * | 2020-06-11 | 2020-10-16 | 北京科技大学 | Wearable blind guiding equipment based on depth vision and server |
| CN111783557B (en) * | 2020-06-11 | 2023-08-15 | 北京科技大学 | Wearable blind guiding equipment based on depth vision and server |
| CN113160167B (en) * | 2021-04-16 | 2022-01-14 | 深圳市铱硙医疗科技有限公司 | Medical image data extraction working method through deep learning network model |
| CN113160167A (en) * | 2021-04-16 | 2021-07-23 | 重庆飞唐网景科技有限公司 | Medical image data extraction working method through deep learning network model |
| CN113034403A (en) * | 2021-04-16 | 2021-06-25 | 重庆飞唐网景科技有限公司 | Working method for denoising medical image data through Gaussian filtering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106709518A (en) | Android platform-based blind way recognition system | |
| CN106821694B (en) | A kind of mobile blind guiding system based on smart phone | |
| Yuan et al. | Robust lane detection for complicated road environment based on normal map | |
| US10445602B2 (en) | Apparatus and method for recognizing traffic signs | |
| CN107506760A (en) | Traffic signals detection method and system based on GPS location and visual pattern processing | |
| Li et al. | Road detection algorithm for autonomous navigation systems based on dark channel prior and vanishing point in complex road scenes | |
| CN106980852A (en) | Based on Corner Detection and the medicine identifying system matched and its recognition methods | |
| CN104915642B (en) | Front vehicles distance measuring method and device | |
| CN107977639A (en) | A kind of face definition judgment method | |
| CN103295009B (en) | Based on the license plate character recognition method of Stroke decomposition | |
| CN112733711B (en) | Remote sensing image damaged building extraction method based on multi-scale scene change detection | |
| CN111462140B (en) | Real-time image instance segmentation method based on block stitching | |
| CN108537143B (en) | A kind of face identification method and system based on key area aspect ratio pair | |
| CN104463138A (en) | Text positioning method and system based on visual structure attribute | |
| CN105046206A (en) | Pedestrian detection method and apparatus based on moving associated prior information in videos | |
| CN114037966A (en) | High-precision map feature extraction method, device, medium and electronic equipment | |
| CN115049954B (en) | Target identification method, device, electronic equipment and medium | |
| CN106650735B (en) | A kind of LED character automatic positioning recognition methods | |
| CN111950345A (en) | Camera identification method and device, electronic equipment and storage medium | |
| KR101967858B1 (en) | Apparatus and method for separating objects based on 3D depth image | |
| Chen et al. | Image segmentation based on mathematical morphological operator | |
| CN104715250A (en) | Cross laser detection method and device | |
| Kim | Detection of traffic signs based on eigen-color model and saliency model in driver assistance systems | |
| CN107463934A (en) | A kind of tunnel slot detection method and device | |
| CN111860084A (en) | Image feature matching and positioning method and device and positioning system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170524 |
|
| RJ01 | Rejection of invention patent application after publication |