CN102323070B - Method and system for detecting abnormality of train - Google Patents
Method and system for detecting abnormality of train Download PDFInfo
- Publication number
- CN102323070B CN102323070B CN201110155792.6A CN201110155792A CN102323070B CN 102323070 B CN102323070 B CN 102323070B CN 201110155792 A CN201110155792 A CN 201110155792A CN 102323070 B CN102323070 B CN 102323070B
- Authority
- CN
- China
- Prior art keywords
- train
- image
- reference picture
- current train
- global
- 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.)
- Active
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a method and system for detecting abnormality of a train. The method comprises the following steps of: acquiring a global image of the current train; determining a number of the current train and selecting a global image corresponding to the number of the current train from a preset image library as a reference image; and aligning and comparing the global image of the current train with the reference image so as to determine a region of the global image of the current train, which is inconsistent with the reference image, as an abnormal region of the train. According to the scheme provided by the embodiment of the invention, the detection of an actual vehicle is converted into the analysis of the image, thereby automatic comparison can be carried out by adopting a computer-assisted mode and the problems of low efficiency and high possibility of leakage detection caused by mainly depending manual labor in the prior art are solved.
Description
Technical field
The present invention relates to technical field of traffic transportation, relate to a kind of train abnormality detection method and system in particular.
Background technology
Transportation by railroad is large with freight volume, quick, security high reliability, in occupation of the position of outbalance in technical field of traffic transportation.
And train is as the core of transportation by railroad, the comprehensive, accurate, quick of its abnormality detection is vital for safety of railway traffic ensures.
But present train abnormality detection mode mainly staff is rule of thumb investigated, and this mode at least exists following shortcoming:
1, train (comprising lorry, passenger vehicle, motor train unit and other type column cars) forms complex structure, and tiny parts are more, adopts manual detection mode to there is inefficiency and easily undetected problem;
2, the factors such as method of operation all can add the difficulty of manual detection, reduce further work efficiency and further increases undetected probability:
For motor-car, present platform height is higher, causes the hidden number of components of train to increase; Its composition structure and traditional train difference are comparatively large, the more difficult normal condition form remembeing each parts of upkeep operation personnel; In addition, after warehouse-in, the interior outer rim of its wheel, tread, wheel rim block due to rail and bogie structure, there is vision dead zone.In addition, motor train unit operation characteristic is: one stands erectly reaches, the dwell time is short and long routing runs, and these features cause it to be overhauled to manually midway.
Therefore need to utilize computing machine automatic abnormality detection mode indirect labor to detect, reduce work difficulty and increase work efficiency.In addition, because Train Parts number is a lot of, fault type is difficult to calculate, and existing fault detection method is difficult to set up accurate and complete fault model, so, described train fault detection method to be noted abnormalities region by global image comparison, and the mode of the further recognition and verification of binding key fault is to abnormal area classification, classification, and legend alarm.
Summary of the invention
In view of this, the invention provides a kind of train abnormality detection method and system, in order to solve inefficiency that prior art manual detection mode brings and easily to occur undetected problem.
For achieving the above object, the invention provides following technical scheme:
A kind of train abnormality detection method, comprising:
Obtain current train global image;
Determine current train license number, in pre-set image storehouse, select the global image corresponding with current train license number as reference image;
Described current train global image is carried out aliging comparison with described reference picture, to determine that the inconsistent region of described current train global image and described reference picture is for train abnormal area.
Preferably, said method also comprises:
Described train abnormal area is carried out legend mark.
Preferably, in said method, described legendization mark of being carried out by described train abnormal area is specially:
Described train abnormal area is carried out classification according to the order of severity, and wherein severely subnormal region utilizes conventional fault detection method to carry out the classification of emphasis Fault Identification, and different brackets, different classes of abnormal area indicate with different colours or shape and show.
Preferably, said method also comprises:
According to the grade setting standard preset, determine the rank of abnormal area;
The abnormal area abnormal rank being greater than to pre-determined threshold carries out legend display.
Preferably, said method also comprises:
After receiving user and cancelling the instruction of abnormal area, in described current global image, cancel the abnormal area that described user specifies.
Preferably, said method also comprises, by described current train global image stored in described pre-set image storehouse.
Preferably, in said method, described acquisition current train image is specially: utilize the mode of linear array or face battle array imaging to obtain current train global image.
Preferably, in said method, the described reference picture corresponding with current train license number is:
The image of the normal train pre-set, or,
At the image with the most contiguous same the car passed through of current time, or,
At multiple global images of same the car passed through with current time vicinity, or,
The reference picture merging or obtain after statistical study is carried out by multiple global images of described same the car passing through with current time vicinity.
Preferably, in said method, described described current train global image and described reference picture are carried out aliging comprising:
For face system of battle formations picture, adopt the mode of global registration;
For linear array images, detect current train speed, and inquiry obtains train speed corresponding to described reference picture;
When the difference of the train speed corresponding with described reference picture when described current train speed is less than default value, the mode of unique point global registration is utilized to align to described current train image and described reference picture, otherwise, in local registration mode, described current train image and described reference picture are alignd.
Preferably, in said method, the mode of described unique point global registration comprises:
SIFT/SURF method is utilized to try to achieve multiple unique points of described current train image and reference picture, and preserve the yardstick of each unique point and the proper vector of direction formation, utilize Euclidean distance method to find out unique point characteristic of correspondence point in described reference picture of current train image respectively, form same place pair;
Erroneous point pair is rejected according to projective rejection according to RANSAC algorithm;
Determine that same place that RANSAC retains is to the coordinate transform mapping relations of correspondence;
Interpolation arithmetic is carried out to reference picture.
Preferably, in said method, the mode of described local registration comprises:
SIFT/SURF method is utilized to try to achieve multiple unique points of described current train image and reference picture, and preserve the yardstick of each unique point and the proper vector of direction formation, utilize Euclidean distance method to find out unique point characteristic of correspondence point in described reference picture of current train image respectively, form same place pair;
The coordinate transform mapping relations of each same place to correspondence carry out interpolation arithmetic to reference picture to utilize projective transformation to determine successively.
Preferably, in said method, described comparison comprises:
Determine the marginal portion of current train image and reference picture, utilize marginal information to compare.
Preferably, in said method, also comprised before the marginal portion determining current train image and reference picture:
Utilize statistics with histogram method that current train image and reference picture are carried out brightness normalized.
Preferably, in said method, also comprise after the marginal portion determining current train image and reference picture: the edge of current train image and reference picture is normalized.
The present invention also discloses a kind of train abnormality detection system, comprising:
Image acquisition unit, for obtaining current train global image;
Train license number determining unit, for determining current train license number;
Reference picture chooses unit, for selecting the global image corresponding with current train license number as reference image in pre-set image storehouse;
Alignment comparing unit, for carrying out aliging comparison by described current train global image with described reference picture;
Train abnormal area determining unit, for the alignment comparison result according to described alignment comparing unit, determine that the inconsistent region of described current train global image and described reference picture is train abnormal area, and according to the order of severity of abnormal area, the classification of emphasis Fault Identification is carried out to severely subnormal.
Preferably, said system also comprises:
Legend unit, for described train abnormal area is carried out legend mark, or when there is multiple abnormal area, abnormal area rank being exceeded to pre-determined threshold carries out legend mark.
Preferably, said system also comprises:
Image stored in unit, for by described current train global image stored in described pre-set image storehouse.
Known via above-mentioned technical scheme, compared with prior art, the analysis that the scheme that the present embodiment provides will be transformed into the detection of actual vehicle image, thus computer-assisted way can be adopted automatically to compare, solve prior art and too much rely on the artificial and inefficiency that causes and easily occur undetected problem.
Accompanying drawing explanation
In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only embodiments of the 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 the accompanying drawing provided.
A kind of train abnormality detection method process flow diagram that Fig. 1 provides for the embodiment of the present invention;
The particular flow sheet of step S14 in a kind of train abnormality detection method process flow diagram that Fig. 2 provides for the embodiment of the present invention;
The particular flow sheet of step S15 in a kind of train abnormality detection method process flow diagram that Fig. 3 provides for the embodiment of the present invention;
The another kind of train abnormality detection method process flow diagram that Fig. 4 provides for the embodiment of the present invention;
Another train abnormality detection method process flow diagram that Fig. 5 provides for the embodiment of the present invention;
The structural representation of a kind of train abnormality detection system that Fig. 6 provides for the embodiment of the present invention;
The structural representation of the another kind of train abnormality detection system that Fig. 7 provides for the embodiment of the present invention;
The structural representation of another train abnormality detection system that Fig. 8 provides for the embodiment of the present invention;
The structural representation of another train abnormality detection system that Fig. 9 provides for the embodiment of the present invention;
The structural representation of another train abnormality detection system that Figure 10 provides for the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
The embodiment of the invention discloses a kind of train abnormality detection method, it mainly determines exception by the mode that image ratio is right, the detection of actual vehicle will be transformed into the analysis to image, thus computer-assisted way can be adopted automatically to compare, solve prior art and too much rely on the artificial and inefficiency that causes and easily occur undetected problem.The substance of the method is: by automatic acquisition current train global image, and determine current train license number, then, in pre-set image storehouse, select the global image corresponding with current train license number as reference image, described current global image and reference picture are compared, finally determines whether current train occurs exception.
Below by several embodiment, technical solution of the present invention is described in detail.
Embodiment one
A kind of train abnormality detection method process flow diagram that the present embodiment provides, comprises the steps:
Step S11, acquisition current train global image.
Described global image refers to the train image photographed from all angles (as bottom, top, left side, right side etc.).
The concrete mode obtaining current train global image can be: by being set in advance in multiple camera picked-ups of certain predeterminated position (bottom is referred to as sensing point) in train driving path.
Its picked-up mode can adopt the imaging mode of linear array or face battle array.
Step S12, determine current train license number.
Described current train license number specifically refers to that train respectively saves () compartment, often saves compartment and all has unique numbering, as ZE210103.
Step S13, in pre-set image storehouse, select the global image corresponding with current train type as reference image.
Described reference picture can be with any one in hypograph:
1, fixed form image
Fixed form image, the standard picture of shooting when namely train dispatches from the factory, associates vehicle number information during storage.Current vehicle image (i.e. current compartment image) and fixed form image can be compared.
2, with the normal picture of time arest neighbors during the described current train global image of shooting, described normal picture refers to determines that train is without exception after alignment comparison, and the concrete mode of described alignment comparison hereafter will introduced in detail.In general, the time is more contiguous, and reference value is maximum, by the highest for the credible result degree of itself and present image compare of analysis.
3, the image that photographs when the most contiguous time crosses certain sensing point of current train;
4, the image that photographs when the most contiguous time crosses other sensing points of current train.
5, multiple images of being photographed in contiguous current time of described current train.
Select image and the present image compare of analysis respectively of multiple time neighbour, by comparison result according to different weight fusion process, reduce false drop rate and loss.
6, based on the image that described current train obtains after statistical mathematics model at multiple images that contiguous current time is photographed.
Step S14, described current train global image to be alignd with described reference picture.
Image " alignment ", process is similar to image registration, is the prerequisite solving change detection.Image alignment is exactly find out the common scene part in two width images, and determine the conversion parameter between them, this two width image normally take in different time, different illumination conditions, based on different resolution or different angles and position etc., conversion between them can be rigid body translation, also can nonlinear affined transformation etc.
Application is detected for train fault, for face system of battle formations picture, adopt the mode of global alignment, for linear array images, " alignment " mainly solves linear deformation problem, and mainly defer to two kinds of patterns execution, one is when reference train and the current train speed of a motor vehicle are more or less the same, and directly utilizes unique point global alignment to linear array images; Two be when the reference train speed of a motor vehicle and current vehicle speed gap larger time, then to linear array images local alignment, below respectively introduce:
Pattern one: global alignment
" alignment " (above-mentioned steps S14) step in global alignment pattern mainly process flow diagram as shown in Figure 2 performs, and shown in Fig. 2, flow process comprises the following steps:
Step S21, feature point extraction.
In the present embodiment, main employing SIFT/SURF is also weighted mode keeping characteristics point in conjunction with harris isocenter detection algorithm by different weight.
Concrete mode is: first, utilize SIFT algorithm to carry out feature detection at metric space, and determine the position of key point and the yardstick residing for key point, then, use the direction character of principal direction as this point of key point neighborhood gradient, to realize the independence of operator to yardstick and direction.
SIFT algorithm can process the characteristic matching problem occurred between two width images in translation, rotation, dimensional variation, illumination variation situation, and can also possess comparatively stable characteristic matching ability to visual angle change, affined transformation to a certain extent.
SURF algorithm is the acceleration version of SIFT algorithm, and by integral image haar differentiate, the coupling that SURF algorithm completes two width objects in images under temperate conditions achieves real-time process substantially.
Harris Corner Detection Algorithm grows up on Moravec algorithm basis.The thought of Moravec Corner Detection Algorithm is: design a local detection window in the picture, when this window makes minute movement along all directions, investigate the average energy change of window, when this energy change value exceedes the threshold value of setting, just the central pixel point of window is extracted as angle point.It is detection window that Harris detection algorithm chooses Gaussian function, extracts angle point again, have certain inhibiting effect to noise to after the smoothing filtering of image.
Step S22, utilize RANSAC method reject erroneous point.
The yardstick of each unique point, direction constitutive characteristic vector can be obtained while trying to achieve unique point in step S21, utilize Euclidean distance method to find out the unique point characteristic of correspondence point in a reference image of present image respectively, form same place pair.
RANSAC algorithm comprises the sample data collection of abnormal data according to one group, calculates the mathematical model parameter of data, obtains effective sample data.In the present embodiment, first detect unique point by SIFT/SURF, then by RANSAC, error hiding is rejected.
In utilizing unique point to align, model is the projective rejection from the unique point a plane to the unique point in another one plane, reacts for projection matrix H.H is 3 × 3 matrixes comprising 8 degree of freedom, and it is minimum can be calculated by pair match point of 4 in two planes, but 3 points on same plane must not conllinear.
Step S23, calculating coordinate change mapping function.
The feature point pairs retained after RANSAC is rejected erroneous point, now requires at least to retain four same places pair, utilizes projective transformation to try to achieve coordinate conversion relation between present image and reference picture.
Step S24, interpolation arithmetic.
The projective transform matrix of trying to achieve according to step S33 with reference to image carries out coordinate transform and interpolation processing.Here consider interpolation and operation efficiency, adopt bilinear interpolation method.Mathematically, bilinear interpolation is the linear interpolation expansion of the interpolating function having Two Variables, and its core concept carries out once linear interpolation respectively in both direction.
Pattern two: local alignment
When the reference train speed of a motor vehicle and current vehicle speed gap larger time, illustrate likely when twice gathers train image, there is speed change in train, conventional overall projective transformation is difficult to try to achieve virtual borderlines relation accurately, in this case, local alignment pattern can be adopted, the particular content of this pattern comprises: utilize SIFT/SURF method to try to achieve multiple unique points of described current train image and reference picture, and preserve the yardstick of each unique point and the proper vector of direction formation, Euclidean distance method is utilized to find out unique point characteristic of correspondence point in described reference picture of current train image respectively, form same place pair, the coordinate transform mapping relations of each same place to correspondence carry out interpolation arithmetic to history image to utilize projective transformation to determine successively.
Can find out, compared with global alignment pattern, Integral Thought is consistent, but has lacked step S22.
After the relatively current train speed of a motor vehicle and the reference train speed of a motor vehicle, can determine according to comparative result the alignment thereof adopting which kind of pattern:
When the difference of the train speed corresponding with described reference picture when described current train speed is less than default value, the mode of unique point global registration (i.e. above-mentioned pattern one) is utilized to align to described current train image and described reference picture, otherwise, in the mode of local registration (i.e. above-mentioned pattern two), described current train image and described reference picture are alignd.
Described current global image after step S15, comparison alignment and reference picture.
After Current vehicle image aligns with corresponding reference picture, carry out image " comparison ", by the method for difference between detected image, find train fault generation area.
Step S16, determine that the inconsistent region of described current train global image and described reference picture is train abnormal area.
Can find out, the analysis that the scheme that the present embodiment provides will be transformed into the detection of actual vehicle image, thus computer-assisted way can be adopted automatically to compare, solve prior art and too much rely on the artificial and inefficiency that causes and easily occur undetected problem.
In said method, the image comparison process described in step S15 mainly by following flow performing, is shown in Fig. 3, comprises the steps:
Step S31, edge extracting.
Extract the fringe region of present image and reference picture respectively.
Step S32, the fringe region of described present image and reference picture to be compared.
The comparison of the fringe region of image is equal to the overall comparison of image by the present embodiment, saves comparison time, raises the efficiency.
It should be noted that, owing to being easy to, by illumination effect, directly to carry out image ratio pair, can comparison result be affected, cause wrong report, therefore in other embodiments, cut down illumination effect by following two kinds of modes in present image and reference picture imaging process:
Preconditioned pattern: utilize statistics with histogram method that present image and reference picture are carried out brightness normalized, then compare;
Real-time tupe: the edge of present image and reference picture being tried to achieve/real-time normalized of gray scale/texture, then carries out difference comparison, reaches the object that illumination effect is removed.
In addition, consider that Edge difference only represents Main Differences, though representative but lack comprehensive, therefore, can outside Edge detected difference, gray scale/textural characteristics the difference of further detection present image and reference picture, also namely based on Edge detected difference, the gray scale/textural characteristics Difference test of present image and reference picture is auxiliary thinking determination image change region.
In other examples, described abnormal area can also be identified, so that operator can understand position and the quantity of abnormal area easily, specifically as shown in Figure 4, comprise following flow process:
Step S41 ~ step S46 is substantially identical with the step S11 in above-mentioned Fig. 1 ~ step S16 content;
Step S47, described abnormal area is carried out identifying rear output.
Carry out described abnormal area to identify and specifically can adopt in various manners (such as, with color or add the forms such as frame) to carry out, come as long as abnormal area and its She region can be distinguished.
In addition, in other embodiments, classification can also be carried out to the influence degree of current train safety to abnormal area according to abnormal area position, and give different marks to the abnormal area of different stage, such as identify the highest grade abnormal area with redness, with the secondary high abnormal area of yellow mark grade.So, operating personnel can determine the state of current train quickly and intuitively according to the difference of mark.
It should be noted that, occur that abnormal area not necessarily just represent that current train occurs abnormal, in the present embodiment, judge abnormal for the right result of image ratio and what carry out warning Main Basis can be the amplitude (i.e. the amplitude of variation of two width image corresponding regions) of region of variation and the size of region of variation, be greater than a certain setting value then think that this region is exception when meeting region of variation amplitude or area.
Can find out, abnormal area shows in the mode of legend by the present embodiment, operating personnel can be given with directly perceived, image, clear and definite instruction, can improve operating personnel and identify abnormal efficiency.
Certainly, different identification is adopted to represent that the abnormal area of different brackets can allow operating personnel understand current train situation intuitively and easily, when the relation still needing operating personnel to various grade and mark is remembered, when grade classification is careful, need operating personnel to remember the relation of a lot of grade and mark, this makes troubles to the judgement of current train state to operating personnel.For this reason, the other embodiment of the present invention provides a kind of scheme, and automatically judge abnormal and report to the police when occurring abnormal, detailed process as shown in Figure 5, comprises following flow process:
Step S51 ~ step S56 is substantially identical with the step S11 in above-mentioned Fig. 1 ~ step S16 content.
Step S57, judge whether the order of severity of abnormal area exceedes pre-determined threshold, if so, enters step S58, otherwise, process ends;
Concrete, judge whether the order of severity of abnormal area exceedes pre-determined threshold and can be specifically: whether current abnormal area attaches most importance to unit failure to utilize conventional fault detection method to judge, or, judge whether the amplitude (i.e. the amplitude of variation of two width image corresponding regions) of region of variation exceedes predetermined threshold, or judge whether the area of region of variation exceedes predetermined threshold etc.
Step S58, to report to the police.
The form of reporting to the police can have a lot, and such as reported to the police by sound, image or motion (vibration), concrete alarm form belongs to prior art, does not repeat them here.
If do not occur abnormal area or abnormal area lower grade, then can think that current train is in normal condition, in this case, accessed current train global image can as the reference picture of follow-up same column train abnormality detection.That is: when described abnormal area quantity be zero or the abnormal rank of described abnormal area lower than pre-determined threshold time, by described current train global image stored in described pre-set image storehouse.
In addition, in a further embodiment, the comparison result that operating personnel can provide according to the present embodiment processes: whether artificial judgment abnormal area really breaks down, if, then place under repair, and after repairing, sent the abnormal area cancelling instruction cancellation correspondence by man-machine interaction unit, otherwise, then direct transmission by man-machine interaction unit cancels the abnormal area that correspondence is cancelled in instruction, and then by described current global image stored in pre-set image storehouse.
Certainly, can also directly by current global image stored in pre-set image storehouse, as with reference to image.In this case, as want subsequent operation personnel recall again as described in global image, abnormal investigation is carried out to the train of correspondence, and upgrades described global image according to investigation result, that is: cancel getting rid of abnormal abnormal area.
So, when calling described global image next time, can be that whether the exception can monitored above it is repaired on the other hand with reference to comparing on the one hand with normal region, or whether become more serious, to take measures to be solved.
For aforesaid each embodiment of the method, in order to simple description, therefore it is all expressed as a series of combination of actions, but those skilled in the art should know, the present invention is not by the restriction of described sequence of movement, because according to the present invention, some step can adopt other orders or carry out simultaneously.Secondly, those skilled in the art also should know, the embodiment described in instructions all belongs to preferred embodiment, and involved action and module might not be that the present invention is necessary.
For the train abnormality detection method proposed above, present invention also offers a kind of train abnormality detection system realizing the method, its a kind of concrete structure as shown in Figure 6, comprise: image acquisition unit 61, train license number determining unit 62, reference picture choose unit 63, alignment comparing unit 64 and train abnormal area determining unit 65, wherein:
Image acquisition unit 61, for obtaining current train global image;
Train license number determining unit 62, for determining current train type;
Reference picture chooses unit 63, for selecting the global image corresponding with current train type as reference image in pre-set image storehouse;
Alignment comparing unit 64, for carrying out aliging comparison by described current train global image with described reference picture;
Train abnormal area determining unit 65, for the alignment comparison result according to described alignment comparing unit, determine that the inconsistent region of described current train global image and described reference picture is train abnormal area, and according to the order of severity of abnormal area, the classification of emphasis Fault Identification is carried out to severely subnormal.
The concrete mode that image acquisition unit 61 obtains the concrete mode of current train global image, train license number determining unit 62 determines current train type, reference picture choose the process that unit 63 chooses the mode of reference picture, alignment comparing unit 64 performs alignment, comparison step, and train abnormal area determining unit 65 determines that the concrete mode of train abnormal area please refer to the content of method part above, does not repeat them here.
The another kind of concrete structure of train abnormality detection system as shown in Figure 7, comprise image acquisition unit 71, train license number determining unit 72, reference picture chooses unit 73, align comparing unit 74, train abnormal area determining unit 75 and legend unit 76, wherein:
Image acquisition unit 71, train license number determining unit 72, reference picture choose unit 73, alignment comparing unit 74 and the function of train abnormal area determining unit 75, choose unit 63, the comparing unit 64 that aligns is substantially identical with the function of train abnormal area determining unit 65 with image acquisition unit 61 above, train license number determining unit 62, reference picture.
Legend unit 76, for carrying out legend mark by described train abnormal area.
The another kind of concrete structure of train abnormality detection system as shown in Figure 8, comprise image acquisition unit 81, train license number determining unit 82, reference picture chooses unit 83, align comparing unit 84, train abnormal area determining unit 85, legend unit 86 and alarm unit 87, wherein:
Image acquisition unit 81, train license number determining unit 82, reference picture choose the function of unit 83, alignment comparing unit 84, train abnormal area determining unit 85 and legend unit 86, choose unit 73, the comparing unit 74 that aligns, train abnormal area determining unit 75 be substantially identical with the function of legend unit 76 with image acquisition unit 71 above, train license number determining unit 72, reference picture.
Alarm unit 87, for when the abnormal rank of abnormal area exceedes pre-set level, reports to the police.
The another kind of concrete structure of train abnormality detection system as shown in Figure 9, comprise image acquisition unit 91, train license number determining unit 92, reference picture choose unit 93, alignment comparing unit 94, train abnormal area determining unit 95, legend unit 96, alarm unit 97 and image stored in unit 98, wherein:
Image acquisition unit 91, train license number determining unit 92, reference picture choose the function of unit 93, alignment comparing unit 94, train abnormal area determining unit 95, legend unit 96, alarm unit 97, choose unit 83, the function of the comparing unit 84 that aligns, train abnormal area determining unit 85, legend unit 86, alarm unit 87 is substantially identical with image acquisition unit 81 above, train license number determining unit 82, reference picture.
Image stored in unit 98, for by described current train global image stored in described pre-set image storehouse, as with reference to image.
In addition, the present embodiment can also comprise man-machine interaction unit, as shown in Figure 10, the present embodiment comprises image acquisition unit 101, train license number determining unit 102, reference picture choose unit 103, alignment comparing unit 104, train abnormal area determining unit 105, legend unit 106, alarm unit 107, image stored in unit 108 and man-machine interaction unit 109, wherein:
Image acquisition unit 101, train license number determining unit 102, reference picture choose unit 103, alignment comparing unit 104, train abnormal area determining unit 105, legend unit 106, alarm unit 107, image stored in unit 108, choose unit 93, the comparing unit 94 that aligns, train abnormal area determining unit 95, legend unit 96, alarm unit 97, image be substantially identical stored in unit 98 with image acquisition unit 91 above, train license number determining unit 92, reference picture;
The current global image determined through described train abnormal area determining unit 85 is shown to operating personnel by described man-machine interaction unit 109 1 aspect, described operating personnel can also be received on the other hand to the indication information sent after the abnormal investigation of train, abnormal area in described global image is upgraded, that is: cancels getting rid of abnormal abnormal area.
The train abnormality detection system that the present embodiment provides carries out the acquisition of image, the comparison alignment of image all can be carried out automatically, without the need to too much artificial participation, improves work efficiency and reduces undetected probability.
For convenience of description, various unit is divided into describe respectively with function when describing above device.Certainly, the function of each unit can be realized in same or multiple software and/or hardware when implementing the application.
It should be noted that, in this instructions, each embodiment adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar portion mutually see.For device disclosed in embodiment, because it corresponds to the method disclosed in Example, so description is fairly simple, relevant part illustrates see method part.
In addition, it should be noted that, in this article, the such as relational terms of first and second grades and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or equipment and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or equipment.When not more restrictions, " comprising one ... " key element of limiting by statement, and be not precluded within process, method, article or the equipment comprising described key element and also there is other identical element.
To the above-mentioned explanation of the disclosed embodiments, professional and technical personnel in the field are realized or uses the present invention.To be apparent for those skilled in the art to the multiple amendment of these embodiments, General Principle as defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments.Therefore, the present invention can not be restricted to these embodiments shown in this article, but will meet the widest scope consistent with principle disclosed herein and features of novelty.
Claims (15)
1. a train abnormality detection method, is characterized in that, comprising:
The mode of linear array or face battle array imaging is utilized to obtain current train global image;
Determine current train license number, in pre-set image storehouse, select the global image corresponding with current train license number as reference image;
Described current train global image is carried out aliging comparison with described reference picture, to determine that the inconsistent region of described current train global image and described reference picture is for train abnormal area;
Described described current train global image and described reference picture are carried out aliging comprising: the global image that face battle array imaging mode is obtained, adopt the mode of global registration; For the global image that linear array imaging mode obtains, determine the speed difference of the train that current train is corresponding with reference picture, when the difference of the train speed corresponding with described reference picture when described current train speed is less than default value, the mode of unique point global registration is utilized to align to described current train global image and described reference picture, otherwise, in local registration mode, described current train global image and described reference picture are alignd.
2. the method for claim 1, is characterized in that, also comprises:
Described train abnormal area is carried out legend mark.
3. method as claimed in claim 2, is characterized in that, described legendization mark of being carried out by described train abnormal area is specially:
Described train abnormal area is carried out classification according to the order of severity, and wherein severely subnormal region utilizes conventional fault detection method to carry out the classification of emphasis Fault Identification, and different brackets, different classes of abnormal area indicate with different colours or shape and show.
4. the method for claim 1, is characterized in that, also comprises:
According to the grade setting standard preset, determine the rank of abnormal area;
The abnormal area abnormal rank being greater than to pre-determined threshold carries out legend display.
5. the method for claim 1, is characterized in that, also comprises:
After receiving user and cancelling the instruction of abnormal area, in described current global image, cancel the abnormal area that described user specifies.
6. method as claimed in claim 5, is characterized in that, also comprise, by described current train global image stored in described pre-set image storehouse.
7. the method for claim 1, is characterized in that, the described reference picture corresponding with current train license number is:
The image of the normal train pre-set, or,
At the image with the most contiguous same the car passed through of current time, or,
At multiple global images of same the car passed through with current time vicinity, or,
The reference picture merging or obtain after statistical study is carried out by multiple global images of described same the car passing through with current time vicinity.
8. the method for claim 1, is characterized in that, the mode of described unique point global registration comprises:
SIFT/SURF method is utilized to try to achieve multiple unique points of described current train image and reference picture, and preserve the yardstick of each unique point and the proper vector of direction formation, utilize Euclidean distance method to find out unique point characteristic of correspondence point in described reference picture of current train image respectively, form same place pair;
Erroneous point pair is rejected according to projective rejection according to RANSAC algorithm;
Determine that same place that RANSAC retains is to the coordinate transform mapping relations of correspondence;
Interpolation arithmetic is carried out to reference picture.
9. the method for claim 1, is characterized in that, the mode of described local registration comprises:
SIFT/SURF method is utilized to try to achieve multiple unique points of described current train image and reference picture, and preserve the yardstick of each unique point and the proper vector of direction formation, utilize Euclidean distance method to find out unique point characteristic of correspondence point in described reference picture of current train image respectively, form same place pair;
The coordinate transform mapping relations of each same place to correspondence carry out interpolation arithmetic to reference picture to utilize projective transformation to determine successively.
10. the method for claim 1, described comparison comprises:
Determine the marginal portion of current train image and reference picture, utilize marginal information to compare.
11. methods as claimed in claim 10, is characterized in that, also comprised before the marginal portion determining current train image and reference picture:
Utilize statistics with histogram method that current train image and reference picture are carried out brightness normalized.
12. methods as claimed in claim 10, is characterized in that, also comprise after the marginal portion determining current train image and reference picture: be normalized at the edge of current train image and reference picture.
13. 1 kinds of train abnormality detection systems, is characterized in that, comprising:
Image acquisition unit, obtains current train global image for utilizing the mode of linear array or face battle array imaging;
Train license number determining unit, for determining current train license number;
Reference picture chooses unit, for selecting the global image corresponding with current train license number as reference image in pre-set image storehouse;
Alignment comparing unit, for carrying out aliging comparison by described current train global image with described reference picture; Described described current train global image and described reference picture are carried out aliging comprising: the global image that face battle array imaging mode is obtained, adopt the mode of global registration; For the global image that linear array imaging mode obtains, determine the speed difference of the train that current train is corresponding with reference picture, when the difference of the train speed corresponding with described reference picture when described current train speed is less than default value, the mode of unique point global registration is utilized to align to described current train global image and described reference picture, otherwise, in local registration mode, described current train global image and described reference picture are alignd;
Train abnormal area determining unit, for the alignment comparison result according to described alignment comparing unit, determine that the inconsistent region of described current train global image and described reference picture is train abnormal area, and according to the order of severity of abnormal area, the classification of emphasis Fault Identification is carried out to severely subnormal.
14. systems as claimed in claim 13, is characterized in that, also comprise:
Legend unit, for described train abnormal area is carried out legend mark, or when there is multiple abnormal area, abnormal area rank being exceeded to pre-determined threshold carries out legend mark.
15. systems as described in claim 13 or 14, is characterized in that, also comprise:
Image stored in unit, for by described current train global image stored in described pre-set image storehouse.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110155792.6A CN102323070B (en) | 2011-06-10 | 2011-06-10 | Method and system for detecting abnormality of train |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110155792.6A CN102323070B (en) | 2011-06-10 | 2011-06-10 | Method and system for detecting abnormality of train |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102323070A CN102323070A (en) | 2012-01-18 |
| CN102323070B true CN102323070B (en) | 2015-04-08 |
Family
ID=45450859
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110155792.6A Active CN102323070B (en) | 2011-06-10 | 2011-06-10 | Method and system for detecting abnormality of train |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102323070B (en) |
Families Citing this family (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103512762B (en) * | 2012-06-29 | 2016-12-21 | 北京华兴致远科技发展有限公司 | Image processing method, device and train failure detection system |
| CN103077526B (en) * | 2013-02-01 | 2016-05-11 | 苏州华兴致远电子科技有限公司 | There is train method for detecting abnormality and the system of depth detection function |
| CN104200464B (en) * | 2014-08-19 | 2018-03-09 | 苏州华兴致远电子科技有限公司 | A kind of detection method and system for identifying that train is abnormal |
| CN104463235B (en) * | 2014-11-18 | 2019-01-18 | 中国铁道科学研究院电子计算技术研究所 | Fault recognition method and device based on EMU operation image |
| CN104634592A (en) * | 2014-12-18 | 2015-05-20 | 成都铁安科技有限责任公司 | Train running gear fault diagnosis method and train running gear fault diagnosis device |
| CN104809427B (en) * | 2014-12-31 | 2019-02-15 | 朱金良 | Failure automatic identification detection system and detection method based on 3D information |
| CN105109521B (en) * | 2015-08-20 | 2017-07-07 | 苏州华兴致远电子科技有限公司 | A kind of track train method for detecting abnormality and system |
| CN106226050B (en) * | 2016-07-15 | 2019-02-01 | 北京航空航天大学 | A kind of TFDS fault picture automatic identifying method based on convolutional neural networks |
| CN106254839A (en) * | 2016-09-30 | 2016-12-21 | 湖南中铁五新重工有限公司 | The anti-method and device of slinging of container truck |
| CN106548147B (en) * | 2016-11-02 | 2019-07-12 | 南京鑫和汇通电子科技有限公司 | A kind of quick noise robustness image foreign matter detection method and TEDS system |
| CN108225706B (en) * | 2016-12-15 | 2019-09-10 | 北京唐智科技发展有限公司 | A kind of identification wheel intensively removes the automated diagnostic method of pitted skin failure |
| CN108510019B (en) * | 2017-02-28 | 2021-02-05 | 北京天乐泰力科技发展有限公司 | Intelligent platform system based on RFID and control method thereof |
| CN107220996B (en) * | 2017-05-17 | 2019-07-09 | 武汉大学 | One kind is based on the consistent unmanned plane linear array of three-legged structure and face battle array image matching method |
| CN111902830A (en) * | 2018-03-23 | 2020-11-06 | 住友重机械工业株式会社 | Support device for construction machine |
| CN108960344A (en) * | 2018-08-03 | 2018-12-07 | 敦煌研究院 | Difference detecting method, device and the terminal device of cultural relic images |
| CN109754410B (en) * | 2019-01-03 | 2020-12-11 | 北京化工大学 | Railway vehicle carriage counting method based on machine vision |
| CN110017998B (en) * | 2019-03-26 | 2021-07-06 | 中国人民财产保险股份有限公司 | Vehicle detection method, device and equipment |
| CN110217265A (en) * | 2019-06-21 | 2019-09-10 | 中国神华能源股份有限公司 | Vehicle flat sliding detection system and method |
| CN111914901A (en) * | 2020-07-06 | 2020-11-10 | 周爱丽 | Seat arrangement uniformity measuring system and method |
| CN113781418A (en) * | 2021-08-30 | 2021-12-10 | 大连地铁集团有限公司 | Subway image anomaly detection method and system based on comparison and storage medium |
| CN116309741B (en) * | 2023-05-22 | 2023-08-11 | 中南大学 | TVDS image registration method, segmentation method, device and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1996194A (en) * | 2005-12-31 | 2007-07-11 | 清华大学 | Moving body positioning and rectifying system and its motion tracking method |
| CN101334266A (en) * | 2008-07-18 | 2008-12-31 | 北京优纳科技有限公司 | Circuit board defect off-line checking method based on large-capacity image storage technology |
| CN101616310A (en) * | 2009-07-17 | 2009-12-30 | 清华大学 | The target image stabilizing method of binocular vision system of variable visual angle and resolution |
| CN101739829A (en) * | 2009-12-03 | 2010-06-16 | 北京中星微电子有限公司 | Video-based vehicle overspeed monitoring method and system |
| CN101794439A (en) * | 2010-03-04 | 2010-08-04 | 哈尔滨工程大学 | Image splicing method based on edge classification information |
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201001338A (en) * | 2008-06-16 | 2010-01-01 | Huper Lab Co Ltd | Method of detecting moving objects |
| US8301577B2 (en) * | 2008-11-21 | 2012-10-30 | National Yunlin University Of Science And Technology | Intelligent monitoring system for establishing reliable background information in a complex image environment |
-
2011
- 2011-06-10 CN CN201110155792.6A patent/CN102323070B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1996194A (en) * | 2005-12-31 | 2007-07-11 | 清华大学 | Moving body positioning and rectifying system and its motion tracking method |
| CN101334266A (en) * | 2008-07-18 | 2008-12-31 | 北京优纳科技有限公司 | Circuit board defect off-line checking method based on large-capacity image storage technology |
| CN101616310A (en) * | 2009-07-17 | 2009-12-30 | 清华大学 | The target image stabilizing method of binocular vision system of variable visual angle and resolution |
| CN101739829A (en) * | 2009-12-03 | 2010-06-16 | 北京中星微电子有限公司 | Video-based vehicle overspeed monitoring method and system |
| CN101794439A (en) * | 2010-03-04 | 2010-08-04 | 哈尔滨工程大学 | Image splicing method based on edge classification information |
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
Non-Patent Citations (1)
| Title |
|---|
| GPS/DR/GIS技术在基于GSM-R列车监控***中应用;刘化冰等;《铁路航测》;20030330(第1期);第27-31页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102323070A (en) | 2012-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102323070B (en) | Method and system for detecting abnormality of train | |
| CN103077526B (en) | There is train method for detecting abnormality and the system of depth detection function | |
| CN103512762B (en) | Image processing method, device and train failure detection system | |
| CN109949578A (en) | A kind of illegal automatic auditing method of vehicle crimping based on deep learning | |
| CN109767626B (en) | Roadside parking inspection method and system, patrol car and server | |
| CN106548147B (en) | A kind of quick noise robustness image foreign matter detection method and TEDS system | |
| CN104573646A (en) | Detection method and system, based on laser radar and binocular camera, for pedestrian in front of vehicle | |
| CN105426864A (en) | Multiple lane line detecting method based on isometric peripheral point matching | |
| CN107273802B (en) | Method and device for detecting fault of brake shoe drill rod ring of railway train | |
| CN103593671B (en) | The wide-range lane line visible detection method worked in coordination with based on three video cameras | |
| CN104567725A (en) | Vehicle operation fault detection system and method | |
| CN104200464A (en) | Train anomaly recognition detection method and system | |
| US11475679B2 (en) | Road map generation system and road map generation method | |
| CN103021182B (en) | Method and device for monitoring motor vehicle in case of regulation violation for running red light | |
| CN204461369U (en) | Vehicle operating fault detection system | |
| CN112488995B (en) | Intelligent damage judging method and system for automatic maintenance of train | |
| CN105472355A (en) | Switch machine notch monitoring system and method based on binocular visual processing identification | |
| CN113516629A (en) | Intelligent detection system for TFDS passing operation | |
| CN105263018A (en) | Method and device for detecting superposed stripe in video image | |
| CN103413325A (en) | Vehicle speed identification method based on vehicle body feature point positioning | |
| CN105109521B (en) | A kind of track train method for detecting abnormality and system | |
| CN115527170A (en) | Method and system for identifying closing fault of door stopper handle of automatic freight car derailing brake device | |
| CN103473925B (en) | A kind of verification method of road vehicle detection system | |
| CN109558765A (en) | The detection method of automobile and lane line, device | |
| CN106951820A (en) | Passenger flow statistical method based on annular template and ellipse fitting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20160426 Address after: Dongping street Suzhou City Industrial Park in Jiangsu province 215123 No. 270 building 4A, Australia Shunchang 4C unit Patentee after: Suzhou New Vision Science and Technology Co., Ltd. Address before: 100193, 107-A, 10 Zhongguancun Software Park, 8 northeast Wang Xi Road, Beijing, Haidian District Patentee before: Beijing Huaxing Zhiyuan Technology Development Co., Ltd. |