• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    PURPOSE: A method for image processing of a flow is provided to optimize the determination of identity of particles to track for performing precise image processing of the flow regardless of intervals between particles. CONSTITUTION: A method for image processing of a flow includes the steps of obtaining continuous flow image for a predetermined time period(t1,t2), determining same particles among continuous images by using a normal distribution prediction method, and computing moving velocities of the same particles per time and position according to a passing route, wherein the same particles are determined by probability estimation after limiting surrounding particles(b) of a particle(a) to track in the continuous images to a predetermined number, and in the normal distribution prediction method, a probability determination becomes more sensitive if an interval between the particle to track and the surrounding particles becomes smaller, and particles having highest connection relation are estimated to be same particles by assigning the probability to connection relations(stream lines).
    公开号:KR20000061378A
    申请号:KR1019990010351
    申请日:1999-03-25
    公开日:2000-10-16
    发明作者:김원중
    申请人:김기삼；학교법인조선대학교；
    IPC主号:
    专利说明:

    Image Processing Method of Flow Field
    The present invention relates to an image processing method of a flow field, and more particularly to an image processing method of a flow field optimized to determine the identity of the tracking particles.
    In general, a method of displaying a fluid flow as an image and analyzing it by a qualitative analysis method is called a flow field image processing method. In the image processing of the flow field, a qualitative method of detecting only the flow of the fluid (that is, the stream line) is mainly used. On the other hand, PTV (Particle Tracking Velocimetry) (hereinafter referred to as "PTV") can also be used to find out the qualitative streamline and quantitative flow rate. The image processing method of the flow field is to analyze the flow and velocity of the air around the driving vehicle, the flow and velocity analysis of the fluid inside the fluid machine (eg engine, pump, washing machine, etc.), the fuselage and the wing of the airplane in flight It is applied to various types of air flow and speed analysis, water flow and speed analysis around the ship in operation.
    Referring to Fig. 1, there is shown a diagram for explaining a conventional flow field image processing method.
    A continuous flow field image is obtained for a period of time. (Step 1) Distribute the fine particles (eg Al powder) in the flow field to visualize the flow of fluid as the movement of the fine particles. At this time, the distribution of particles visualized in one frame is shown in FIG. 2. Thus, by dividing the flow field image into a plurality of frames at predetermined time intervals from the visualized flow field, a continuous image is obtained as shown in Fig. 1A.
    Estimate the same particle between successive images. (Second Step) As shown in Fig. 1B, the same particles between successive images are estimated from successive images at predetermined intervals for a predetermined time t1 to t4. For example, if you described example, if the same report by applying a wire that meets the wire of t1 when the particle at the time t2 (P t2), the particles of t1 when (P t1) and the particle (P t2) of t2 when the same Estimate with particles. Then, if the same report by applying a wire that meets the wire of t2 when the particle (P t3) of time t3, the particles (P t2) and the particle (P t3) of t3 when teeth t2 when the estimated in the same particles Done. When it is in the t4 particle (P t4 ) having a satisfactory streamline by the same method, it is reasonable to estimate the particles (P t1 to P t4 ) from t1 to t4 as the same particle.
    Track same particle. (Third Step) As shown in Fig. 1C, the movement path of the particles estimated to be the same particles is tracked.
    Calculate the velocity of each particle. (Fourth Step) As shown in FIG. 1D, the moving speed for each particle P t1 to P t4 according to the moving path of the particles is calculated for each time and for each position. Image processing of the flow field is performed by the first to fourth steps.
    On the other hand, in the conventional image processing method of the flow field after estimating the same particles according to the time and position of the specific particles, it is to track the estimated movement path of the particles. However, in the conventional image processing method of the flow field, when the spacing of particles is dense or vice versa, it is difficult to estimate the same particles, which leads to a problem that it is difficult to accurately image the flow field. Accordingly, there is an urgent need for an image processing method for a new flow field that can easily estimate the same particles.
    Accordingly, it is an object of the present invention to provide an image processing method of a flow field optimized to determine the identity of the tracking particles.
    1 is a diagram for explaining a conventional flow field image processing method.
    2 shows the distribution of particles visualized in one frame.
    3 to 5 are diagrams for explaining the principle of the flow field image processing method according to the present invention.
    6 is a view for explaining the discrimination method of the same particle.
    7 is a view showing a flow field image processing method according to the present invention according to the procedure.
    8 to 9 are diagrams showing the distribution of particles represented by superimposing four consecutive frames of images.
    10 is a view showing a distribution of particles output using the flow field image processing method according to the present invention.
    In order to achieve the above object, the image processing method of the flow field according to the present invention comprises the steps of obtaining a continuous flow field image for a predetermined time, and determining the same particles between the continuous images using a normal distribution prediction method, Comprising a step of calculating the movement speed by time, location according to the movement path of the particles.
    Other objects and features of the present invention other than the above object will become apparent from the description of the embodiments with reference to the accompanying drawings.
    With reference to Figures 3 to 10 will be described a preferred embodiment of the present invention.
    3 to 5 will be described with reference to the principle of the image processing method of the flow field according to the present invention.
    The image processing method of the flow field according to the present invention will increase the measurable dynamic range and accuracy. This will be described in detail with reference to FIG. 3. a t1, assuming a matching particle of the particles to a t2 particles, b t2 particles enables you think that this assumption is similar to the b speed of t1 particles in the vicinity of a certain pertinent a t1 particles to see if with a t1 particles The probability of a t1 particle's velocity hypothesis is increased by checking whether the By carrying out this process in the same way for the number of particles b t1 t1 around a particle it is determined to fit the speed of a home t1 particles.
    On the other hand, if the matching of the particle assumed to be a particle t1 a t2 particles as shown in Figure 4 the speed of a particle t1 is assumed to be V1. This assumption is satisfied for a t1 select b t1 particle around the particle to the assumed that by the principle of the viscosity of the continuous a t1 particles and b t1 particle velocity is similar to the assumption of b t1 particle velocity V1 to determine is right Consider the effects of b t2 and b ' t2 particles in determining whether there are b t2 particles available. Since the particles b t2 b speed Vb1 of the case to determine the matching of the grain particles and t1, b 't2 particles a b When comparing the speed Vb'1 when deciding a matching of the particle particle Vb1 t1 <t2 particles Vb'1 b Is likely to be a matching particle and has a higher probability of determining the velocity assumption V1 of the a t1 particle. Accordingly, the smaller Vb1 has a greater effect on the probability of determining the speed.
    Herein, the role of R, which is a distance between the a t1 particle and the a t2 particle, will be described with reference to FIG. 5. In other words, it is intended to know how the spacing between a t1 particles and surrounding b t1 particles affects the velocity determination. The smaller R is, the more similar the velocity of the a t1 particle and the velocity of the a t2 particle are. In this case, as shown in (a) of FIG. 6, when R is small, the probability determination is sensitive, and as R is larger, the insensitivity is lower. In other words, R is a higher value in both the rate of other b t1 particles a t1 particles and b t2 Even though it has a particle R is small b t1 particles have a t1 particle velocity determine the probability of the more b t1 particle R is greater same Will be added. Therefore, the smaller the area that assumes similar flow when determining the velocity of a t1 particles, the more accurate it can be determined.However, in areas where the distribution of particles is sparse, it is an obstacle to probability determination. In this region, the number of b t1 particles closest to the a t1 particle (for example, four) is extracted within the region, and the particles determine the speed of the a t1 particle. That is, to optimize the particle search area, the number of adjacent particles is limited to a predetermined number (for example, four), and the same particle discrimination method is used using a normal distribution that is a stochastic estimation as shown in FIG. Will be decided. In this case, a normal distribution prediction technique is introduced into the particles around the tracking particles to verify the identity of the tracking particles. In this case, the probability function is shown in Equation 1.
    Here, 0 x 100, m is 50, is the standard deviation, 2 means variance.
    Referring to FIG. 7, there is shown a diagram for explaining an image processing method of a flow field according to the present invention.
    A continuous flow field image is obtained for a period of time. (Step 11) Distribute the fine particles (eg Al powder) in the flow field to visualize the flow of fluid as the movement of the fine particles. At this time, the continuous image is divided by dividing the flow field image into a plurality of frames (for example, two) as shown in FIG. 7A from the visualized flow field at a predetermined time interval t. You can get it. In this case, the continuous image is shown in Fig. 7B. 8 and 9 show the distribution of particles obtained in the actual flow field. FIG. 8 is a diagram showing overlapping images of four consecutive frames, and FIG. 9 is a diagram illustrating particle flow information of a last frame in a large circle in FIG.
    Normal particle estimation is used to estimate the same particles between successive images. (Twelfth Step) As shown in Fig. 7C, the same particles between successive images are estimated for a predetermined time t1 to t2. In this case, the normal distribution prediction technique is applied as shown in FIG. 7 (d) to increase the estimated probability of the same particles. At this time, the probability is assigned to the linking relationship (ie, the streamline) to estimate the particles having the highest linking relationship as the same particle.
    Calculate the movement speed by time and location according to the movement path of the same particles. (Thirteenth Step) As shown in FIG. 1D, the moving speed for each particle P t1 to P t4 according to the moving path of the particles is calculated for each time and for each position. This process results in flow field image processing. At this time, the distribution form of particles by the image processing method of the flow field according to the present invention is shown in FIG. In the image processing method of the flow field according to the present invention, the particle tracking capability is improved by applying the normal distribution prediction technique to the estimation of the same particles, thereby making it possible to accurately image the flow field.
    As described above, the image processing method of the flow field according to the present invention has an advantage that the image processing of the flow field can be precisely performed.
    Those skilled in the art will appreciate that various changes and modifications can be made without departing from the technical spirit of the present invention. Therefore, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification but should be defined by the claims.
    权利要求:
    Claims (4)
    [1" claim-type="Currently amended] In the image processing method of the flow field to represent the flow of the fluid in the image analysis,
    Obtaining a continuous flow field image for a period of time,
    Judging the same particles between the consecutive images using a normal distribution prediction technique;
    Comprising the step of calculating the movement speed by time, location according to the movement path of the same particles.
    [2" claim-type="Currently amended] The method of claim 1,
    The same particle determination step,
    And limiting the surrounding particles of the tracked particles of the continuous image to a predetermined number and determining the same particles by probabilistic estimation.
    [3" claim-type="Currently amended] The method of claim 2,
    In the normal distribution prediction method, if the distance between the tracer and its surrounding particles is small, the probability determination is sensitive, and the larger the distance between the tracer and its surrounding particles is less sensitive to the probability determination.
    [4" claim-type="Currently amended] The method of claim 3, wherein
    And the probability determination assigns to the linking relationship to estimate particles having the highest linking relationship as the same particles.
    类似技术:
    公开号 | 公开日 | 专利标题
    US10657815B2|2020-05-19|Vehicular information systems and methods
    CN104657740B|2020-01-03|Method and device for segmenting an occupancy grid of an environmental model
    US20170013409A1|2017-01-12|Systems and methods to determine a location of a mobile device
    US9891057B2|2018-02-13|Information processing device, computer readable storage medium, and map data updating system
    CN106323308B|2019-02-05|It pre-processes the method for road data, drive road vehicle and its control method certainly
    Lin et al.1999|Experimental study of real-time bus arrival time prediction with GPS data
    Fathy et al.1998|A window-based image processing technique for quantitative and qualitative analysis of road traffic parameters
    Huang et al.1994|Automatic symbolic traffic scene analysis using belief networks
    Petty et al.1998|Accurate estimation of travel times from single-loop detectors
    KR100446636B1|2004-09-04|Apparatus and method for measuring ego motion of autonomous vehicles and 3D shape of object in front of autonomous vehicles
    US7027615B2|2006-04-11|Vision-based highway overhead structure detection system
    Schubert et al.2011|Empirical evaluation of vehicular models for ego motion estimation
    EP2806411B1|2016-05-18|Driving model generation device, driving model generation method, driving evaluation device, driving evaluation method, and driving support system
    DE102012201472A1|2013-08-01|Method for providing parking information on free parking spaces
    US6430303B1|2002-08-06|Image processing apparatus
    WO2015134434A1|2015-09-11|Traffic flow rates
    Danescu et al.2009|Probabilistic lane tracking in difficult road scenarios using stereovision
    EP2709064B1|2019-06-26|Image processing to derive movement characteristics for a plurality of queue objects
    US20050008194A1|2005-01-13|Apparatus and method for image processing
    CN104282020A|2015-01-14|Vehicle speed detection method based on target motion track
    EP2107392A1|2009-10-07|Object recognition system for autonomous mobile body
    TW201324383A|2013-06-16|Method and apparatus for video analytics based object counting
    DE102004032118B4|2006-09-21|Object recognition method for vehicles
    CN103324977B|2016-03-30|A kind of destination number detection method and equipment
    Buczko et al.2016|How to distinguish inliers from outliers in visual odometry for high-speed automotive applications
    同族专利:
    公开号 | 公开日
    KR100345918B1|2002-07-27|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    法律状态:
    1999-03-25|Application filed by 김기삼, 학교법인조선대학교
    1999-03-25|Priority to KR1019990010351A
    2000-10-16|Publication of KR20000061378A
    2002-07-27|Application granted
    2002-07-27|Publication of KR100345918B1
    优先权:
    申请号 | 申请日 | 专利标题
    KR1019990010351A|KR100345918B1|1999-03-25|1999-03-25|Image Processing Method of Flow Field|
    [返回顶部]