TWI220969B - System and method for image detection - Google Patents

Abstract

The present invention is a detection system for applying the automatic reading of image signal to determine the occupation and moving status for the objects moving along a certain path and direction on a certain plane in a specific space, and particularly for the detection of road traffic status. The method includes: capturing images through video cameras and digitization to obtain the image intensity data for each pixel; processing these pixel data with a macro detection manner; using the conversion relations between a specific object and the corresponding image on the screen in a detected space, and using an image capture band defined on the image screen and a detection unit for the operation of various status indexes for a specific object; and, updating the background image for the detection space according to the change of the weather or environment to obtain various detection data.

Description

(-6 09 2 12 Printed by A7 _B7_ of the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs V. Description of the Invention (I) Field of the Invention The present invention relates to the use of real-time images to obtain the possession and movement of objects in a certain space, especially for Detection of road traffic conditions. BACKGROUND OF THE INVENTION In the past, traffic detection used a single point to obtain the traffic, speed, and occupancy data of a road section. Traditionally, most of the methods are to embed an induction coil under the road. To detect. However, a single point detection method often makes the detection data lack of accuracy due to insufficient representation, which makes it impossible to detect the traffic jam on the entire road segment. In addition, due to the induction The coil is easily damaged due to long-term rolling of the vehicle. In addition to the inability to perform its functions, the coil needs to be dug from time to time to replace or repair the coil. In addition to causing waste of manpower and material resources, it also has a considerable impact on traffic. Nissho Sumitomo Electric Industrial Co., Ltd. discloses a road traffic measurement in the Republic of China Patent Bulletin No. 349211 Method, based on the road shot by the camera and the vehicles passing on the road are drawn into a plurality of continuous portraits and animated to measure the traffic jam on the road. First set a plurality of sampling points in the above portrait and take the brightness 値. A sampling point at which a vehicle is presumed to exist at a specific time in the portrait is detected as a presence sampling point, and then a sampling point at which a vehicle is presumed to be moving at a specific time in the continuous portrait is taken as _ 4ACS / 200002TW_1 This paper standard applies Chinese National Standard (CNS ) A4 specification (210X297mm) IWI- —Γ ----- :: installed —----------- I (Please read the precautions on the back before filling this page) 1220969 A7 B7 5 2. Description of the invention (1) Make moving sampling points to detect. Using the characteristics of the vehicle's appearance is mostly rectangular, determine the moving block and the traffic jam block according to the proportion and use it to measure the traffic jam state on the road section under test at a certain time. Electric Industrial Co., Ltd. also discloses a vehicle motion detection device in US Patent No. 5,590,217. The vehicle is used as a detection object, and a two-dimensional surface is used by a camera. The image acquisition method captures the vehicle image data in the detection area and stores it in the device, and then uses the computer to process the image data to obtain road traffic conditions. The above-disclosed method and device have the following disadvantages: 1. Detection object: It can only detect vehicles on the lane, and cannot be used for the detection of other objects. 2 · Sampling method: Sampling at the same distance on the cross section of the lane as the actual distance, so the farther away from the camera, seen on the image screen The smaller the sampling point interval is. Because the sampling points are not adjacent to each other continuously arranged, there will be the disadvantage of insufficient sampling representative in the close range near the camera. Then you must smooth the data to increase the overall Data processing time. 3. Detection grid: Divide the sampling points into several grids according to the direction of the vehicle, but because the sampling points are sampled in a non-adjacent way based on the cross section of the lane, and the sampling interval is large (about 1 to 2) Seconds), which is prone to data loss. 4 · Detection perspective: Although the purpose is to detect vehicle occupancy in lane space and traffic flow, etc., the vehicle is still used for detection due to the judgment process. 4ACS / 200002TW 2 This paper standard is applicable to China National Standard (CNS) A4 specifications. (210X297 mm (please read the precautions on the back before filling this page)

1T printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 1220969 printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention According to the sampling points in it, it is determined whether there is a vehicle at the sampling point or it is moving to determine whether it is a moving block or a traffic jam block. This algorithm is too complicated, so it is not efficient. 5. Processing time: one processing cycle time is 1 to 2 seconds. If the speed is 40 km / h, the speed per second is about 11 meters / second. Therefore, the distance traveled by the vehicle in 1 second has exceeded the length of ordinary cars, which may cause misjudgment. Therefore, although it can handle road conditions with lower speeds, it cannot cover situations with higher speeds. 6. Background image initialization: No method has been proposed that can automatically achieve background image initialization. 7. Background image updating method: The calculation formula used is a variable coefficient. It is judged that the higher the vehicle occupation or traffic jam, the smaller the variable coefficient. In fact, it is not necessary to update the background when it is determined that the vehicle is in possession, and the traffic jam is not directly related to the background update. Therefore, the present invention uses more advanced technology to perform image detection by using sampling bands. In addition to improving the detection sensitivity, the obtained data has a smooth effect. By using the method provided by the present invention, in addition to reducing the data processing time, the computing efficiency can also be improved. Even if the vehicle speed in the lane is as high as 90 km / h or more, there will still be no misjudgment. 4ACS / 200002TW 3 This paper size is applicable to Chinese National Standard (CNS) A4 specification (210X297 mm) ΙΚ.ΙΓ ----- ^ Loading ----------- i (Please read the note on the back first Matters will be continued on this page) ο 6 9 20 12 A7 __B7_ V. Description of the invention (+) Abstract description (please read the precautions on the back before continuing on this page) The invention automatically reads and judges the detection of the image signal by applying the image signal Detection system for measuring the movement and possession of objects in space. A specific embodiment of the present invention is applied to the detection of road traffic conditions, but it can also be applied to security systems or other systems. The system of the present invention includes an image acquisition unit (A / D converter section), a memory storage section, a data processing section, and an input / output control section. After the image data transmitted from an image source (which can be a camera or any image source) is converted into a digital data by the image capture unit, it is temporarily stored in the memory storage unit. After the data processing unit obtains the required data from the memory storage unit, it performs various data calculations, and the object state detection result obtained by the calculation is output by the input / output control unit. If applied to traffic detection, the calculated road traffic status results can be used as a reference for road traffic data or dynamic traffic control. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economics The focus of the present invention is on the operations performed by the data processing unit on the captured images in order to obtain information that is substantially cost effective. First, the visible space on the image screen is defined as a detection area according to user requirements. At least four fixed points are selected as control points on the periphery of the detection area, and the relative coordinates of each control point are measured at the site to set the detection. Correspondence with each control in the picture _ 4ACS / 200Q02TW _4 This paper size applies Chinese National Standard (CNS) A4 specification (210 X297 mm) ο 6 9 20 2 A7 _ _B7______ 5. The image coordinates corresponding to the point of the invention (S) Calculate the geometric correspondence between the image in the detection area and the actual space. Second, within the range of space covered by the detection area, according to the main path of object movement, define more than one sampling band so that it can cover the area of the space to be detected on the image screen. The direction and number of sampling bands are determined according to the detection needs. The number of linear sampling point groups included in the sampling zone represents its detection sensitivity. If the number of groups is greater, the sensitivity will be higher, but the number of processing sampling points will be more. Third, the defined sampling band is divided into a plurality of detection units. In principle, the lengths of the individual detection units on the image screen, after being transformed into lengths in actual space through coordinates, represent equal distances. That is to say, the detection unit displayed on the image screen has the visual phenomenon that the distance is short and the distance is long as viewed by the human eye. After the detection unit is defined, the initial image background of the space to be detected on the image frame must be determined. The present invention can directly use a manually selected picture in which no moving objects appear in the detection area as the background picture to read the background image data of the detection area, but a simpler method is to use a background image extraction method to automatically read the detection data. Background of the survey area. For the intensity (intensity) of multiple images continuously captured by the same pixel over a period of time, a statistical number distribution is made, and the non-representative intensity 値 is regarded as noise and filtered. After noise filtering, the number of consecutive non-zero times is combined into a number of data blocks, and the largest sum of data is selected. The most frequently occurring image intensity is regarded as the background image intensity of the image point. After the background of all the image points in each detection unit is stable, the pre-detection operation comes to an end. __ 4ACS / 2000Q2TW _5_ This paper size applies Chinese National Standard (CNS) A4 specification (210X: 297 mm) (Please read the notes on the back before writing this page). Printed by the cooperative 1220969 A7 B7 V. Description of the invention (t) Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs Next, the measured pixel intensity is calculated. The present invention includes two main detection and calculation operations: object occupancy state detection and object movement state detection. In object occupancy detection, firstly read the pixel data in the individual detection unit, and compare the pixel data with the background of the pixel. If the difference is large, it is regarded as different from the background. In a certain detection unit, if the number of pixels different from its background accounts for the total number of pixels in the detection unit greater than a preset ratio, the detection unit is considered to be occupied by an object. So implemented sequentially. After all the detection units have completed the comparison, the input / output control unit can output the object occupancy status or perform further statistical analysis. Another main operation of the present invention is detecting the moving state of an object. The individual pixel data in each detection unit is compared with the pixel data of the same coordinates in the image frame of the previous acquisition time, and multiple consecutive data of the same image point at different times are taken into consideration. If the difference is significant, it is considered that a moving object appears in this image point during this time interval. Then compare the movement status of each image point in the same detection unit during this time interval. If it is determined that the number of image points where a moving object appears is greater than a preset ratio 値, it is considered that there is an object moving in the space to which the detection unit belongs. After all the detection units have completed the comparison, the input / output control unit outputs the moving state of the object or conducts further statistical analysis. Since the aforementioned image points that are different from the background represent object occupancy, if the coordinates of the image points representing the image points that are different from the background in continuous time are plotted, the visual movement trajectory of the object on the image screen can be obtained. If converted to the coordinates of real space, the object can be moved h--Γ—one——: one pack ----- ΓΙ1Τ: ------ ^ (Please read the precautions on the back before writing this page trivially ) 4ACS / 200002TW 6 This paper size applies to Chinese National Standard (CNS) A4 specification (210X297 mm) 1220969

Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 5. Description of the invention (q) The actual spatial trajectory of movement, and it can be determined whether it is moving or stationary. After the above calculations of the object occupancy state and the object movement state are completed, the background of the image area of the detection area is updated. Since the range to be detected may change with the change of sky and weather, the action of background update is extremely important. First, the image intensity 値 of an image point in the detection unit is compared with the initial background 値 of the image point to obtain a difference 値. If the difference 値 is larger than the preset threshold, the background update factor of the pixel is set to 1; otherwise, it is set to 0. In addition, a first background update parameter ai and a second background update parameter α 2 are set. Bring these data into the calculation formula, you can get the background update result of the pixel. The implication of this calculation formula is that if there is an object appearing in the comparison of the respective pixel data and the background pixel data at the same position, the background will not be updated, or it will only be updated at a very small proportion. If it is judged that no object appears, and the background changes only slightly due to the change of light, the data of each image point is updated at a preset ratio. Therefore, the second background update parameter α2 is often set to be very small 値 or 0. After the background of all the image points in the detection unit is updated, it can enter the next processing cycle to continue the detection and calculation of the object occupancy state or the object movement state. The content of the image data of the present invention is expressed by the image intensity of each pixel. If a monochrome image source is used, its data format is represented by grayscale; if a colorful image source is used, its data format is represented by a combination of more than one band intensity. 4ACS / 200002TW 7 This paper size is applicable to China National Standard (CNS) A4 specification (210X297mm) I --------- ϋ, installed -----: I1Τ, ------ ^ ( Please read the precautions on the back and write this page first) c > 6 09 2 2 A7 B7 V. Description of the invention () Simple explanation of the drawings (please read the precautions on the back before writing this page) Figure 1 is Schematic diagram of the system architecture of the present invention. FIG. 2 is a data processing flowchart of the present invention. FIG. 3A shows the layout of the detection area and the selection of control points. FIG. 3B illustrates the definition of the sampling zone. Figure 3C shows the geometric correspondence between the image frame and the real space, and explains the meaning of the detection unit. (Example 1) Figure 3D shows the geometric correspondence between the image screen and the real space, and explains the meaning of the detection unit. (Example 2) Figure 4 is a flowchart of the background image extraction method. Fig. 5A is a graph showing the statistical distribution of image intensity and frequency in the background image extraction method. Figure 5B is the data block after the noise filtering step in the statistical allocation chart of Figure 5A. FIG. 6 is a flowchart of a method for calculating an occupied state of an object. FIG. 7A is a drawing of an image track line of an object. FIG. 7B is a plot of the trajectory of the object corresponding to the actual space. FIG. 8 is a flowchart of a method for calculating an object moving state. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs Figure 9 is the background image update flowchart. Figure 10 is a schematic diagram of an example of lane space occupancy data. 4ACS / 200002TW 8 This paper size applies Chinese National Standard (CNS) A4 specification (210X297 mm) 1220969 A7 B7_ V. Detailed description of the invention (^)) Detailed description of the invention The invention is a detection system for detecting the image of an object. The image signal automatically reads and judges the movement and occupation status of each object in the space to be detected. A specific embodiment of the present invention is applied to the detection of road traffic conditions, but it can also be applied to security systems or other systems. In the following, the detection of road traffic conditions is a specific embodiment of the present invention 'and will be described in detail based thereon. FIG. 1 is a schematic diagram of a system architecture of the present invention. The image signals of the vehicle and background are input by image source 1. This image source 1 can be a video camera installed on the roadside to transmit real-time road traffic images, or it can be recorded in advance. Good video. The main system 7 includes an image acquisition unit (A / D converter section) 2, a memory storage section (data storage section) 3, a data processing section (data processing section) 4, and an input / output control section (input / output control section) 5 Composition. The image signal input from the image source 1 is converted into digital data by the image capturing unit 2 and its content is the intensity of the pixel. The memory storage unit 3 is connected to the # image capture unit 2 and the data processing unit 4 and is responsible for the storage of these digital data and the temporary storage of the calculation data at each stage. The steps and details of the traffic state calculation performed by the data processing unit 4 of the present invention will be described later. The road traffic detection result processed by the data processing unit 4 is output to the external terminal device 6 by the input / output control unit 5 'Supply: set 4ACS / 200002TW 9 This paper size applies to China National Standard (CNS) Α4 specification (210 × 297) (%) (Please read the notes on the back before filling out this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Dingquan Ministry of Economy 1220969 A7 B7 V. Description of the invention ((0) References for road traffic information or dynamic traffic control. 2 is a data processing flowchart of the present invention. Step 41 includes three actions: definition of detection area, definition of sampling zone, and definition of detection unit. First, define a clear lane space on the image screen as the detection area. As shown in FIG. 3A, a number of fixed points are selected as the control points 411, and the relative coordinates of each control point 411 are measured in situ to calculate the relationship 414 corresponding to the coordinate transformation in the image frame (as shown in FIG. 3C). Within the space of the lane covered by the detection area, more than one sampling band 412 is defined according to the image geometric scale, so that it matches the image length of the detection area. , And can just cover enough lane space area (as shown in Figure 3B). In this specific embodiment, the sampling zone 412 is defined along the direction of traffic flow, that is, defined along the direction of the lane. Finally, the sampling zone 412 is cut into several easy The understanding detection unit 413 is used as a basic unit for reading and analyzing detection data. As shown in FIG. 3C, these detection units 413 correspond to the same lengths represented in the actual space. In other words, on the image screen The detection unit seen, as seen by the human eye, generally has short-distance and long-distance visual phenomena. Step 42 is the background image initialization step. There are two methods for the background image initialization. The first is to manually select the detection The picture with no moving objects in the measurement area is used as the initial background to read the background image data that does not contain vehicles or other moving objects in the detection area. However, such background-like screen images are not easy to obtain. The present invention provides an automatic completion of dynamics. Background image extraction method. Figure 4 is a flowchart of this background image extraction method. It is for continuous acquisition 4ACS / 200002TW 10 This paper is applicable in this paper National Standard (CNS) A4 Specification (210X297 mm) (Please read the precautions on the back before continuing on this page), \ 呑, printed by the Consumers ’Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 1220969 A7 B7___ V. Description of the invention (U ) The intensity of the individual image points in a sequence screen is calculated by assigning the number of possible intensities in the data of the most recent period and filtering out the background image. The following uses a monochrome image as an example. As step 421 , Capture the image intensity 同一 (the number of samples is determined empirically) of the same point on the image screen at successive moments, and make a statistical distribution of intensity and frequency. For the convenience of explanation, this statistical distribution is made Figure 5A shows the frequency-intensity / statistics chart. In step 422, the intensity of the number of times that is less than a preset threshold 値 is regarded as noise and filtered. Allocation of times after filtering out noise As shown in Figure 5B, this example can be roughly divided into three data blocks. Step 423: Each consecutive non-zero intensity number of times is the same data block, and the accumulation of the intensity number of times of each data block is regarded as the area of each data block. In step 424, the area of each data block is compared, and the data block with the largest area is selected, that is, data block 2 in this example. In step 425, the data block (the data block 2 in this example) with the largest area is selected, and the intensity 値 that occurs most frequently, that is, the maximum value 値 2 in this example. Think of this intensity 値 as the initial background 値 of the image point. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 1 ,! · 1 ----- install I -------. I order 1 (Please read the precautions on the back before filling this page) Step 41 and Step 42 above are pre-operational stages. After the background initialization of all pixels in the detection area is completed, further calculations are performed on the measured intensity of the pixels. As shown in Fig. 2, the present invention includes two main detection and calculation operations: object occupancy state detection (step 43) and object movement state detection (step 45). FIG. 6 is a flowchart of a calculation method for detecting an object occupancy state. Taking a specific embodiment of the present invention as an example, in step 431, 4ACS / 200002TW at a certain time in an individual detection unit is first read according to the definition of the detection area. Specifications (210X297 mm) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economy 1220969 A7 B7

V. Description of the invention (\ 2J pixel data. Step 432, compare the individual pixel data of each detection unit with the pixel data at the same position in the background image, and obtain the image intensity of the pixel at that moment and its background Step 433, set a threshold 値 A, and compare the difference 所得 obtained in step 432 with this threshold 値. If the difference 所得 obtained in step 432 is greater than the preset threshold 値, go to step 434, and make it The possession state 値 is 1; if not, let it be 0. After all the image points in the detection unit have been determined, proceed to step 435. Add the determination 所有 of all the image points and divide by the detection unit. The total number of image points, if the quotient is greater than another preset threshold 値 B, the detection unit is determined to be occupied by the vehicle, and it proceeds to step 436, and the input / output control unit 5 outputs the vehicle occupation signal; if not, it is regarded as There is no vehicle for this detection unit. The threshold 上述 A above can be set by considering factors such as light and dark, the real distance represented by the image point to the camera model, and other factors; the threshold 値 B is related to the detection sensitivity, the smaller the , The more sensitive. After all the detection units have completed the discrimination, the processing cycle of the moving state of the object is completed. The coordinates of each image point in the occupied state (1) after a comparison with the background in a continuous period of time represent the output of the image, which can be obtained Figure 7A shows the visual trajectory of a vehicle. If the image coordinates of individual pixels are transformed into coordinates in real space and outputted as a map, the true trajectory of a vehicle is shown in Figure 7B. Figure 2 The present invention further includes an object movement state detection operation in step 45. Fig. 8 is a flowchart of an object movement state detection method. According to the present invention, 4ACS / 200002TW 12 This paper size is applicable to China National Standard (CNS) 8 4 specifications (210X297) Li) — — · I -----: equipment -----: 1 order: ---— Quan (please read the precautions on the back before filling this page) 1220969 A7 B7___ V. Description of the invention ( 13) In a specific embodiment, in step 451, the pixel data at a certain moment in the individual detection unit is first read according to the definition of the detection area. Step 452 'the individual pixel data of each detection unit is compared with the previous Same position in an image frame The pixel data is compared, and the difference in image intensity between the current point and the previous point is obtained. In step 453, a threshold value is set. If the difference between the comparison results of the individual pixel points is greater than the threshold value, the process proceeds to step 454. , Let its moving state flag 値 be 1; if not, let it be 0. Step 455, the moving state flag 状态 of individual pixels at different times is continuously accumulated by η pens, where η is a continuous judgment of a preset moving state.次数 Accumulation times. If the quotient obtained by dividing the sum by dividing by η is greater than another threshold 値, the image point is regarded as having a moving object ', then proceed to step 456, and make its moving state discriminate 値 to 1. Step 457, calculate the same detection The total number of image points in the measurement unit where there are moving objects. If the ratio of the total number of image points to the total number of image points of the detection unit is greater than a preset threshold, then step 458 is performed to output the object movement signal of the detection unit; if not, it is determined that the detection unit belongs to No vehicle moves in the lane space. When all the detection units have completed the discrimination, the process of detecting the movement state of the object is completed. When the object movement state detection operation is completed, the process proceeds to step 46 and the detection result is output. Then, the background image update in step 47 is performed. Figure 9 is a flowchart of background image update. In step 471, the image intensity Ft of the image point in the individual detection unit at time t is first read according to the definition of the detection area. In step 472, the read image point intensity Ft is compared with the image point background 値 Bt at that moment, and a difference 値 Dt is obtained. Step 473 'Will Dt 4ACS / 200002TW 13 This paper size applies Chinese National Standard (CNS) A4 specification (210X297 mm) (please read the precautions on the back before writing this page) Order-Quan · Employee of Intellectual Property Bureau, Ministry of Economic Affairs Printed by the Consumer Cooperative 1220969 A7 B7 V. Description of the invention (ι +) Compare with a preset door threshold 'if Dt is greater than this preset threshold 令' Make the background update coefficient Mt 1; if not, make it be 〇, then go to step 474 or 475. These two steps are mainly to set the two background update parameters α ^, “2” and to set the two background update parameters d 1, ^ 2 and the background 値 Bt, the difference 値 Dt and the background update into the following formula:

Bt + 1 = Bt + [a i (l-Mt) + a 2Mt] Dt is to obtain the background tBt + 1 of the pixel at time t + 1. In step 476, the background 値 Bt + i 'is output as the new background 此 of the image point in the next detection cycle operation. After the backgrounds of all the pixels in all detection units are updated, the background update processing cycle is completed. Because the range of lanes to be detected may change with the change of sky, weather or light, the original spatial background may also change with the appearance of fixed objects. These factors are the main purpose of background update. So the implication of 'background update' is that when the reading of the individual pixel data and the pixel data at the same position as the background shows that a vehicle appears, the background is updated with a very small proportion, or not even updated. Therefore, the second background update parameter α 2 is often set to be very small or zero. If it is determined that no vehicle appears, the difference is purely due to slight changes in the background caused by the change in light, the background update step of the present invention can dynamically detect the changes to maintain the correctness of the detection. The pixel sampling content of the present invention is expressed by the image intensity of each pixel. If a monochrome image source is used, its data format is represented by grayscale; if a colorful image source is used, its data format is represented by a combination of more than one band intensity. The comparison of pixel image intensity mainly reflects 4ACS / 200002TW 14 This paper size applies Chinese National Standard (CNS) A4 specification (210X297 mm)

If0¾ (Please read the notes on the back before filling this page)

I Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 1220969 A7 B7 V. Description of the invention (&) The degree of difference in the intensity of the image information, including characteristics such as brightness or / and color. As shown in FIG. 2, a specific embodiment of the present invention After the detection result is output in step 46, the obtained data can be further used for several applications. The following three index calculation methods are described below: lane time occupancy, lane space occupancy, and vehicle delay space distribution such as parking. 1. Lane time occupancy When the invention adopts a small-area, single-detection-unit detection area definition method for an individual lane, its occupancy calculation method is the same as the traditional time occupancy definition, and can be used to replace the traditional detector. It is explained below: Measurement time (seconds); h is the time (seconds) that the i-th vehicle possesses the individual detection unit; M is the total number of passing vehicles (vehicles); τ is the time occupation rate (%) of the individual detection units. The actual detection data is a discontinuous data type with a fixed time interval scanning, so the approach formula is as follows: IL- — · 1 ----- T, equipment ------. I1T ( Please read the precautions on the back before filling in this page) οτ Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs τ where bl is the vehicle ownership judgment of the individual detection unit at time i. If there is a vehicle, it is 1, Otherwise, 値 is 0; t is the scanning interval (seconds); N is the total number of scans (times) in the measurement time.

4ACS / 200002TW 15 This paper size applies to Chinese National Standard (CNS) A4 specification (210X297 mm) 1220969 A7 B7 V. Description of invention (l [) 2 · Lane space occupation (please read the precautions on the back before filling this page ) The foregoing is the definition of the detection area with a small area and a single detection unit. If the present invention adopts the definition method of a large range and a plurality of detection units, the distribution situation of the vehicle occupation rate in the lane space can be considered at the same time, which is called the lane space occupation rate. It is expressed as the spatial distribution of the time occupancy of individual detection units, see Figure 10. The definition of time occupancy of individual detection units is the same as above. 3. Spatial distribution of delays such as stoppages. Traditionally, manual data collection of vehicle stoppages using intersection vehicle delay survey methods is used to collect information on vehicles at a fixed time interval. The lag estimation formula is as follows ...

N

Dv = ί ^ η (/ = 1 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, where t is the counting time interval (seconds); N is the number of records (pens); it is called the i-th record of the number of parked vehicles, etc. (Vehicles); Dv is the total delay (seconds) of vehicle stopping, which is measured as vehicle delay (vehicle-seconds). In practice, investigators often have too many vehicles to stop counting due to stoppages (and often because of humans). Vision can no longer discern the outline of the car body and cannot count) the actual number of vehicles waiting and so on, so the space of the lane occupied by the vehicle fleet is often used to estimate the number of vehicles that may be stopping. , Which is similar to this view. The estimation formula is as follows: NN Μ

Ds -ti7 ^ dsi ~) / = 1 / = ι y = i where t is the scanning time interval (seconds); L is the truth covered by the detection unit -____ 4ACS / 200002TW 16 _ This paper standard is applicable to Chinese national standards (CNS ) A4 specification (210X297 mm) 1220969 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 _____B7 V. Description of the invention () Lane length (meters); N is the number of data (pen); M is the number of detection units;

Si is the number of detection units (units) identified in the data record of time i as the vehicle stopped; \ is the determination of the vehicle movement status of the jth detection unit at time i. 1. Otherwise, 値 is 0; Ds is the total time and space delay (meter-second) when the vehicle is stopped. Its measurement also considers the characteristics of time and space. The specific embodiment of the present invention can not only be the same as the traditional traffic detection point of view, but mainly count micro-data such as the number of vehicles at a certain point, the occupancy rate, and the driving speed. The traffic flow parameters such as the distribution of traffic congestion and the movement of vehicles as seen by the human eye can be converted into traffic flow data with both time and space characteristics. It can provide reference data that is more in line with human intuitiveness and easy to understand. The methods of image signal processing are based on one-dimensional data as a point-by-point judgment, rather than using two-dimensional data, so it can reduce the time for unnecessary calculations. The advantages of the present invention over the previous case are as follows: 1. Detection object: an object that can continuously or intermittently move. Applies to vehicles but is not limited to vehicles. 2. Sampling method: Sampling bands are arranged along the main common movement path of objects in the detection area. The sampling zone can be a group or complex array corresponding to real space, parallel to the object's moving path, and the points are continuously arranged adjacent to each other in a linear pattern. Since the pixel data has the characteristics of intermittent grid-like distribution, this layout method can naturally achieve the line segment close to the real space, and the data is smoothed, so no additional smoothing is required. (The number of sampling point groups included in the sampling zone _ 4ACS / 2000Q2TW 17 This paper size applies to the Chinese National Standard (CNS) M specifications (210X297 mm) IIII .----- ^ 装 ----- II Order r ( Please read the notes on the back before filling this page) 9 6 09 2 12 A7 B7 V. Description of the invention (18) (Please read the notes on the back before filling this page) It represents its detection sensitivity. More, the sensitivity is higher, but the number of processing sampling points is more; otherwise, the sensitivity is lower, but the sampling points to be processed are fewer.) 3. Detection unit: In the detection area, move the object along the direction of the object. It is divided into several detection units that are connected back and forth and represent equal lengths in real space, so sampling in the direction of object movement is uninterrupted. 4. Detection perspective: Use the same phenomenon as the visual observation to quantify, determine the state of objects in the detection area, such as the state of possession and movement, and directly use the space in the area as the detection object. Therefore, it is not limited to the judging process that takes a moving object as the detection object, and it is not necessary to assume the size of the object in advance, so the calculation efficiency is high. 5. Processing time: processing 5 or more detection cycles per second with high accuracy. Taking vehicle detection as an example, even if the vehicle speed is as high as 90 km / h, it is not easy for misjudgment to occur. 6. Background image initialization: Contains the number of statistics of intensity distribution of each image point, and performs noise filtering first. Select the non-zero data block with the largest cumulative sum and use the highest intensity 値 as the background to update the background image of the pixel. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 7. Background image update: The method used is to consider whether the object is occupied or not and decide whether to update the image. Although the present invention has been described above with reference to the preferred embodiment, the description is only an example of applying the present invention and cannot be used as a limitation. Any modification, omission or combination will be included in the present invention. Therefore, the protection scope of the present invention shall be determined by the scope of the attached patent application. _ 4ACS / 200002TW_18 This paper size applies to China National Standard (CNS) A4 (210X 297 mm)

Claims (1)

Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 1220969 A7 B7 VI. Patent application scope: Except where the total number of samples allocated for the number of statistics and the selection of the first threshold 门 are set so that the distribution of the number of statistics after filtering includes At least one set of non-zero numbers; the combination of consecutive non-zero numbers is regarded as a data block, and the total number of the same data block is regarded as the area of each data block, and the largest area is selected The data block, in which the image intensity 发生, which occurs most frequently in the data block, is the initial background of the sampling point. 4. The method according to item 1 of the scope of patent application, wherein the fifth step comprises: comparing the background of the sampling point at time t 値 Bt with the image intensity of the sampling point 取样 R A first difference 値 Dt is obtained later; a second threshold 値 is set, and the first difference 値 Dt is compared with the second threshold 値; a background update coefficient Mt is set, and if the first difference 値 Dt is greater than the first Two thresholds 値, the background update coefficient Mt of the image point is 1, if the first difference 値 is less than or equal to the second threshold 値, the background update coefficient 'of the image point is 0; set a first background The update parameter α! And a second background update parameter α2, the background 値 Bt, the first difference 値 Dt, the background update 値 Mt, the first background update parameter α !, and the second background update parameter " 2 Substitute into the following formula: Bt + i ~ Bt + [〇! I (l-Mt) + a 2Mt] Dt ______________ 士 4 ^ _ ^ J Ί τ = I (Please read the precautions on the back before filling this page) -Line 丨 This paper size applies to China National Standard (CNS) A4 (210x297 mm) Intellectual Property Bureau, Ministry of Economic Affairs Consumer cooperative work printed 1220969 A7 B7 six, patented range: to obtain the the t + image point in time of the background Ι Zhi Bt + 1. 5. The method as described in item 4 of the scope of patent application, wherein the second background update parameter α 2 is a minimum value. 6. The method according to item 4 of the scope of patent application, wherein the second background update parameter α 2 is zero. 7. The method as described in item 1 of the scope of patent application, wherein the data calculation in the third step is an object occupancy detection operation, and the steps include: comparing the image intensity to the background of the sampling point値 Compare and get a second difference 値; Set a third threshold 値, compare the second difference 値 with the third threshold 値; Set an possession state 若, if the second difference 値 is greater than the The third threshold 値, the occupation state 値 of the sampling point is 1, and if the second difference 値 is less than or equal to the third threshold 値, the occupation state 该 of the sampling point is 0; a fourth threshold 値 is set If the quotient of the total occupancy status of each sampling point in the detection unit divided by the total number of the sampling points in the detection unit is greater than the fourth threshold 门, determine the detection corresponding to the detection unit Area is occupied by objects. 8. The method as described in item 1 of the scope of patent application, wherein the data calculation in the third step is an object movement state detection operation, and the steps include: comparing the time t and the time t + 1 The image intensity is 値, and you get a 4 pack. Φ · — 丨 Order · (Read the precautions on the back before filling in this page) -Line-This paper size is applicable to China National Standard (CNS) A4 (210x297 mm) Wisdom of the Ministry of Economic Affairs Printed by the Consumer Cooperative of the Property Bureau 1220969 A7 B7 6. Scope of patent application: The third difference 値; set a fifth threshold 値, compare the third difference 値 with the fifth threshold ；; set a moving state flag値, if the third difference 値 is greater than the fifth threshold 値, the moving state flag 値 of the image point is 1; if the third difference 値 is less than or equal to the fifth threshold 値, the image point should be The moving state flag 値 is 0; a sixth threshold 値 is set, and the moving state flag 値 is accumulated by η pens. If the sum of the consecutive η pens of the moving state flag 値 divided by η is greater than the sixth threshold値, the sampling point A moving object appears; a seventh threshold is set; if the ratio of the sampling points of the detecting unit to the total sampling points of the detecting unit is greater than the seventh threshold, the detecting unit is regarded as There are moving objects. 9. The method according to item 7 of the scope of the patent application, wherein the coordinates of the sampling points at the neighboring time at which the occupied state is 11 are plotted, and one of the objects can occupy the visual trajectory line. 10. The method as described in item 1 of the scope of the patent application, wherein the image intensity 値 is expressed in grayscale. 11. The method according to item 1 of the scope of patent application, wherein the image intensity is represented by a combination of more than one color band intensity. 12. An image detection system, the system includes: at least one A / D converter section, which converts an image source into a digital data; _____________ 士 ^ —— —II ^ —4 5-^ Gukou I (please read the notes on the back before filling in this page), line _ This paper size applies to China National Standard (CNS) A4 (210x297 mm) 1220969 Employee Consumer Cooperatives, Intellectual Property Bureau, Ministry of Economic Affairs Print A7 B7 6. Scope of patent application: At least one memory storage section, connected to the at least one image capture unit, for storing one data; at least one image data processing section, Linked with the at least one memory storage unit, 'using at least one sampling zone and at least one detection unit, processing calculation and judgment of image intensity data, and calculation of object possession or movement state data; and at least one data input / output control A data input / output section is connected to the at least one data processing unit and controls the at least one data processing unit and an external Data transmission between the terminal apparatus. 13.—A method for detecting road traffic conditions, the method includes the following steps: a first step: defining a lane range to be detected on an image screen as a detection area, and performing a real object in the detection area; A standard conversion with an image of the real object, defines at least one sampling zone in the image of the detection area, and divides the at least one sampling zone into a plurality of detection units, wherein the at least one sampling zone and the detection zone The image area of the measurement area is equal and includes at least one linear sampling point group; a second step: determining the background of a sampling point in the detection unit; a third step: detecting each of the detection units in the detection unit An image intensity of one of the sampling points, the image intensity obtained after a calculation of a detection result; this paper size applies the Chinese National Standard (CNS) A4 specification (210x297 mm) (Please read the precautions on the back before filling (This page) = -line-1220969 A7 B7_ 6. Scope of patent application: a fourth step: output the detection result; and a fifth step: dynamically update the background frame. 14. The method according to item 13 of the scope of patent application, wherein the sampling zone is defined along the direction of the traffic flow. _____________ 士 ”t ___ ^ 1J ^, words 丨 (Please read the notes on the back before filling this page) -Line 丨 Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, this paper is printed in accordance with China National Standards (CNS) A4 specifications ( 210x297 mm)