TWI749770B - Flooding warning method - Google Patents
Flooding warning method Download PDFInfo
- Publication number
- TWI749770B TWI749770B TW109131901A TW109131901A TWI749770B TW I749770 B TWI749770 B TW I749770B TW 109131901 A TW109131901 A TW 109131901A TW 109131901 A TW109131901 A TW 109131901A TW I749770 B TWI749770 B TW I749770B
- Authority
- TW
- Taiwan
- Prior art keywords
- processor
- area
- warning
- detection field
- image
- Prior art date
Links
Images
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
Description
本發明係關於一種淹水警示方法，特別係關於一種可以判定關注區域是否有淹水或積水的狀況，並據以發出警示通知的淹水警示方法。The present invention relates to a flood warning method, in particular to a flood warning method that can determine whether there is flooding or stagnant water in an area of interest and issue a warning notice accordingly.
為了避免因淹水造成的財產受損或人員傷亡，不論是路面或是河邊、湖邊等鄰近水源的區域多設有水位偵測系統，以在偵測到水位過高時可以迅速地發出警示通知到相關單位，以降低淹水造成的損害。In order to avoid property damage or casualties caused by flooding, water level detection systems are often installed on roads, rivers, lakes, and other areas adjacent to water sources to quickly send out when the water level is too high. The warning is notified to relevant units to reduce the damage caused by flooding.
目前的水位偵測器的原理為，當偵測器內的電極接觸到液體時，會透過接觸到的液體形成封閉迴路，並依據流通的電流檢測其水位。因此，在判斷是否有淹水狀況時，可以藉由判斷電極間的阻抗值，並依據該阻抗值是否高於或低於域設的阻抗值來判斷該場域是否有積水或淹水的狀況。The principle of the current water level detector is that when the electrode in the detector contacts a liquid, a closed circuit is formed through the contacted liquid, and the water level is detected according to the current flowing. Therefore, when judging whether there is a flooding condition, you can judge whether there is water accumulation or flooding in the field by judging the impedance value between the electrodes and whether the impedance value is higher or lower than the impedance value of the domain setting. .
然而，藉由水位偵測器偵測淹水狀況，除了電極部位可能因交替暴露在艷陽及雨天的情況下而產生腐蝕的狀況，不僅會因替換偵測器而產生過高的耗材成本，且在耗損的偵測器尚未被替換的情況下，更可能使得偵測器的偵測準確度下降。此外，一個場域常需要裝設多個水位偵測器才能較全面地檢測該場域的淹水狀況，然即使裝設多個水位偵測器仍無法確定該些水位偵測器係裝設在容易淹水的地點，導致偵測結果不夠精確。However, the use of water level detectors to detect flooding conditions, in addition to the corrosion of the electrode parts due to alternate exposure to the sun and rain, not only will the replacement of the detector cause excessively high consumables costs, but also In the case where the worn-out detector has not been replaced, it is more likely that the detection accuracy of the detector will decrease. In addition, a field often needs to install multiple water level detectors to more comprehensively detect the flooding status of the field, but even if multiple water level detectors are installed, it is still impossible to determine that the water level detectors are installed. In locations that are prone to flooding, the detection results are not accurate enough.
鑒於上述，本發明提供一種以滿足上述需求的淹水警示方法。In view of the above, the present invention provides a flood warning method that meets the above needs.
依據本發明一實施例的淹水警示方法，適用於一偵測場域，該方法包含：以一攝像機拍攝該偵測場域以取得關聯於該偵測場域的一原始影像，其中該原始影像包含一預設偵測區域；以一處理器對該原始影像執行一影像處理程序以取得一處理後影像，且將該預設偵測區域重疊於該處理後影像；以該處理器計算該處理後影像的一積水區域與該預設偵測區域重疊的一比例值；以該處理器判斷該比例值是否落入一警示值域；以及當判斷該比例值落入該警示值域時，以該處理器輸出一警示通知。The flood warning method according to an embodiment of the present invention is suitable for a detection field, and the method includes: shooting the detection field with a camera to obtain an original image associated with the detection field, wherein the original The image includes a predetermined detection area; a processor executes an image processing procedure on the original image to obtain a processed image, and the predetermined detection area is overlapped with the processed image; the processor calculates the A scale value at which a water accumulation area of the processed image overlaps the preset detection area; the processor determines whether the scale value falls within a warning value range; and when it is judged that the scale value falls within the warning value range, A warning notice is output by the processor.
依據本發明一實施例的淹水警示方法，適用於一偵測場域，該方法包含：以一攝像機拍攝該偵測場域以取得關聯於該偵測場域的一原始影像，其中該原始影像包含多個關鍵點；以一處理器對該原始影像執行一影像處理程序以取得一處理後影像，且將該些關鍵點重疊於該處理後影像；以該處理器計算一水位估算值，該水位估算值關聯於該處理後影像的該些關鍵點之中涵蓋於一積水區域內的關鍵點；以該處理器判斷該水位估算值是否落入一警示值域；以及當判斷該水位估算值落入該警示值域時，以該處理器輸出一警示通知。The flood warning method according to an embodiment of the present invention is suitable for a detection field, and the method includes: shooting the detection field with a camera to obtain an original image associated with the detection field, wherein the original The image includes a plurality of key points; an image processing program is executed on the original image by a processor to obtain a processed image, and the key points are superimposed on the processed image; the processor calculates an estimated water level, The water level estimation value is associated with the key points in the water accumulation area among the key points of the processed image; the processor is used to determine whether the water level estimation value falls into a warning range; and when the water level estimation is judged When the value falls into the warning value range, the processor outputs a warning notice.
綜上所述，依據本發明一或多個實施例所示的淹水警示方法，可以即時地偵測各個案場的積水或淹水狀況，且相較於習知的水位偵測器，本發明所示的淹水警示方法不會因天氣交替變換而導致偵測裝置受損，甚至影響到偵測準確度的問題。並且，依據本發明一或多個實施例所示的淹水警示方法，可以在不需耗費高額成本裝設多個偵測裝置的情況下，仍可有效地判斷該區域的積水或淹水狀況。此外，依據本發明一或多個實施例所示的淹水警示方法，不僅可以針對每個偵測區域有對應的偵測標準，更可以適時地更新偵測標準，以使判斷是否淹水的結果可以更準確。且若預先規劃的偵測區域不利於觀察該區域的積水或淹水狀況，藉由本發明的淹水警示方法更可以重新劃分偵測區域，而不需人工前往偵測區域的地點替換偵測裝置的設置位置。In summary, according to the flood warning method shown in one or more embodiments of the present invention, the accumulated water or flooding condition of each case can be detected in real time, and compared with the conventional water level detector, this The flood warning method shown in the invention will not cause damage to the detection device due to alternate weather changes, and even affect the problem of detection accuracy. Moreover, according to the flood warning method shown in one or more embodiments of the present invention, it is possible to effectively determine the water accumulation or flooding condition of the area without the need to install multiple detection devices at high cost. . In addition, according to the flood warning method shown in one or more embodiments of the present invention, not only can there be a corresponding detection standard for each detection area, but also the detection standard can be updated in time to determine whether it is flooded or not. The result can be more accurate. And if the pre-planned detection area is not conducive to observing the accumulation or flooding of the area, the flood warning method of the present invention can further divide the detection area without manually going to the location of the detection area to replace the detection device The setting location.
以上之關於本揭露內容之說明及以下之實施方式之說明係用以示範與解釋本發明之精神與原理，並且提供本發明之專利申請範圍更進一步之解釋。The above description of the disclosure and the following description of the implementation manners are used to demonstrate and explain the spirit and principle of the present invention, and to provide a further explanation of the patent application scope of the present invention.
以下在實施方式中詳細敘述本發明之詳細特徵以及優點，其內容足以使任何熟習相關技藝者了解本發明之技術內容並據以實施，且根據本說明書所揭露之內容、申請專利範圍及圖式，任何熟習相關技藝者可輕易地理解本發明相關之目的及優點。以下之實施例係進一步詳細說明本發明之觀點，但非以任何觀點限制本發明之範疇。The detailed features and advantages of the present invention will be described in detail in the following embodiments. The content is sufficient to enable anyone familiar with the relevant art to understand the technical content of the present invention and implement it accordingly, and in accordance with the content disclosed in this specification, the scope of patent application and the drawings Anyone who is familiar with relevant skills can easily understand the purpose and advantages of the present invention. The following examples further illustrate the viewpoints of the present invention in detail, but do not limit the scope of the present invention by any viewpoint.
本發明所示的淹水警示方法適用於例如為路面、海港、水庫、河川、溝渠等偵測場域，且每一偵測場域較佳具有設置於路燈、交通號誌燈等的路側裝置的一攝像機以及一處理器，以基於該攝像機拍攝得的影像執行淹水警示方法。The flood warning method shown in the present invention is suitable for detection fields such as roads, seaports, reservoirs, rivers, ditches, etc., and each detection field preferably has a roadside device installed on street lights, traffic lights, etc. A camera and a processor of, to execute the flood warning method based on the image captured by the camera.
請一併參考圖1以及圖2A及2B，其中圖1係依據本發明一實施例所繪示的淹水警示方法的流程圖；圖2A及圖2B係依據本發明一實施例所繪示的即時影像及處理後影像的示意圖。Please refer to FIG. 1 and FIGS. 2A and 2B together, in which FIG. 1 is a flowchart of a flood warning method according to an embodiment of the present invention; FIGS. 2A and 2B are illustrated according to an embodiment of the present invention Schematic diagram of the real-time image and the processed image.
步驟S10：拍攝偵測場域以取得關聯於偵測場域的原始影像，其中該原始影像包含一預設偵測區域。Step S10: Shoot the detection field to obtain an original image related to the detection field, wherein the original image includes a predetermined detection area.
以圖2A為例，圖2A為即為攝像機所拍得的原始影像，而原始影像中的道路區域則為偵測場域，並且原始影像更包含預設偵測區域DA。換言之，在基於原始影像偵測該偵測場域的水位之前，可以先設定對應該偵測場域的一或多個感興趣區域（region of interest，ROI），並且所述的感興趣區域即可作為預設偵測區域DA。Taking FIG. 2A as an example, FIG. 2A is the original image taken by the camera, and the road area in the original image is the detection field, and the original image further includes the preset detection area DA. In other words, before detecting the water level of the detection field based on the original image, one or more regions of interest (ROI) corresponding to the detection field can be set first, and the region of interest is Can be used as the default detection area DA.
步驟S20：對原始影像執行影像處理程序以取得處理後影像，且將預設偵測區域重疊於處理後影像。Step S20: Perform an image processing procedure on the original image to obtain the processed image, and overlap the preset detection area on the processed image.
處理器可以基於預先取得的先前原始影像以及先前處理後影像對一人工智慧模型進行訓練，以使訓練後的人工智慧模型能夠對原始影像進行影像處理程序以輸出處理後影像。據此，訓練後的人工智慧模型即可在後續收到即時的原始影像時，可以逐像素（pixelwise）的方式對收到的原始影像執行影像處理程序以輸出對應的處理後影像。此述的人工智慧模型例如包含卷積類神經網路（convolutional neural network，CNN）、遞歸類神經網路（recurrent neural network，RNN），或兩者的組合，本發明不對人工智慧模型的類型予以限制。The processor can train an artificial intelligence model based on the previous original image obtained in advance and the previously processed image, so that the trained artificial intelligence model can perform an image processing program on the original image to output the processed image. Accordingly, when the trained artificial intelligence model subsequently receives the real-time original image, it can execute the image processing program on the received original image in a pixelwise manner to output the corresponding processed image. The artificial intelligence model described herein includes, for example, a convolutional neural network (CNN), a recurrent neural network (RNN), or a combination of the two. The present invention does not deal with the type of artificial intelligence model. Be restricted.
換言之，經過影像處理程序後的原始影像可以成為如圖2B所示的處理後影像，預設偵測區域DA可以重疊於該處理後影像。亦即，處理器執行影像處理程序以辨識出原始影像中是否包含積水區域，並且若原始影像中包含積水區域，則在處理後影像中則會包含如圖2B所示的積水區域PA（即圖2B中所示的白色區塊）。需特別注意的是，此述的積水區域僅為示例，處理器執行影像處理程序除了判定影像中是否有積水區域之外，更可以藉由辨識路面是否乾燥或潮濕、或車輛經過時產生水花與否等方式，判定積水的程度。In other words, the original image after the image processing procedure can become the processed image as shown in FIG. 2B, and the preset detection area DA can overlap the processed image. That is, the processor executes an image processing program to identify whether the original image contains a water accumulation area, and if the original image contains a water accumulation area, the processed image will include the water accumulation area PA as shown in Figure 2B (i.e. The white block shown in 2B). It should be noted that the water accumulation area described here is only an example. In addition to determining whether there is a water accumulation area in the image, the processor executes the image processing program. No, etc., to determine the degree of stagnant water.
步驟S30：計算處理後影像的積水區域與預設偵測區域重疊的比例值。Step S30: Calculate the ratio of the overlap between the water accumulation area of the processed image and the preset detection area.
亦即，因預設偵測區域DA重疊於處理後影像，故當處理後影像中包含積水區域PA時，則積水區域PA可以與處理後影像中的預設偵測區域DA重疊。藉此，處理器可以計算積水區域PA與預設偵測區域DA重疊的比例以取得一比例值。That is, because the preset detection area DA overlaps the processed image, when the processed image includes the water accumulation area PA, the water accumulation area PA may overlap the preset detection area DA in the processed image. In this way, the processor can calculate the overlap ratio of the water accumulation area PA and the preset detection area DA to obtain a ratio value.
步驟S40：判斷比例值是否落入警示值域。Step S40: Determine whether the ratio value falls within the warning value range.
警示值域較佳係作為判斷積水是否達淹水或危險標準的依據，因此，當積水區域PA與預設偵測區域DA重疊的比例值落入警示值域時，則表示偵測場域中的預設偵測區域DA已達淹水或可能淹水的程度，其中警示值域例如是大於等於50%的值域，然本發明不對警示值域的實際數值予以限制。The warning range is better to be used as a basis for judging whether the stagnant water meets the flooding or dangerous standard. Therefore, when the overlap ratio of the stagnant water area PA and the preset detection area DA falls into the warning range, it means that the detection area is The preset detection area DA has reached the level of flooding or possible flooding, where the warning value range is, for example, a value range greater than or equal to 50%, but the present invention does not limit the actual value of the warning value range.
步驟S50：輸出警示通知。Step S50: output a warning notice.
亦即，當處理器於步驟S40判斷比例值落入警示值域時，處理器即可輸出警示通知至相關單位，其中警示通知除了包含淹水通知之外，較佳還包含該偵測場域的位置資訊。That is, when the processor determines in step S40 that the ratio value falls within the warning range, the processor can output a warning notice to the relevant unit, where the warning notice preferably includes the detection field in addition to the flooding notice. Location information.
請回到步驟S40，當於步驟S40判斷比例值未落入警示值域時，表示積水區域PA未與偵測區域DA重疊，或是積水區域PA與偵測區域DA重疊的比例未達危險或警示程度，故處理器可以執行步驟S60結束方法，或是回到步驟S10再持續取得偵測場域的即時的原始影像，並再持續基於取得的原始影像判斷偵測場域的偵測區域DA的積水狀況。Please go back to step S40. When it is determined in step S40 that the ratio value does not fall into the warning range, it means that the water accumulation area PA does not overlap with the detection area DA, or the overlap ratio of the water accumulation area PA and the detection area DA is not dangerous or Warning level, the processor can execute step S60 to end the method, or return to step S10 to continuously obtain the real-time original image of the detection field, and continue to determine the detection area DA of the detection field based on the obtained original image The status of stagnant water.
此外，處理器較佳為一邊緣運算裝置（edge computing device），因此邊緣運算裝置較佳可以將警示通知輸出至相關單位的雲端伺服器，邊緣運算裝置亦可以是將處理後影像輸出至雲端伺服器，並由雲端伺服器計算處理後影像的積水區域與預設偵測區域重疊的比例值。換言之，邊緣運算裝置可以是將處理後影像輸出至雲端伺服器，以由雲端伺服器基於處理後影像判斷預設偵測區域的積水狀況；邊緣運算裝置亦可以是基於處理後影像判斷預設偵測區域的積水狀況，且於判斷比例值落入警示值域時再輸出警示通至雲端伺服器，本發明不以此為限。In addition, the processor is preferably an edge computing device. Therefore, the edge computing device can preferably output the warning notification to the cloud server of the relevant unit, and the edge computing device can also output the processed image to the cloud server. The cloud server calculates the ratio of the overlap between the water accumulation area of the processed image and the preset detection area. In other words, the edge computing device can output the processed image to the cloud server, so that the cloud server can determine the water accumulation in the preset detection area based on the processed image; the edge computing device can also determine the preset detection based on the processed image. Measure the water accumulation condition of the area, and when it is judged that the ratio value falls within the warning value range, then output a warning to the cloud server. The present invention is not limited to this.
且，實現為邊緣運算裝置的處理器可以係被密封在箱中（例如，路燈控制箱）並設置在遠離地面的位置，因此當天氣環境惡劣時，密封在箱中的處理器仍可以正常運作而不受到損害。藉由將處理器以邊緣運算裝置的方式實現，可以避免為了將多幀即時影像傳輸至雲端伺服器而佔用過多的傳輸量，以及避免雲端伺服器承載過多的運算量。Moreover, the processor implemented as an edge computing device can be sealed in a box (for example, a street light control box) and set away from the ground, so when the weather is bad, the processor sealed in the box can still operate normally Without being harmed. By implementing the processor as an edge computing device, it is possible to avoid excessive transmission volume for transmitting multi-frame real-time images to the cloud server, and avoid the cloud server from carrying too much computing volume.
請參考圖3，圖3係繪示圖1的步驟S30的細部流程圖。在取得處理後影像並將預設偵測區域重疊於處理後影像（圖1的步驟S20）之後，處理器可以執行步驟S30的子步驟S301、S303及S305，以較精準地劃分出處理後影像中的積水區域。Please refer to FIG. 3, which is a detailed flowchart of step S30 in FIG. 1. After obtaining the processed image and superimposing the preset detection area on the processed image (step S20 in FIG. 1), the processor can execute the sub-steps S301, S303, and S305 of step S30 to more accurately divide the processed image In the stagnant water area.
步驟S301：判斷對應於多個原始影像的多個處理後影像中的多個初估區域。Step S301: Determine a plurality of preliminary estimation regions in a plurality of processed images corresponding to a plurality of original images.
有鑑於偵測場域例如是車輛往來頻繁的道路，因此若處理器僅基於一幀原始影像進行偵測，可能會因車輛等物件剛好位於原始影像中的積水區塊而導致處理後影像中的積水區域失準。In view of the fact that the detection field is, for example, a road with frequent vehicles, if the processor only detects based on one frame of the original image, objects such as vehicles may just be located in the water block in the original image. The area of stagnant water is inaccurate.
因此，處理器取得的原始影像較佳是連續的多幀影像（例如，5～6幀的影像）或是在一間隔時段內依序取得的多幀影像，而在對該些原始影像執行影像處理程序之後即可取得多個處理後影像。並且，若該些原始影像具有如圖2A所示的積水的區塊時，則該些原始影像中的積水的區塊在經過影像處理程序後即可作為所述的多個初估區域。Therefore, the original image obtained by the processor is preferably a continuous multi-frame image (for example, an image of 5-6 frames) or a multi-frame image sequentially obtained in an interval, and the image is executed on these original images. After the processing procedure, multiple processed images can be obtained. Moreover, if the original images have water-filled areas as shown in FIG. 2A, the water-filled areas in the original images can be used as the multiple preliminary estimation areas after the image processing procedure.
步驟S303：以該些初估區域的聯集區域作為積水區域。Step S303: Use the combined area of the preliminary estimation areas as the stagnant water area.
處理器可以將在步驟S301所取得的該些初估區域進行聯集以取得聯集區域，並以該聯集區域作為積水區域，以精確地取得積水區域。The processor may combine the preliminary estimated areas obtained in step S301 to obtain a union area, and use the union area as a stagnant area to accurately obtain the stagnant area.
舉例而言，若在取得多幀的原始影像時有交通物件經過一或多個該些初估區域，則可能會使得初估區域的一部份被經過的交通物件遮蔽，因此處理器可以基於該些初估區域取得聯集區域，並以聯集區域作為積水區域，以取得精準的積水區域。For example, if a traffic object passes through one or more of these preliminary estimation areas when acquiring multiple frames of original images, a part of the preliminary estimation area may be obscured by the passing traffic objects, so the processor can be based on These preliminary estimated areas are to obtain a joint area, and the joint area is used as a water accumulation area to obtain an accurate water accumulation area.
步驟S305：計算積水區域與預設偵測區域之間的重疊比例以取得比例值。Step S305: Calculate the overlap ratio between the water accumulation area and the preset detection area to obtain a ratio value.
在取得積水區域後，處理器即可計算積水區域與預設偵測區域之間的重疊比例，據以取得用以判斷積水是否達警示狀態的比例值。After obtaining the stagnant water area, the processor can calculate the overlap ratio between the stagnant water area and the preset detection area to obtain a ratio value for judging whether the stagnant water reaches the warning state.
請參考圖4並搭配參考圖5，其中圖4係依據本發明另一實施例所繪示的淹水警示方法的流程圖，而圖5係依據本發明另一實施例所繪示的統計模型的示例圖。圖4所示的步驟S01、S03以及S05較佳係執行於圖1的步驟S10之前，以在實際對偵測場域進行積水偵測前，先建立對應於該偵測場域的統計模型，以取得適於該偵測場域的警示值域。Please refer to FIG. 4 in conjunction with FIG. 5. FIG. 4 is a flowchart of a flood warning method according to another embodiment of the present invention, and FIG. 5 is a statistical model according to another embodiment of the present invention. Example graph. Steps S01, S03, and S05 shown in FIG. 4 are preferably performed before step S10 in FIG. 1, so as to establish a statistical model corresponding to the detection field before actually performing the water accumulation detection on the detection field. In order to obtain the warning value range suitable for the detection field.
在以攝像機拍攝偵測場域以取得關聯於該偵測場域的原始影像（步驟S10）之前，處理器更可以先執行步驟S01：取得關聯於偵測場域的多個先前處理後影像的多個建模比例值。Before shooting the detection field with a camera to obtain the original image associated with the detection field (step S10), the processor may first perform step S01: obtain a plurality of previously processed images associated with the detection field Multiple modeling scale values.
換言之，攝像機可以預先拍攝偵測場域持續一段時間，以取得關聯於該偵測場域的多個原始影像，處理器即可對該些原始影像執行影像分析程序以取得所述的多個先前處理後影像。In other words, the camera can shoot the detection field for a period of time in advance to obtain multiple original images related to the detection field, and the processor can execute the image analysis program on the original images to obtain the multiple previous images. The processed image.
並且，處理器可以將預設偵測區域重疊於該些先前處理後影像，並執行如前述步驟S30的內容，以計算出每一該些先前處理後影像中的積水區域與預設偵測區域重疊的比例值，並以該些比例值作為該些建模比例值。In addition, the processor may overlap the preset detection area on the previously processed images, and execute the content of the aforementioned step S30 to calculate the water accumulation area and the preset detection area in each of the previously processed images Overlapping scale values, and use these scale values as the modeling scale values.
步驟S03：基於該些先前處理後影像的數量以及該些建模比例值建立統計模型。Step S03: Establish a statistical model based on the number of previously processed images and the modeling scale values.
換言之，處理器可以基於該些先前處理後影像的數量以及該些建模比例值建立如圖5所示的高斯分佈圖以作為該統計模型，並且以先前處理後影像的數量作為高斯分佈圖的縱軸（影像數量）；以建模比例值作為高斯分佈圖的橫軸（比例值）。藉此，建立出來的統計模型可以較符合偵測場域的實際狀況。然此述的高斯分佈圖以及其縱軸、橫軸的設定僅為示例，本發明不對統計模型的實現方式予以限制。In other words, the processor can create a Gaussian distribution map as shown in FIG. 5 based on the number of previously processed images and the modeling scale values as the statistical model, and use the number of previously processed images as the Gaussian distribution map. Vertical axis (number of images); take the modeling scale value as the horizontal axis (scale value) of the Gaussian distribution map. In this way, the established statistical model can be more in line with the actual conditions of the detection field. However, the Gaussian distribution chart described here and the settings of its vertical axis and horizontal axis are only examples, and the present invention does not limit the implementation of the statistical model.
步驟S05：以統計模型的指定信賴區間的上限值作為警示值域的下限值。Step S05: Use the upper limit of the designated confidence interval of the statistical model as the lower limit of the warning range.
亦即，統計模型可以具有指定信賴區間CI，因此處理器即可以指定信賴區間CI的上限值對應到的統計模型的比例值作為警示值域Warn的下限值V。舉例而言，圖5所示的統計模型的指定信賴區間CI例如為95%以內的區間，而若95%的指定信賴區間CI的上限值係對應到為50%的比例值V，則處理器可以將50%的比例值V（即取指定信賴區間CI的上限值對應到統計模型的橫軸的值為比例值V）作為警示值域Warn的下限值。因此，若後續在步驟S30所計算得的比例值高於50%，則可以判斷此比例值已落入警示值域Warn，並據此輸出警示通知。That is, the statistical model may have a designated confidence interval CI, so the processor may designate the proportional value of the statistical model corresponding to the upper limit value of the confidence interval CI as the lower limit value V of the warning range Warn. For example, the designated confidence interval CI of the statistical model shown in FIG. 5 is, for example, an interval within 95%, and if the upper limit value of the 95% designated confidence interval CI corresponds to a ratio value V of 50%, then the processing The device can take the 50% proportional value V (that is, take the upper limit value of the designated confidence interval CI corresponding to the value of the horizontal axis of the statistical model as the proportional value V) as the lower limit value of the warning range Warn. Therefore, if the ratio value calculated in step S30 is higher than 50%, it can be judged that the ratio value has fallen into the warning range Warn, and a warning notification is output accordingly.
並且，在實際偵測時，不論計算出的比例值為何，該比例值皆可用以更新統計模型，以使統計模型及依據指定信賴區間所得的警示值域可以更符合偵測場域的當前或常態狀況。舉例而言，在經常下雨的一個偵測場域中，若對應該偵測場域的一預設偵測區域的比例值常態性地較高，則統計而得的指定信賴區間的上限值可能也較高，故對應該預設偵測區域的警示值域的下限值可能也較高，以避免誤發警示通知。Moreover, in actual detection, regardless of the calculated ratio value, the ratio value can be used to update the statistical model, so that the statistical model and the warning range based on the specified confidence interval can be more in line with the current or the detection field. Normal state. For example, in a detection field where it rains frequently, if the proportion of a predetermined detection area corresponding to the detection field is normally higher, the upper limit of the specified confidence interval obtained by statistics The value may also be higher, so the lower limit of the warning value range corresponding to the preset detection area may also be higher to avoid false warning notifications.
此外，若預設偵測區域內有暫時性的異常狀況時（例如，因施工而造成的暫時性積水），則藉由更新統計模型亦可以避免因該暫時性的異常狀況而不斷地發出警示通知。In addition, if there is a temporary abnormal situation in the preset detection area (for example, temporary stagnant water caused by construction), updating the statistical model can also avoid continuous warnings due to the temporary abnormal situation Notice.
請一併參考圖6及圖7A及7B，其中圖6係依據本發明另一實施例所繪示的淹水警示方法的流程圖；圖7A及7B係依據本發明一實施例所繪示的即時影像及處理後影像的示意圖。Please refer to FIGS. 6 and 7A and 7B together, in which FIG. 6 is a flowchart of a flood warning method according to another embodiment of the present invention; FIGS. 7A and 7B are illustrated according to an embodiment of the present invention Schematic diagram of the real-time image and the processed image.
圖6所示的步驟S10’、S50’及S60’的實現方式相同於圖1所示的步驟S10、S50及S60，故相同之處不再於此贅述，惟不同處在於圖6所示的淹水警示方法的步驟S20’、S30’及S40’不同於圖1的步驟S20、S30及S40。The implementation of steps S10', S50', and S60' shown in FIG. 6 is the same as that of steps S10, S50, and S60 shown in FIG. The steps S20', S30', and S40' of the flood warning method are different from the steps S20, S30, and S40 of FIG. 1.
根據圖6，在取得原始影像（步驟S10’）後，處理器可以接著執行步驟S20’：對原始影像執行影像處理程序以取得處理後影像，且將多個關鍵點重疊於處理後影像。According to FIG. 6, after obtaining the original image (step S10'), the processor may then perform step S20': perform an image processing program on the original image to obtain the processed image, and overlap multiple key points on the processed image.
如圖7A所示，原始影像可以包含多個關鍵點KP，因此在取得如圖7B所示的處理後影像之後，處理器可以將該些關鍵點KP重疊到如圖7B所示的處理後影像。As shown in FIG. 7A, the original image may contain multiple key points KP, so after obtaining the processed image as shown in FIG. 7B, the processor can overlap these key points KP to the processed image as shown in FIG. 7B .
步驟S30’：計算關聯於處理後影像中的關鍵點的水位估算值。Step S30': Calculate the estimated water level associated with the key points in the processed image.
所述的水位估算值關聯於處理後影像的該些關鍵點之中涵蓋於積水區域內的關鍵點。詳言之，水位估算值例如係該積水區域所涵蓋該些關鍵點的數量對該些關鍵點的總數的比例值，以圖7B為例，因積水區域PA未涵蓋任何一個關鍵點KP，故於步驟S30’所計算出的水位估算值為0。在未繪示的另一實施例中，若另一處理後影像的關鍵點KP總數為16個，且該另一處理後影像中的積水區域PA涵蓋4個關鍵點KP，則對應該另一處理後影像的水位估算值即為25%。The estimated water level is related to the key points in the stagnant water area among the key points of the processed image. In detail, the estimated water level is, for example, the ratio of the number of key points covered by the stagnant area to the total number of key points. Taking Figure 7B as an example, because the stagnant area PA does not cover any key point KP, The estimated water level calculated in step S30' is zero. In another embodiment not shown, if the total number of key points KP of another processed image is 16, and the puddle area PA in the another processed image covers 4 key points KP, it corresponds to another The estimated water level of the processed image is 25%.
在未繪示的又一實施例中，每一關鍵點KP可以具有不同的權重值，例如較靠近道路的關鍵點KP可以具有大於1的權重值，而距離道路較遠的其他關鍵點KP則可以具有等於1或小於1的權重值，因此當積水區域PA涵蓋具有大於1的權重值的關鍵點KP時，水位估算值即會適應性地增加。據此，藉由對不同位置的關鍵點KP設定不同的權重值，可以強化偵測場域中重點區域的監控。In another embodiment not shown, each key point KP may have a different weight value. For example, a key point KP closer to the road may have a weight value greater than 1, while other key points KP farther away from the road may have a weight value greater than 1. It can have a weight value equal to 1 or less than 1, so when the water accumulation area PA covers the key point KP with a weight value greater than 1, the water level estimate value will increase adaptively. Accordingly, by setting different weight values for the key points KP at different positions, the monitoring of the key areas in the detection field can be strengthened.
此外，相似於圖3所示的步驟S30的子步驟，圖6的步驟S30’中取得水位估算值的方式亦可以是先判斷多個處理後影像中的多個初估區域，並以該些初估區域的聯集區域作為積水區域，再計算積水區域所涵蓋該些關鍵點的數量，並以積水區域所涵蓋關鍵點的數量對該些關鍵點的總數的比例值作為水位估算值。In addition, similar to the sub-steps of step S30 shown in FIG. 3, the method of obtaining the water level estimation value in step S30' of FIG. The joint area of the preliminary estimation area is regarded as the stagnant area, and then the number of these key points covered by the stagnant area is calculated, and the ratio of the number of key points covered by the stagnant area to the total number of key points is used as the water level estimation value.
在計算出水位估算值後，處理器即可執行圖6的步驟S40’以判斷水位估算值是否落入警示值域，並於判斷水位估算值落入警示值域時執行步驟S50’輸出警示通知；於判斷水位估算值未落入警示值域時執行步驟S60’結束方法。After calculating the estimated water level, the processor can execute step S40' in Figure 6 to determine whether the estimated water level falls within the warning range, and execute step S50' to output a warning notification when it is judged that the estimated water level falls within the warning range. ; When it is judged that the estimated water level does not fall into the warning value range, step S60' is executed to end the method.
此外，請回到步驟S10’，在取得原始影像之前，處理器可以執行如圖4所示的步驟S01、S03及S05以基於建模比例值建立對應該偵測區域的統計模型，其中需特別注意的是，處理器取得用於圖6的實施例的建模比例值係基於關鍵點KP的數量所取得，亦即圖6的實施例的建模比例值係該些先前處理後影像中的積水區域所涵蓋的關鍵點KP數量，對該些先前處理後影像中的關鍵點KP的總數的比例，其中建模比例值亦可以係基於具有不同權重值的關鍵點KP所計算出的比例值所取得。In addition, please go back to step S10'. Before obtaining the original image, the processor can execute steps S01, S03, and S05 as shown in Figure 4 to establish a statistical model corresponding to the detection area based on the modeling scale value. Note that the processor obtains the modeling scale value used in the embodiment of FIG. 6 based on the number of key points KP, that is, the modeling scale value of the embodiment in FIG. 6 is the value of the previously processed images. The number of key points KP covered by the stagnant area is the ratio of the total number of key points KP in the previously processed images. The modeling scale value can also be calculated based on the key points KP with different weight values. Obtained.
並且，於步驟S30’取得的水位估算值可以用於更新該統計模型，以使統計模型可以符合偵測場域的當前狀況。Moreover, the estimated water level obtained in step S30' can be used to update the statistical model so that the statistical model can conform to the current conditions of the detection field.
綜上所述，依據本發明一或多個實施例所示的淹水警示方法，可以即時地偵測各個案場的積水或淹水狀況，且相較於習知的水位偵測器，本發明所示的淹水警示方法不會因天氣交替變換而導致偵測裝置受損，甚至影響到偵測準確度的問題。並且，依據本發明一或多個實施例所示的淹水警示方法，可以在不需耗費高額成本裝設多個偵測裝置的情況下，仍可有效地判斷該區域的積水或淹水狀況。此外，依據本發明一或多個實施例所示的淹水警示方法，不僅可以針對每個偵測區域有對應的偵測標準，更可以適時地更新偵測標準，以使判斷是否淹水的結果可以更準確。且若預先規劃的偵測區域不利於觀察該區域的積水或淹水狀況，藉由本發明的淹水警示方法更可以重新劃分偵測區域，而不需人工前往偵測區域的地點替換偵測裝置的設置位置。In summary, according to the flood warning method shown in one or more embodiments of the present invention, the accumulated water or flooding condition of each case can be detected in real time, and compared with the conventional water level detector, this The flood warning method shown in the invention will not cause damage to the detection device due to alternate weather changes, and even affect the problem of detection accuracy. Moreover, according to the flood warning method shown in one or more embodiments of the present invention, it is possible to effectively determine the water accumulation or flooding condition of the area without the need to install multiple detection devices at high cost. . In addition, according to the flood warning method shown in one or more embodiments of the present invention, not only can there be a corresponding detection standard for each detection area, but also the detection standard can be updated in time to determine whether it is flooded or not. The result can be more accurate. And if the pre-planned detection area is not conducive to observing the accumulation or flooding of the area, the flood warning method of the present invention can further divide the detection area without manually going to the location of the detection area to replace the detection device The setting location.
雖然本發明以前述之實施例揭露如上，然其並非用以限定本發明。在不脫離本發明之精神和範圍內，所為之更動與潤飾，均屬本發明之專利保護範圍。關於本發明所界定之保護範圍請參考所附之申請專利範圍。Although the present invention is disclosed as above in the foregoing embodiments, it is not intended to limit the present invention. All changes and modifications made without departing from the spirit and scope of the present invention fall within the scope of the patent protection of the present invention. For the scope of protection defined by the present invention, please refer to the attached patent scope.
DA:預設偵測區域 PA:積水區域 KP:關鍵點 CI:指定信賴區間 V:比例值 Warn:警示值域 DA: Default detection area PA: stagnant water area KP: key points CI: designated confidence interval V: proportional value Warn: warning range
圖1係依據本發明一實施例所繪示的淹水警示方法的流程圖。 圖2A及2B係所繪示即時影像及處理後影像的示意圖。 圖3係繪示圖1的步驟S30的細部流程圖。 圖4係依據本發明另一實施例所繪示的淹水警示方法的流程圖。 圖5係依據本發明另一實施例所繪示的統計模型的示例圖。 圖6係依據本發明又一實施例所繪示的淹水警示方法的流程圖。 圖7A及7B係繪示即時影像及處理後影像的示意圖。 FIG. 1 is a flowchart of a flood warning method according to an embodiment of the present invention. 2A and 2B are schematic diagrams of real-time images and processed images. FIG. 3 is a detailed flowchart of step S30 in FIG. 1. Fig. 4 is a flowchart of a flood warning method according to another embodiment of the present invention. FIG. 5 is an example diagram of a statistical model drawn according to another embodiment of the present invention. Fig. 6 is a flowchart of a flood warning method according to another embodiment of the present invention. 7A and 7B are schematic diagrams showing real-time images and processed images.
Claims (8)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW109131901A TWI749770B (en) | 2020-09-16 | 2020-09-16 | Flooding warning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW109131901A TWI749770B (en) | 2020-09-16 | 2020-09-16 | Flooding warning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TWI749770B true TWI749770B (en) | 2021-12-11 |
| TW202213285A TW202213285A (en) | 2022-04-01 |
Family
ID=80681087
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW109131901A TWI749770B (en) | 2020-09-16 | 2020-09-16 | Flooding warning method |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI749770B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWM486043U (en) * | 2014-01-13 | 2014-09-11 | Water Resources Agenacy Ministry Of Economic Affairs | Community flood warning sensing system |
| WO2018058856A1 (en) * | 2016-09-28 | 2018-04-05 | 深圳市元征科技股份有限公司 | Road-surface ponding detection method and device |
| CN109670404A (en) * | 2018-11-23 | 2019-04-23 | 江苏理工学院 | A kind of road ponding image detection method for early warning based on mixed model |
| CN109741391A (en) * | 2018-12-26 | 2019-05-10 | 斑马网络技术有限公司 | Detection method, device and the storage medium of surface gathered water depth |
| CN109801327A (en) * | 2018-12-21 | 2019-05-24 | 杭州电子科技大学 | A kind of urban waterlogging depth of accumulated water information extracting method based on video data |
| US20190313510A1 (en) * | 2016-09-22 | 2019-10-10 | Signify Holding B.V. | Flooding localization and signalling via intelligent lighting |
| CN110675415A (en) * | 2019-12-05 | 2020-01-10 | 北京同方软件有限公司 | Road ponding area detection method based on deep learning enhanced example segmentation |
-
2020
- 2020-09-16 TW TW109131901A patent/TWI749770B/en active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWM486043U (en) * | 2014-01-13 | 2014-09-11 | Water Resources Agenacy Ministry Of Economic Affairs | Community flood warning sensing system |
| US20190313510A1 (en) * | 2016-09-22 | 2019-10-10 | Signify Holding B.V. | Flooding localization and signalling via intelligent lighting |
| WO2018058856A1 (en) * | 2016-09-28 | 2018-04-05 | 深圳市元征科技股份有限公司 | Road-surface ponding detection method and device |
| CN109670404A (en) * | 2018-11-23 | 2019-04-23 | 江苏理工学院 | A kind of road ponding image detection method for early warning based on mixed model |
| CN109801327A (en) * | 2018-12-21 | 2019-05-24 | 杭州电子科技大学 | A kind of urban waterlogging depth of accumulated water information extracting method based on video data |
| CN109741391A (en) * | 2018-12-26 | 2019-05-10 | 斑马网络技术有限公司 | Detection method, device and the storage medium of surface gathered water depth |
| CN110675415A (en) * | 2019-12-05 | 2020-01-10 | 北京同方软件有限公司 | Road ponding area detection method based on deep learning enhanced example segmentation |
Non-Patent Citations (1)
| Title |
|---|
| 曾建澔、羅世瑋、陳倫奇、吳志泓、林芳邦,"基於影像分析技術之自動化水位偵測警戒系統",TANET2013臺灣網際網路研討會-【論文集】,2013年 * |
Also Published As
| Publication number | Publication date |
|---|---|
| TW202213285A (en) | 2022-04-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101060031B1 (en) | Device and method for monitoring water level | |
| US10872247B2 (en) | Image feature emphasis device, road surface feature analysis device, image feature emphasis method, and road surface feature analysis method | |
| JP6187584B2 (en) | Method and apparatus for ensuring minimum contrast for machine vision systems | |
| AU2019249431B2 (en) | System and method for early identification and monitoring of defects in transportation infrastructure | |
| JP2008057994A (en) | Water level observation system by image processing | |
| CN110929676A (en) | Deep learning-based real-time detection method for illegal turning around | |
| US20190078882A1 (en) | Water level measurement system and water level measurement method | |
| US11842617B2 (en) | Flood warning method | |
| WO2015193043A1 (en) | Flash flooding detection system | |
| JP2016108827A (en) | Irrigation channel monitoring server and irrigation channel monitoring system | |
| CN105554494B (en) | Snow point image detecting method and device and video quality detection device and system | |
| KR101949155B1 (en) | Method and cctv for monitoring flooding of road and the same | |
| CN115186943A (en) | Urban drainage waterlogging prediction modeling method and system and electronic equipment | |
| KR20220015003A (en) | Monotoring system for sewer | |
| TWI720644B (en) | Method for size estimating by image recognition of specific target | |
| TWI749770B (en) | Flooding warning method | |
| KR20180074572A (en) | Speed-sensitive water level system and predicting dangerous situations | |
| KR102533185B1 (en) | General river safety management device using CCTV camera images | |
| TWI481824B (en) | Method of water level surveillance | |
| US20200049856A1 (en) | Real-time computation of an atmospheric precipitation rate from a digital image of an environment where an atmospheric precipitation is taking place | |
| CN110660229A (en) | Vehicle speed measuring method and device and vehicle | |
| CN110658353B (en) | Method and device for measuring speed of moving object and vehicle | |
| TW201418671A (en) | Method of water level measurement | |
| CN115830460A (en) | Real-time monitoring method and system for river flood prevention | |
| CN116091456A (en) | Road visibility detection method |