Multi-source hidden danger analysis information fusion method

Abstract

The invention discloses a multi-source hidden danger analysis information fusion method, and belongs to the field of computer vision. According to the method, information of hidden danger analysis results generated by a plurality of hidden danger analysis systems in various latitudes such as hidden danger types, coordinate positions, confidence coefficients and hidden danger areas is fused; through the information fusion mode, the consistency explanation and description of the hidden danger analysis result are obtained, and a fused analysis result is formed. Therefore, manual analysis of hidden danger results generated by the hidden danger analysis system is reduced, the processing difficulty of the user is reduced, and finally the generation probability of hidden danger missed notices is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the field of computer vision, and in particular to a multi-source hidden danger analysis information fusion method. Background Art

[0002] In modern life and industrial production, electric energy is an indispensable and important green energy. In power transmission, transmission lines are an important link in power transmission. In order to ensure the stable operation of transmission lines and improve the reliability and safety of power transmission, the visualization of transmission line channels has become an important research topic today.

[0003] In the transmission line channel visualization technology, the usual practice is to install intelligent terminal devices at the possible hidden danger locations in the transmission line, and transmit the collected image data to the transmission line channel visualization platform through wireless communication for hidden danger analysis and processing. In this process, there are several processing methods. The first is to use a predetermined analysis model on the terminal device to perform hidden danger analysis, and upload the analysis results to the transmission line channel visualization platform as hidden danger result processing; the second is to use a secondary analysis method. Specifically, the terminal device performs the first analysis and uploads the analysis results to the back-end hidden danger analysis system of the transmission line channel visualization platform. The back-end hidden danger analysis system will receive the analysis results and the pictures with hidden dangers for secondary analysis. Since the analysis model and computing performance are better than the terminal device, the results obtained are relatively more accurate.

[0004] As business develops, multiple manufacturers' hidden danger analysis systems are deployed on the transmission line channel visualization platform to enable multiple hidden danger analysis systems to analyze and identify the same image data separately, so as to improve the accuracy of hidden danger identification and reduce the hidden danger missed reporting rate. At present, the monitoring user needs to switch between multiple hidden danger analysis results and manually judge the analysis results, which is difficult for users to handle and increases the probability of hidden danger missed reporting. Summary of the invention

[0005] The present invention provides a multi-source hidden danger analysis information fusion method, which fuses the hidden danger analysis results generated by multiple hidden danger analysis systems in terms of hidden danger type, coordinate position, confidence, hidden danger area and other multi-dimensional information, thereby reducing the processing difficulty for users and ultimately effectively reducing the probability of hidden danger omissions.

[0006] The present invention provides the following technical solutions:

[0007] A multi-source hidden danger analysis information fusion method, the method comprising:

[0008] S10: multiple hidden danger analysis systems analyze and identify the same image data respectively to obtain multiple hidden danger analysis results;

[0009] The hidden danger analysis results include hidden danger area, hidden danger confidence and hidden danger identification;

[0010] S20: convert various hidden danger identifications of the same hidden danger object by different hidden danger analysis systems into the same hidden danger identification;

[0011] S30: converting various rectangular coordinate systems of origin defined by different hidden danger analysis systems for the position of the hidden danger object in the image into a preset standard coordinate system;

[0012] S40: Convert various hidden danger confidences of different hidden danger analysis systems into the same standard range;

[0013] S50: The hidden danger analysis results are integrated according to the hidden danger areas, hidden danger confidences and hidden danger identifications of different hidden danger analysis systems.

[0014] Furthermore, S20 includes:

[0015] S21: Pre-establish a dictionary of hidden danger object names and hidden danger identifiers, and establish a corresponding relationship with the hidden danger identifiers of each hidden danger analysis system;

[0016] S22: According to the dictionary, various hidden danger identifications of the same hidden danger object generated by different hidden danger analysis systems are uniformly converted into the same hidden danger identification.

[0017] Further, the S50 includes:

[0018] S51: when the two potential hazard areas have no overlap or inclusion relationship, retain the potential hazard analysis results corresponding to the two potential hazard areas;

[0019] S52: when two hidden danger areas overlap and have no inclusion relationship, the overlapping area percentage of the two overlapping hidden danger areas is calculated, and the two hidden danger analysis results are merged according to the overlapping area percentage;

[0020] S53: When the two hidden danger areas are in a containment relationship, the two hidden danger analysis results are merged according to the hidden danger identifiers corresponding to the two hidden danger areas.

[0021] Furthermore, the calculation formula of the overlapping area percentage r is as follows:

[0022] r=A / (A1+A2-A)*100%

[0023] Where A is the overlapping area of ​​the two hidden danger areas, and A1 and A2 are the areas of the two hidden danger areas respectively.

[0024] Further, the S52 includes:

[0025] When the percentage of overlapping area exceeds the set threshold and the hidden danger identifications of two hidden danger analysis results are the same, the hidden danger analysis result with a higher hidden danger confidence is retained, and the hidden danger analysis result with a lower hidden danger confidence is discarded;

[0026] When the percentage of overlapping area exceeds the set threshold and the hidden danger identifications of the two hidden danger analysis results are different, the hidden danger area with high hidden danger confidence is retained, and the hidden danger area with low hidden danger confidence is discarded, and the hidden danger identifications of the two hidden danger analysis results are retained;

[0027] When the overlapping area percentage does not exceed the set threshold, the two hidden danger analysis results are retained.

[0028] Further, the S53 includes:

[0029] When two potential danger areas are in a containment relationship and the potential danger identifiers corresponding to the two potential danger areas are the same, the potential danger analysis result of the larger potential danger area is retained, and the potential danger analysis result of the smaller potential danger area is discarded;

[0030] When two potential hazard areas are in an inclusion relationship and the potential hazard identifiers corresponding to the two potential hazard areas are different, the potential hazard analysis results corresponding to the two potential hazard areas are retained.

[0031] The present invention has the following beneficial effects:

[0032] The present invention fuses the hidden danger analysis results generated by multiple hidden danger analysis systems in terms of hidden danger type, coordinate position, confidence, hidden danger area and other latitudes; through this information fusion method, a consistent explanation and description of the hidden danger analysis results is obtained to form a fused analysis result. This reduces the need for manual analysis of the hidden danger results generated by the hidden danger analysis system, reduces the processing difficulty for users, and ultimately effectively reduces the probability of missed hidden danger reports. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The present invention is a flowchart of the multi-source hidden danger analysis information fusion method. DETAILED DESCRIPTION

[0034] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0035] The embodiment of the present invention provides a multi-source hidden danger analysis information fusion method, such as Figure 1 As shown, the method includes:

[0036] S10: Multiple hidden danger analysis systems analyze and identify the same image data respectively to obtain multiple hidden danger analysis results.

[0037] Among them, the hidden danger analysis results include hidden danger area, hidden danger confidence and hidden danger identification.

[0038] S20: Convert various hazard identifications of the same hazard object by different hazard analysis systems into the same hazard identification to achieve hazard type fusion.

[0039] Since different hazard analysis systems define different names or identifiers for the same hazard, it is necessary to merge the hazard types, that is, to convert hazard types of the same type but with different names into the same hazard identifier.

[0040] An implementation of the hidden danger type fusion of the present invention includes:

[0041] S21: Pre-establish a dictionary of hidden danger object names and hidden danger identifiers, and establish a corresponding relationship with the hidden danger identifiers of each hidden danger analysis system.

[0042] For example, the dictionary is created as follows:

[0043]

[0044] S22: According to the dictionary, various hazard identifications of the same hazard object by different hazard analysis systems are uniformly converted into the same hazard identification.

[0045] For example, the hidden danger identifications wjj, wajj, etc. of each hidden danger analysis system of the excavator are uniformly converted into the same hidden danger identification WaJueJi.

[0046] S30: Convert various rectangular coordinate systems of origin defined by different hidden danger analysis systems for the position of the hidden danger object in the image into a preset standard coordinate system.

[0047] Different hidden danger analysis systems define different origin rectangular coordinate systems for the location of hidden danger objects in the image. There are three commonly used methods:

[0048] (1) Take the upper left corner of the image as the origin, the right is the x-axis direction, and the downward is the y-axis direction.

[0049] (2) Take the lower left corner of the image as the origin, the right direction is the x-axis, and the upward direction is the y-axis.

[0050] (3) The center point of the image is taken as the origin, the right is the x-axis direction, and the upward is the y-axis method.

[0051] This step is to fuse the coordinates of the image where the hidden danger is located. This fusion is to fuse the origin rectangular coordinate systems defined by different hidden danger analysis systems for the position of the hidden danger in the image into the same preset standard coordinate system. The preset standard coordinate system can be pre-set or selected from the coordinate systems of multiple hidden danger analysis systems.

[0052] For example: the second and third coordinate modes are converted into the first coordinate mode.

[0053] The specific method is:

[0054] Define the height h and width w of the image;

[0055] Convert the point coordinates (x2, y2) of the second method into the coordinate formula of method 1:

[0056] x1=x2;

[0057] y1=h-x2;

[0058] Convert the point coordinates (x3, y3) of method 3 into the coordinate formula of method 1:

[0059] x1=w / 2+x3;

[0060] y1=h / 2-y3.

[0061] S40: Convert various hidden danger confidence levels of different hidden danger analysis systems into the same standard range.

[0062] Different hidden danger analysis systems use different intervals to express the confidence of hidden danger results. There are two common ones. The first one is [0-1], where 0 means 0% confidence and 1 means 100% confidence. The second one is [0-100], where 0 means 0% confidence and 100 means 100% confidence.

[0063] This step is confidence fusion, which converts the confidence of hidden danger results from different hidden danger analysis systems into the same standard range.

[0064] S50: The hidden danger analysis results are integrated according to the hidden danger areas, hidden danger confidences and hidden danger identifications of different hidden danger analysis systems.

[0065] This step is to merge the hidden danger area and hidden danger results, and merge the hidden danger area and hidden danger results analyzed by the multi-source hidden danger analysis system into one hidden danger area result.

[0066] In one example, this step is implemented as follows:

[0067] S51: When there is no overlap or inclusion relationship between the two potential hazard areas, the potential hazard analysis results corresponding to the two potential hazard areas are retained.

[0068] S52: When two hidden danger areas overlap and have no inclusion relationship, the overlapping area percentage of the two overlapping hidden danger areas is calculated, and the two hidden danger analysis results are merged according to the overlapping area percentage.

[0069] The calculation formula for the overlap area percentage r is as follows:

[0070] r=A / (A1+A2-A)*100%

[0071] Where A is the overlapping area of ​​the two hidden danger areas, and A1 and A2 are the areas of the two hidden danger areas respectively.

[0072] Before fusion, it is necessary to first set a threshold for the percentage of the overlapping area to the total area of ​​the two potential danger areas. The higher the threshold, the more similar the two potential danger areas are considered to be, and the lower the threshold, the more different the two potential danger areas are considered to be. For example, a threshold in the range of [70%-80%] is more reasonable.

[0073] Then, according to the threshold, fusion is performed according to the following process:

[0074] When the percentage of overlapping area exceeds the set threshold and the hidden danger identifications of the two hidden danger analysis results are the same, the hidden danger analysis result with a higher hidden danger confidence is retained, and the hidden danger analysis result with a lower hidden danger confidence is discarded. If the hidden danger confidences are the same, either hidden danger analysis result is retained.

[0075] When the percentage of overlapping area exceeds the set threshold, and the hidden danger identifications of the two hidden danger analysis results are different, the hidden danger area with high hidden danger confidence is retained, and the hidden danger area with low hidden danger confidence is discarded, while the hidden danger identifications of the two hidden danger analysis results are retained. If the hidden danger confidences are the same, either hidden danger area is retained.

[0076] When the overlapping area percentage does not exceed the set threshold, the two hidden danger analysis results are retained.

[0077] S53: When the two hidden danger areas are in a containment relationship, the two hidden danger analysis results are merged according to the hidden danger identifiers corresponding to the two hidden danger areas.

[0078] Specifically, when two potential danger areas are in a containment relationship and the potential danger identifiers corresponding to the two potential danger areas are the same, the potential danger analysis result of the larger potential danger area is retained, and the potential danger analysis result of the smaller potential danger area is discarded.

[0079] When two potential hazard areas are in an inclusion relationship and the potential hazard identifiers corresponding to the two potential hazard areas are different, the potential hazard analysis results corresponding to the two potential hazard areas are retained.

[0080] Through the above steps, the hidden danger information produced by the multi-source hidden danger analysis system is unified in hidden danger type, coordinate origin, confidence standard, hidden danger area and hidden danger result, and finally integrated into a hidden danger analysis result.

[0081] The present invention fuses the hidden danger analysis results generated by multiple hidden danger analysis systems in terms of hidden danger type, coordinate position, confidence, hidden danger area and other latitudes; through this information fusion method, a consistent explanation and description of the hidden danger analysis results is obtained to form a fused analysis result. This reduces the need for manual analysis of the hidden danger results generated by the hidden danger analysis system, reduces the processing difficulty for users, and ultimately effectively reduces the probability of missed hidden danger reports.

[0082] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multi-source hidden danger analysis information fusion method, characterized in that: The method comprises: S10: multiple hidden danger analysis systems analyze and identify the same image data respectively to obtain multiple hidden danger analysis results; The hidden danger analysis results include hidden danger area, hidden danger confidence and hidden danger identification; S20: convert various hidden danger identifications of the same hidden danger object by different hidden danger analysis systems into the same hidden danger identification; S30: converting various rectangular coordinate systems of origin defined by different hidden danger analysis systems for the position of the hidden danger object in the image into a preset standard coordinate system; S40: Convert various hidden danger confidences of different hidden danger analysis systems into the same standard range; S50: The hidden danger analysis results are integrated according to the hidden danger areas, hidden danger confidences and hidden danger identifications of different hidden danger analysis systems.

2. The multi-source hidden danger analysis information fusion method according to claim 1 is characterized in that: The S20 includes: S21: Pre-establish a dictionary of hidden danger object names and hidden danger identifiers, and establish a corresponding relationship with the hidden danger identifiers of each hidden danger analysis system; S22: According to the dictionary, various hidden danger identifications of the same hidden danger object generated by different hidden danger analysis systems are uniformly converted into the same hidden danger identification.

3. The multi-source hidden danger analysis information fusion method according to claim 1 or 2, characterized in that: The S50 includes: S51: when the two potential hazard areas have no overlap or inclusion relationship, retain the potential hazard analysis results corresponding to the two potential hazard areas; S52: when two hidden danger areas overlap and have no inclusion relationship, the overlapping area percentage of the two overlapping hidden danger areas is calculated, and the two hidden danger analysis results are merged according to the overlapping area percentage; S53: When the two hidden danger areas are in a containment relationship, the two hidden danger analysis results are merged according to the hidden danger identifiers corresponding to the two hidden danger areas.

4. The multi-source hidden danger analysis information fusion method according to claim 3 is characterized in that: The calculation formula of the overlapping area percentage r is as follows: r=A / (A1+A2-A)*100% Where A is the overlapping area of ​​the two hidden danger areas, and A1 and A2 are the areas of the two hidden danger areas respectively.

5. The multi-source hidden danger analysis information fusion method according to claim 4 is characterized in that: The S52 includes: When the percentage of overlapping area exceeds the set threshold and the hidden danger identifications of two hidden danger analysis results are the same, the hidden danger analysis result with a higher hidden danger confidence is retained, and the hidden danger analysis result with a lower hidden danger confidence is discarded; When the percentage of overlapping area exceeds the set threshold and the hidden danger identifications of the two hidden danger analysis results are different, the hidden danger area with high hidden danger confidence is retained, and the hidden danger area with low hidden danger confidence is discarded, and the hidden danger identifications of the two hidden danger analysis results are retained; When the overlapping area percentage does not exceed the set threshold, the two hidden danger analysis results are retained.

6. The multi-source hidden danger analysis information fusion method according to claim 5 is characterized in that: The S53 includes: When two potential danger areas are in a containment relationship and the potential danger identifiers corresponding to the two potential danger areas are the same, the potential danger analysis result of the larger potential danger area is retained, and the potential danger analysis result of the smaller potential danger area is discarded; When two potential hazard areas are in an inclusion relationship and the potential hazard identifiers corresponding to the two potential hazard areas are different, the potential hazard analysis results corresponding to the two potential hazard areas are retained.

Images