CN109166336B - A real-time road condition information collection and push method based on blockchain technology - Google Patents

Abstract

The invention relates to the technical field of traffic road conditions, in particular to a method for collecting and pushing real-time road condition information based on blockchain technology, comprising the following steps: 1) running a vehicle to install road condition application software on a vehicle terminal or a driver's mobile phone, The driving recorder connected to the vehicle terminal or the driver's mobile phone collects the real-time road condition video of the current road; 2) The road condition video is processed by the video processing module of the road condition application software to obtain the road condition of the running vehicle; The road condition update module of the road condition application software updates the road condition condition in the road condition inquiry module of the road condition application software; 4) The vehicle owner inquires the road condition through the road condition inquiry module to select an appropriate driving route. The real-time road condition information collection and push method of the present invention can make full use of social public resources and participate in the collection of road traffic road condition data through the public, which is not only efficient, but also more flexible than the traditional mechanism.

Description

A real-time road condition information collection and push method based on blockchain technology

technical field

The invention relates to the technical field of traffic road conditions, in particular to a method for collecting and pushing real-time road condition information based on blockchain technology.

Background technique

With the rapid development of the national economy and the accelerated pace of life, more and more people choose to travel by car, which leads to more and more congestion in urban traffic. traffic situation. The traditional real-time road condition information collection system generally sets up monitoring systems or speed monitors in each road section. The collection, release, and update of road condition information are all fully controlled by the traffic department. Such information collection methods not only It takes a lot of manpower and material resources to install the equipment, and the efficiency is low and the real-time performance is poor.

Moreover, the traditional real-time road condition information collection system usually requires a central platform to integrate and process the collected information. With the increase of the data volume, the efficiency of the system will become lower and lower, and there will be many difficulties in management. Therefore, how to collect traffic information in real time and transmit the information to every traffic participant in time is an urgent problem to be solved at present.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the prior art, the present invention proposes a method for collecting and pushing real-time road condition information based on blockchain technology.

The technical solution adopted by the present invention to solve the technical problem is: a method for collecting and pushing real-time road condition information based on blockchain technology, comprising the following steps:

Step 1) Install the road condition application software on the vehicle terminal or the driver's mobile phone by running the vehicle, and collect the real-time road condition video of the current road through the driving recorder connected with the vehicle terminal or the driver's mobile phone;

Step 2) Process the road condition video through the video processing module of the road condition application software to obtain the road condition where the running vehicle is located;

Step 3) updating the road condition condition in the road condition query module of the road condition application software through the road condition update module of the road condition application software;

Step 4) The vehicle owner inquires the road conditions through the road condition inquiry module to select a suitable driving route.

Preferably, the step 2) specifically includes, 21) Real-time segmentation of the road condition video at a time interval of 20 seconds to obtain a short road condition video;

22) Extracting the S-frame to-be-detected pictures of the short video of the road condition at equal intervals;

23) Detect the number of vehicles in each frame of the picture to be detected and the distance between the running vehicle and the vehicle directly ahead in each frame of the picture to be detected, if the number of vehicles in the picture to be detected is greater than the threshold of the number of vehicles and the distance between the vehicles in the picture to be detected is less than the vehicle distance threshold, the road condition where the running vehicle is located is judged as the first congestion, otherwise it is judged as not crowded;

24) If 95% of the pictures to be detected in the frame S to be detected all judge the road condition where the running vehicle is located as the first congestion, then the road condition where the running vehicle is located is judged as the second congestion, Otherwise, it is judged to be not crowded;

25) If the road condition of the running vehicle is judged as the second congested in three consecutive short videos of road conditions, then the road condition of the running vehicle is judged as the third congested, otherwise it is judged as not congested.

Preferably, in the step 3), the road condition application software further includes a navigation module, and when the road condition where the running vehicle is located is judged to be the third congested, the road condition update module combines with the navigation module to update the The running vehicle is displayed in the road condition query module in the form of red highlights.

Preferably, in the step 23), the background difference method is used to calculate the number of vehicles in the to-be-detected picture, which specifically includes the following steps:

231) A road background library is set on a cloud server, and the cloud server is connected to the vehicle terminal or the driver's mobile phone;

232) Call the primary road background image corresponding to the location of the running vehicle according to the navigation module;

233) Combine the primary road background picture with the to-be-detected picture and obtain a final road background picture through a background calculation model;

234) Difference between the to-be-detected picture and the final road background picture to obtain a vehicle picture;

235) Perform vehicle number statistics on the vehicle picture.

Preferably, the background calculation model is established by a road background training set and perfected by a road background test set, and the road background training set and the road background test set are deduplicated by a descriptor matching method, which specifically includes the following steps:

L1. Extract the respective feature points of all the pictures in the road background training set, and calculate the descriptors of the corresponding pictures according to the feature points;

L2. extract a test picture and calculate the test feature point of the test picture in order in the road background test set, and calculate the test descriptor of the test picture according to the test feature point;

L3. obtain N candidate pictures most similar to the test picture in the road background training set according to the test descriptor in conjunction with the DBOW algorithm;

L4. Select a candidate picture from N candidate pictures in order, and match the descriptor of the candidate picture with the test descriptor of the test picture. If the matching result is the same, then the test picture in the test set is matched. Delete and return to step 2); otherwise, continue to perform step 4) until all N candidate pictures are matched and return to step 2).

Preferably, the matching process in the step L4 is specifically:

L41. Perform brute force matching between the test descriptor of the test picture and the descriptor of the candidate picture, and obtain a set of test descriptors matching the test picture and the candidate picture descriptor;

L42. Delete the test descriptor whose distance between the test descriptor set 1 and the closest test descriptor is greater than a certain threshold D to obtain the test descriptor set 2;

L43. Delete the test descriptors that do not meet the rotational invariance in the second test descriptor set to obtain the third test descriptor set;

L44. Delete the test descriptors that do not meet the scaling invariance in the test descriptor set three to obtain the test descriptor set four;

L45. Calculate the number of test descriptors in the test descriptor set four, and when the number is greater than the threshold M, enter the next step; otherwise, determine that the matching result of the test picture and the candidate picture is not the same;

L46. Determine whether the test descriptors in the test descriptor set 4 are matched on the watermark, and if so, determine that the matching results of the test picture and the candidate picture are not the same, otherwise determine that the test picture and the candidate picture match The matching result is the same.

Preferably, the distance between the two test descriptors in the step L42 is the modulus of the difference between the two test descriptors;

The rotation invariance in the step L43 is the angle formed by the test angle formed between a certain test descriptor in the second test descriptor set and any other two test descriptors and the descriptor corresponding to the candidate picture. equal.

Preferably, the scaling invariance calculation process in the step L44 is specifically:

L441 pairs any pair of test descriptors in the test descriptor set three to form multiple groups of test descriptors;

L442 calculates the test distance between two test descriptors in each group of test descriptors, and calculates the candidate distance between the descriptors corresponding to each group of test descriptors in the candidate picture;

L443 calculates the ratio between each set of test distances and the corresponding candidate distances, and calculates the ratio average of all ratios;

L444 makes a difference between the ratio obtained by each group of test descriptors and the average value of the ratio, and when the difference is greater than a certain threshold, the two test descriptors in the group do not meet the scaling invariance.

Preferably, the step L46 is specifically to calculate the average value of the distance between the two test descriptors in the test descriptor set 4. If the average value is less than 10% of the diagonal length of the test picture, it is determined that the The test descriptors in the test descriptor set four match on the watermark, otherwise it is determined that the test descriptors in the test descriptor set four do not match on the watermark.

Preferably, in the N candidate pictures, N is 5, the threshold D is 30, and the threshold M is 10.

The beneficial effect of the present invention is that the real-time road condition information collection and push method of the present invention can make full use of social public resources and participate in the collection of road traffic road condition data through the public, which is not only efficient, but also more flexible than the traditional mechanism; Going to the central platform in the traditional sense, through the form of distributed data storage and point-to-point transmission, the application of the driving recorder enables every traffic participant to directly participate in the collection process of road condition information.

Description of drawings

FIG. 1 is a flowchart of a method for collecting and pushing real-time road condition information based on blockchain technology according to the present invention.

Detailed ways

The technical solutions of the present invention are further described below with reference to the accompanying drawings and through specific embodiments.

As shown in Figure 1, a method for collecting and pushing real-time road condition information based on blockchain technology includes the following steps:

Step 1) Install the road condition application software on the vehicle terminal or the driver's mobile phone by running the vehicle, and collect the real-time road condition video of the current road through the driving recorder connected with the vehicle terminal or the driver's mobile phone.

Step 2) The road condition video is processed by the video processing module of the road condition application software to obtain the road condition where the running vehicle is located. The video processing module only processes the road condition video collected by its corresponding running vehicle. Specifically, 21) segment the road condition video in real time at 20-second time intervals to obtain a short road condition video. 22) Extract the S frames of the pictures to be detected in the short video of the road condition at equal intervals. 23) Detect the number of vehicles in each frame of the picture to be detected and the distance between the running vehicle and the vehicle directly ahead in each frame of the picture to be detected, if the number of vehicles in the picture to be detected is greater than the threshold of the number of vehicles and the distance between the vehicles in the picture to be detected If it is less than the vehicle distance threshold, the road condition where the running vehicle is located is judged as the first congestion, otherwise it is judged as not crowded. The threshold for the number of vehicles may be 5, and the threshold for the distance between vehicles may be 4 meters. 24) If 95% of the pictures to be detected in the frame S to be detected all judge the road condition where the running vehicle is located as the first congestion, then the road condition where the running vehicle is located is judged as the second congestion, Otherwise, it is judged not to be crowded. 25) If the road condition of the running vehicle is judged as the second congested in three consecutive short videos of road conditions, then the road condition of the running vehicle is judged as the third congested, otherwise it is judged as not congested.

Step 3) Update the road condition condition in the road condition query module of the road condition application software through the road condition update module of the road condition application software. The road condition application software also includes a navigation module, when the road condition where the running vehicle is located is judged to be the third congestion, the road condition updating module combines with the navigation module to display the running vehicle in the form of a red bright spot on the running vehicle. Describe the road condition query module. The road condition update module will update all road condition application software with the latest road condition information of the corresponding running vehicle. The running vehicles whose road conditions are the first crowded, the second crowded and the uncrowded are not displayed, and only the running vehicles whose road conditions are the third crowded are displayed in the form of red highlights in the road condition query module. When the road conditions of the 3rd congestion change to other road conditions, the corresponding red highlights will disappear.

Step 4) The vehicle owner inquires the road conditions through the road condition inquiry module to select a suitable driving route. The car owner can query the road condition information of the relevant road through the road condition query module. For example, when there are many red bright spots in a certain road section, the car owner can choose other roads to drive.

In the step 23), the number of vehicles in the picture to be detected is calculated by the background difference method, which specifically includes the following steps:

231) A road background library is set on a cloud server, and the cloud server is connected to the vehicle terminal or the driver's mobile phone. 232) Call the primary road background image corresponding to the location of the running vehicle according to the navigation module. A primary road background image is used within a certain range for the same road section and the same lane. 233) Combine the primary road background picture with the to-be-detected picture and obtain a final road background picture through a background calculation model. The primary road background image is optimized by the background calculation model and the information of the image to be detected to obtain the final road background image. The background computational model collects existing models. 234) Difference between the image to be detected and the final road background image to obtain a vehicle image. 235) Perform vehicle number statistics on the vehicle picture.

The background calculation model is established by the road background training set and perfected by the road background test set. The road background training set and the road background test set are deduplicated by the descriptor matching method to avoid over-fitting, which specifically includes the following: Step, L1. Extract the respective feature points of all the pictures in the road background training set, and calculate the descriptors of the corresponding pictures according to the feature points. The feature point extraction can use the FAST algorithm, and the descriptor calculation can use the ORB algorithm. Feature points are those points in the image where the light and dark changes are prominent. The descriptor is the identification of each feature point, which is generally obtained by extracting the pixel information near the feature point. It can be used to determine which feature points in different pictures correspond to the same points on the object. This process is called feature point matching. One of the characteristics of the descriptor is that after the image is zoomed, rotated, and translated, the descriptor does not change, and the descriptor is finally presented in the form of a vector, such as 64 numbers of 1 byte.

L2. Extract a test picture from the road background test set in sequence, calculate the test feature points of the test picture, and calculate the test descriptor of the test picture according to the test feature points. The test feature point extraction can use the FAST algorithm, and the test descriptor calculation can use the ORB algorithm.

L3. Obtain N candidate pictures most similar to the test picture in the road background training set according to the test descriptor in combination with the DBOW algorithm. In order to speed up the matching speed of descriptors, this method also uses the DBOW algorithm to speed up. Usually, the number of road background training set images is in the hundreds of thousands, and the number of road background test sets is in the tens of thousands. If the brute force matching is performed directly two by two, it will take a lot of time and greatly reduce the matching efficiency of the present invention. After using the DBOW algorithm, only one comparison is needed to find the most similar pictures, and then it is determined whether the two pictures are the same slice through the matching of descriptors and the positional relationship of matching feature points. The DBOW algorithm is an efficient loop closure detection algorithm. The full name of the DBOW algorithm is Bags of binary words for fast place recognition in image sequence. The feature detection algorithm used is FAST, and the descriptor uses the Brief descriptor.

L4. Select a candidate picture from N candidate pictures in order, and match the descriptor of the candidate picture with the test descriptor of the test picture. If the matching result is the same, then the test picture in the test set is matched. Delete and return to step 2); otherwise, continue to perform step 4) until all N candidate pictures are matched and return to step 2).

Through the matching degree of the descriptors between the two pictures, it can be judged whether the pictures are the same. For example, the number of feature points that can be matched on the same picture is much larger than that of different pictures. And because of the translation, scaling, and rotation invariance of the descriptor, even if the image is deformed, it can be matched. And because the descriptors are local features, after adding different watermarks to the image, they do not affect the matching. The matching process in the step L4 is specifically, L41. Perform violent matching between the test descriptor of the test picture and the descriptor of the candidate picture, and obtain the test descriptor matching the test picture and the candidate picture descriptor. Collection one. Simply matching descriptors is not completely reliable. Those wrong matches are called mismatches. When matching correctly, no matter how the image is deformed, the relative relationship between the feature points remains unchanged, but the feature points that are mismatched are disorganized, so you can pass This rule is used to rule out mismatches.

L42. Delete the test descriptors whose distance from the test descriptor set with the closest distance is greater than a certain threshold D in the test descriptor set 1 to obtain the test descriptor set 2. In step L42, the distance between the two test descriptors is the modulus of the difference between the two test descriptors. In this step, independent matching feature points can be deleted.

L43. Delete the test descriptors in the second test descriptor set that do not meet the rotation invariance to obtain the third test descriptor set. The rotation invariance in step L43 is that the test angle formed between a certain test descriptor in the second test descriptor set and any other two test descriptors is equal to the angle formed by the descriptor corresponding to the candidate picture. For example, the candidate picture has feature point 1, feature point 2 and feature point 3, the three feature points are connected to form triangle 1, the test picture has test feature point 1, test feature point 2 and test feature point 3, three test feature points After the connection, triangle 2 is formed, and feature point 1 matches with test feature point 1, feature point 2 matches with test feature point 2, and feature point 3 matches with test feature point 3. Assuming that the candidate picture and the test picture are the same, triangle 1 matches with the test feature point 3. The shape of triangle 2 is the same, otherwise, the shape of triangle 1 and triangle 2 are very likely to be different.

L44. Delete the test descriptors in the third test descriptor set that do not meet the scaling invariance to obtain the fourth test descriptor set. The scaling invariance calculation process in step L44 is specifically: L441 pairs any two test descriptors in the test descriptor set three to form multiple groups of test descriptors; L442 calculates the difference between two test descriptors in each group of test descriptors. L443 calculates the ratio between each set of test distances and the corresponding candidate distances, and calculates the ratio average of all ratios value; L444 makes the difference between the ratio obtained by each group of test descriptors and the average value of the ratio, and when the difference is greater than a certain threshold, the two test descriptors in this group do not meet the scaling invariance.

For example, the candidate picture has feature point d1, feature point d2, feature point d3 and feature point d4, the test picture has test feature point s1, test feature point s2, test feature point s3 and test feature point s4, and feature point d1 and test feature point s4 The feature point s1 matches, the feature point d2 matches the test feature point s2, the feature point d3 matches the test feature point s3, and the feature point d4 matches the test feature point s4, wherein the distance ratio between the test feature point s1 and the test feature point s2 The distance between the upper feature point d1 and the feature point d2 is the ratio 1, and the distance between the test feature point s3 and the test feature point s4 is the ratio 2 compared with the distance between the upper feature point d3 and the feature point d4. The test images are the same, then the ratio 1 and the ratio 2 are the same. The present invention first obtains the average value of the ratio, and then compares the actual ratio with the average ratio, so as to determine whether the descriptor of the test feature point and the descriptor of the candidate picture really match.

L45. Calculate the number of test descriptors in the test descriptor set 4, when the number is greater than the threshold M, go to the next step, otherwise it is determined that the matching results of the test picture and the candidate picture are not the same.

L46. Determine whether the test descriptors in the test descriptor set 4 are matched on the watermark, and if so, determine that the matching results of the test picture and the candidate picture are not the same, otherwise determine that the test picture and the candidate picture match The matching result is the same. Step L46 is specifically to calculate the average value of the distance between the test descriptors in the test descriptor set 4. If the average value is less than 10% of the diagonal length of the test picture, it is determined that the test descriptor set 4 The test descriptor in the set matches the watermark, otherwise it is determined that the test descriptor in the test descriptor set four does not match the watermark.

The value of N in the N candidate pictures may be 5, the value of the threshold D may be 30, and the value of the threshold M may be 10.

The above-mentioned embodiments are only to describe the preferred embodiments of the present invention, and do not limit the concept and scope of the present invention. Under the premise of not departing from the design concept of the present invention, various modifications and improvements made by those of ordinary skill in the art to the technical solutions of the present invention should fall within the protection scope of the present invention, and the technical content claimed in the present invention has been fully recorded in the claims.

Claims (2)

1. a real-time road condition information collection and push method based on block chain technology, is characterized in that: comprise the following steps, Step 1) Install the road condition application software on the vehicle terminal or the driver's mobile phone by running the vehicle, and collect the real-time road condition video of the current road through the driving recorder connected with the vehicle terminal or the driver's mobile phone; Step 2) Process the road condition video through the video processing module of the road condition application software to obtain the road condition where the running vehicle is located; Step 3) updating the road condition condition in the road condition query module of the road condition application software through the road condition update module of the road condition application software; Step 4) The car owner inquires the road conditions through the road condition inquiry module to select a suitable driving route; The step 2) specifically includes, 21) Real-time segmentation of the road condition video at a time interval of 20 seconds to obtain a short road condition video; 22) Extracting the S-frame to-be-detected pictures of the short video of the road condition at equal intervals; 23) Detect the number of vehicles in each frame of the picture to be detected and the distance between the running vehicle and the vehicle directly ahead in each frame of the picture to be detected, if the number of vehicles in the picture to be detected is greater than the threshold of the number of vehicles and the distance between the vehicles in the picture to be detected is less than the vehicle distance threshold, the road condition where the running vehicle is located is judged as the first congestion, otherwise it is judged as not crowded; 24) If 95% of the pictures to be detected in the frame S to be detected all judge the road condition where the running vehicle is located as the first congestion, then the road condition where the running vehicle is located is judged as the second congestion, Otherwise, it is judged to be not crowded; 25) If three consecutive short videos of road conditions all judge the road condition where the running vehicle is located as the second congestion, then the road condition where the running vehicle is located is judged as the third congestion, otherwise it is judged as not crowded; In the step 3), the road condition application software further includes a navigation module, when the road condition where the running vehicle is located is judged to be the third congested, the road condition update module combines the navigation module to update the running vehicle to the third congested. The form of red highlights is displayed in the road condition query module; In the step 23), the number of vehicles in the picture to be detected is calculated by the background difference method, which specifically includes the following steps: 231) A road background library is set on a cloud server, and the cloud server is connected to the vehicle terminal or the driver's mobile phone; 232) Call the primary road background image corresponding to the location of the running vehicle according to the navigation module; 233) Combine the primary road background picture with the to-be-detected picture and obtain a final road background picture through a background calculation model; 234) Difference between the to-be-detected picture and the final road background picture to obtain a vehicle picture; 235) Counting the number of vehicles on the vehicle picture; The background calculation model is established through a road background training set and perfected through a road background test set, and the road background training set and the road background test set are deduplicated by a descriptor matching method, which specifically includes the following steps: L1. Extract the respective feature points of all the pictures in the road background training set, and calculate the descriptors of the corresponding pictures according to the feature points; L2. extract a test picture and calculate the test feature point of the test picture in order in the road background test set, and calculate the test descriptor of the test picture according to the test feature point; L3. obtain N candidate pictures most similar to the test picture in the road background training set according to the test descriptor in conjunction with the DBOW algorithm; L4. Select a candidate picture from N candidate pictures in order, and match the descriptor of the candidate picture with the test descriptor of the test picture. If the matching result is the same, then the test picture in the test set is matched. Delete and return to step L2; otherwise, continue to perform step L4 to N candidate pictures are all matched and return to step L2; The matching process in the step L4 is specifically, L41. Perform brute force matching between the test descriptor of the test picture and the descriptor of the candidate picture, and obtain a set of test descriptors matching the test picture and the candidate picture descriptor; L42. Delete the test descriptor whose distance between the test descriptor set 1 and the closest test descriptor is greater than a certain threshold D to obtain the test descriptor set 2; L43. Delete the test descriptors that do not meet the rotational invariance in the second test descriptor set to obtain the third test descriptor set; L44. Delete the test descriptors that do not meet the scaling invariance in the test descriptor set three to obtain the test descriptor set four; L45. Calculate the number of test descriptors in the test descriptor set four, and when the number is greater than the threshold M, enter the next step; otherwise, it is determined that the matching results of the test picture and the candidate picture are not the same; L46. Determine whether the test descriptors in the test descriptor set 4 are matched on the watermark, and if yes, determine that the matching results of the test picture and the candidate picture are different, otherwise determine that the test picture and the candidate picture match The matching result is the same; The distance between the two test descriptors in the step L42 is the modulus of the difference between the two test descriptors; The rotation invariance in the step L43 is the angle formed by the test angle formed between a certain test descriptor in the second test descriptor set and any other two test descriptors and the descriptor corresponding to the candidate picture. equal; The scaling invariance calculation process in the step L44 is specifically: L441 pairs any pair of test descriptors in the test descriptor set three to form multiple groups of test descriptors; L442 calculates the test distance between two test descriptors in each group of test descriptors, and calculates the candidate distance between the descriptors corresponding to each group of test descriptors in the candidate picture; L443 calculates the ratio between each set of test distances and the corresponding candidate distances, and calculates the ratio average of all ratios; L444 makes a difference between the ratio obtained by each group of test descriptors and the average value of the ratio, and when the difference is greater than a certain threshold, the two test descriptors in the group do not meet the scaling invariance; The step L46 is specifically to calculate the average value of the distance between the two test descriptors in the test descriptor set 4. If the average value is less than 10% of the diagonal length of the test picture, the test descriptor is determined. The test descriptor in set four matches the watermark, otherwise it is determined that the test descriptor in the test descriptor set four does not match on the watermark. 2. a kind of real-time road condition information collection and push method based on blockchain technology according to claim 1, is characterized in that: in N candidate pictures, N is 5, the threshold D is 30, and the threshold M is 30. is 10.

Images