Passenger flow tracking method

Abstract

The invention discloses a passenger flow tracking method, and belongs to the technical field of image processing. The method specifically comprises the steps of obtaining multiple frames of continuous images shot by a camera, and solving a depth map through stereo matching; eliminating background interference, and segmenting each human head into an independent connected region; constructing a plurality of people counting structural bodies; calculating the minimum value of the sum of optimal matching different rows and different columns of the bipartite graphs of the plurality of two-dimensional arrays by using a KM algorithm; and solving the Euclidean distance between the center coordinate of the connected region of the previous frame and the first point of the track point of the structure body. When the method is used for tracking passenger flow, even if passengers get on the bus shake in a visible area, the tracking algorithm can still track the passengers all the time and is not easy to lose, the method is not only suitable for linear motion, but also suitable for tracking curvilinear motion and random motion, and the algorithm can realize real-time and accurate tracking of multiple persons even if the passengers get on and off the bus are crowded, and the missing metering and overmetering problems are reduced.

Description

Technical field[0001]The present invention relates to the field of image processing, and more particularly to a tracking method of passenger flow.Background technique[0002]The bus is an important public transport in people. Through the statistics of the passenger flow, we can learn about the traffic of each time, each bus, each station, etc., using these data intelligent scheduling, allowing public resources to be more fully utilized, the accuracy of passenger flow statistics Directly affect the operation of intelligent transportation and bus operations. By showing the status of the current passenger and changes, the bus stop can take measures to prevent emergencies in situations of the number of people.[0003]The existing passenger flow statistics technology main means there is a weighing statistical, single-graphic image statistics, and binocular image statistics, which is the most accurate due to its additional stereo height (distance) information on the basis of single-dimensiona...

AI Technical Summary

Problems solved by technology

[0004] Specifically, because the occupants in the car shake in the recognition area, the existing tracking algorithms are likely to cause miscalculations. For example, some people use Kalman filter tracking, which is only suitable for straight line tracking. The shaking of people is a random movement, so At this time, Kalman filter tracking is used to track the shaking of the identification area, which is often intermittent, which is easy to cause overcounting, and the accuracy rate is not high

[0005] In addition, in a very crowded situation, the mutual interference between passengers, using particle filter, TLD and other tracking, not only the effect on the depth map or black and white binary image is not ideal, but also the power consumption of this kind of algorithm is relatively high, generally The real-time effect cannot be achieved on the embedded board, therefore, the present invention is produced in order to solve these problems

Images