Crowd counting method and device based on high and low altitude information fusion, and storage medium

Abstract

The invention discloses a crowd counting method and device based on high and low altitude information fusion, and a storage medium. The method comprises the following steps: constructing a crowd counting model; inputting the high-altitude image into a crowd counting model to obtain a crowd distribution density map; mapping the area of the high-altitude image to the area of the corresponding low-altitude image to obtain a registration image; discretizing the crowd distribution density map into a plurality of density levels, and counting the number of pixels included in each density level; selecting a sub-group included in each density level in the area of the high-altitude image, and counting the number of pixels of the sub-group; inputting the low-altitude image region corresponding to the sub-group into a crowd counting model to obtain the number of people of the sub-group; and according to the number of pixels included in each density level, the number of pixels of the sub-groups and the number of people of the sub-groups, obtaining the global number of people in the specified area. By adopting the technical scheme of the invention, the method is suitable for large-scale and high-density crowd scenes, and the accuracy of crowd counting is greatly improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a crowd counting method, device and storage medium based on high and low altitude information fusion. Background technique [0002] Crowd counting is a very important technique in the public security industry. For a given image or a video, it is automatically processed by a computer to analyze the number of people in it. With the rapid development of society, the traditional manual counting method cannot accurately and quickly obtain the number of people in large-scale scenarios, which brings greater difficulty to security control. The crowd counting algorithm based on machine vision has become one of the current research hotspots due to its technical advantages such as accuracy, real-time performance, and stability. [0003] At present, crowd counting technology has achieved rapid development, and a large number of scholars have devoted themselves to the r...

AI Technical Summary

Problems solved by technology

[0004] In image acquisition, it is often difficult to obtain clear and complete crowd images, and there are serious occlusions or light effects, and the effect of traditional algorithms is not ideal; most of the existing crowd counting algorithms are processed for single-view images or videos with clear pedestrian features However, in the actual crowd control scene, there are more high-density and high-area scenes. The monitoring images contain more people, the occlusion is serious, the image quality is not good, and the coordinate position of each head cannot be accurately marked.

Images