Single-moving-object real-time detection method in electric wheelchair movement process

Abstract

The invention relates to a single-moving-object real-time detection method in the electric wheelchair movement process. The single-moving-object real-time detection method comprises the steps that firstly, a visual collection device fixed to an electric wheelchair collects a video sequence of a moving object; three continuous-frame images in the video sequence are obtained, and the zooming quantity and translation quantity, relative to the image in the middle frame, of the first-frame image and the zooming quantity and translation quantity, relative to the image in the middle frame, of the end-frame image are registered through phase correlation and Fourier-Mellin transform according to the movement state of the electric wheelchair; then, the outline of a movement area is detected through difference multiplication of the three continuous-frame registered images; a foreground mark template and a background mark template are generated through combination of morphology corrosion, expansion operation and a projection method; a foreground mark and a background mark are extracted through a layered projection method according to the foreground mark template and the background mark; finally, the complete movement area is segmented out through a watershed according to a reconstructed gradient image. The single-moving-object real-time detection method in the electric wheelchair movement process is applicable to the electric wheelchair in the movement state, has good robustness in illumination, and can rapidly, accurately and completely detect the moving object.

Description

technical field [0001] The invention belongs to the fields of computer vision and intelligent wheelchairs, and relates to a real-time detection method for a single moving target under the motion of an electric wheelchair. Background technique [0002] Moving target detection in video sequence images is an important research topic in the field of computer vision and an important part of vision-based intelligent systems. In real life, a large amount of meaningful visual information is contained in motion, and the key to further image analysis and processing is to quickly and accurately separate the moving target from the real-time changing background. Moving target detection is widely used in robotics, artificial intelligence, human-machine interface, intent recognition, military systems and other fields, and its effect will directly affect the follow-up work. Therefore, the video sequence image moving target detection technology has important research value and application s...

AI Technical Summary

Problems solved by technology

[0005] The invention provides a real-time detection method for a single moving target under the motion of an electric wheelchair, which can solve the problems of real-time registration of image sequences, extraction of the contour of the moving target, and detection of the moving target when the electric wheelchair is moving, and has better robustness to light. At the same time, it can quickly, accurately and completely extract the moving target from the image, providing the basis for the analysis and processing of the later intention behavior

Images