This paper proposes a fast moving object detection algorithm based on color single channel proportional difference cumulative projection. This method uses the principle that when the numerator and denominator change in the same direction, the ratio difference is small. The method accumulates and projects the ratio difference between the color single channel and the gray value, and detects and locates the moving object through the projection waveform. Compared with the traditional method, it overcomes the noise problem caused by the slow change of illumination by utilizing the differential idea of the change ratio of the color curve, and achieves the purpose of real-time detection and positioning. The experiment shows the rapidity and feasibility of this method. With the increasing level of automation and the widespread application of monitoring equipment, video motion detection system [7] has become a new research hotspot in computer vision. This system generally requires a large storage space and a fast computing speed. This is because the input of the system is an image sequence that reflects the dynamic changes of the scene, and the data contained is very large. Therefore, video motion detection requires a relatively high algorithm speed. In recent years, people have studied and proposed many algorithms for moving object detection. These include a human-based motion shape model [4] and a neural network based on a snakes model to classify human and non-human shapes [5]. These algorithms are complex and time-consuming. Another type uses grayscale or simple information to directly subtract the previous and next images [3] [6]. When the results are subtracted, a large amount of random noise caused by slight changes in illumination is left behind, and a large amount of information about the moving object, such as color and contour, is discarded, which is limited in practical application. In order to solve the problem of illumination influence in direct subtraction and quickly detect moving objects, we conducted research and found that when the illumination changes slowly, the image color single channel and grayscale value are both affected by illumination and have the same change trend, while their proportional relationship is not greatly affected by illumination changes. Therefore, this paper proposes the idea of color single channel proportional difference. That is, when the target object appears in the next frame, the color single channel value and grayscale value in the two images are compared respectively, and the comparison values are cumulatively projected. The difference between the projections of the two images is the position where the moving object appears.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore