Algorithm of Moving Vehicle Detection by Improved Gaussian Mixture Model Based on ROI

Abstract

Detection of moving object is a hot topic in computer vision. Traditionally, it is detected for every pixel in whole image by Gaussian mixture background model, which may waste more time and space. In order to improving the computational efficiency, an advanced Gaussian mixture model based on Region of Interest was proposed. Firstly, the solution finds out the most probably region where the target may turn up. And then Gaussian mixture background model is built in this area. Finally, morphological filter algorithm is used for improving integrity of the detected targets. Results show that the improved method could have a more perfect detection but no more time increasing than typical method.

Research on real-time perception method of key targets in the comprehensive excavation working face of coal mine

“…Traditional target detection models realize target detection by manually extracting target features and separating the complex background of the coal mine. Aiming at the poor quality of coal mine surveillance video images, with much noise and lighting prone to sudden changes, [1] used an improved hybrid Gaussian model to extract the video background to inhibit the interference of complex background information, and to realize target detection in the process of transportation operations. Du and Hao [2] proposed an edge detection method that fuses an improved holistic nested edge detection neural network with the Canny operator for detecting large foreign objects during the transportation of coal streams.…”

Moving Object Detection Based on Improved Background Updating Method for Gaussian Mixture Model

“…Reference [8] introduces illumination variation factor, adopting self-adapting learning rate to update background model, adopting a small learning rate when the illumination breaks solely, and a large learning rate when the illumination breaks quickly. When the illumination mutation is underway, reference [9] use Frame difference to detect moving object, meantime we update the background model at a high speed. All the methods above can improve learning ability to illumination mutation, but still they can't update the background model timely and effectively.…”