ViBe (Visual Background Extractor) algorithm and SLIC (Simple Linear Iterative Cluster) superpixel based background difference method

Abstract

A background difference method based on ViBe algorithm and SLIC superpixel belongs to the technical field of image processing. The invention comprises the following steps: background modeling based on superpixel segmentation, foreground detection and background update. The invention proposes a background modeling foreground detection algorithm combined with SLIC superpixels under the ViBe algorithm framework. While effectively utilizing the advantages of the ViBe algorithm, it expands the neighborhood range of pixels by combining with SLIC superpixels, and makes full use of the spatial correlation of background pixels so that it can better deal with phenomena such as wind and grass and camera shake. The background update method of ViBe algorithm for random replacement of the background library is changed to the update method of Gaussian distribution, which avoids the instability caused by the random replacement strategy to the algorithm. In the experimental part, three sets of frame sequences in the I2R dataset are used for background modeling and foreground detection experiments, and the GMM algorithm and ViBe algorithm are used to compare with our algorithm.

Description

technical field [0001] The invention relates to a background difference method based on a ViBe algorithm and a SLIC superpixel, and belongs to the technical field of image processing. Background technique [0002] In recent years, with the increasing demand for real-time monitoring, background modeling and foreground detection technology has been widely used in the field of video surveillance. At the same time, there are endless types of monitoring scenarios, from indoors to outdoors, from static scenes to complex scenes with strong disturbances. However, complex scenes also pose some challenges for background modeling for foreground detection. [0003] Illumination mutation: The illumination mutation causes strong changes in the background pixels, so it is easy to be misjudged as the foreground during foreground detection. This situation is very common in outdoor scenes in cloudy weather. [0004] Background object displacement: When there is background object displaceme...

AI Technical Summary

Problems solved by technology

However, whether it is dealing with 4 neighborhoods or 8 neighborhoods, the neighborhood range of a single pixel is still very limited, so it cannot cope well with complex scenes with strong disturbances in the background.

Moreover, in order to make the algorithm faster, ViBe uses a random method to update the background library, which brings a certain degree of instability to the algorithm.

Images