Moving Object Detection by Fuzzy Aggregation Using Low Rank Weightage Representation

摘要

We envisage a new algorithm, to detect moving objects having dynamic and challenging background conditions, by applying low rank weightage and fuzzy aggregated multi-feature similarity method. Model level fuzzy aggregation measure driven background model maintenance is used to ensure more robustness. The model and current feature vectors are evaluated between corresponding elements to find out the similarity functions. To compute fuzzy similarities from the ordered similarity function values for each model concepts of Sugeno and Choquet integrals are incorporated in our algorithm. A fuzzy integral set is using model updating and foreground/background classification decision methods. Sugeno Integral calculates only minimum and maximum weightage. We use choquet concept because it has the same functionality as Sugeno but it also uses additional operations like arithmetic mean and Ordered Weighted Averaging (OWA). Here we explain to segment the object by fuzzy aggregation with low rank weightage concept for extracting moving objects with accurate shape in dynamic background. PSNR, MSE and SSIM values are calculated to do performance evaluation.