Wavelet-based Moving Object Segmentation: From Scalar Wavelets to Dual-tree Complex Filter Banks

A framework for moving object segmentation under rapidly changing illumination conditions in complex wavelet domain

“…In the proposed approach, we addressed the issues mentioned [11][12][13][14][15][16][17] using dynamic background modeling step in complex wavelet domain. In this proposed approach, we have use six major steps applied on given video frames which include: wavelet de-composition of frame using complex wavelet transform; use of change detection on detail coefficients (LH, HL, HH); use of improved Gaussian mixture based dynamic background modeling on approximate co-efficient (LL sub-band); use of soft thresholding for noise removal; strong edge detection; inverse wavelet transformation for reconstruction; and finally using closing morphology operator.…”

“…Huang et al [13][14] proposed an algorithm for moving object segmentation using single change detection and double change detection applied in wavelet domain. Baradarani [15][16] refined the work of Huang et al [13][14] using dual tree complex filter bank in wavelet domain. Khare et al [17] also refine the work of Baradarani and Huang using Daubechies complex wavelet.…”

“…All the above discussed methods using change detection in wavelet domain to detect the moving object [11][12][13][14][15][16][17] they do not use any background modeling step to adapt the dynamic background due to various problems such as ghost, shadow, and noise. Motivated by these facts, we have proposed a new method for segmentation of moving object which solve the above problem of [11][12][13][14][15][16][17] using change detection and background modeling applied on Daubechies complex wavelet transform using LL sub-band. The proposed method is compared with other methods namely: Khare et al [17], Kim et al method [6] and Hsia et al method [29].…”

A Framework of Moving Object Segmentation in Maritime Surveillance Inside a Dynamic Background

“…Huang et al [11,12] proposed an algorithm for moving object segmentation to solve the double-edge problem in the spatial domain using a change detection method with different thresholds in four wavelet sub-bands. Baradarani [13,14] refined the work of Huang et al [11,12] using dual tree complex filter bank in wavelet domain. These methods [11][12][13][14] suffer from the problem of noise disturbances and distortion of moving segmented object due to change in speed of objects.…”

“…These methods [11][12][13][14] suffer from the problem of noise disturbances and distortion of moving segmented object due to change in speed of objects. Khare et al [15] refine the work of Baradarani [13,14] and Huang et al [11,12] using Daubechies complex wavelet. The method proposed by Khare et al [15] reduces the noise disturbance and speed change, but it suffers from the problem of dynamic background changes and shadow detection and due to this segmenting coherence occurs [16].…”