Video Inpainting for Arbitrary Foreground Object Removal

Siddique, Ashraf; Lee, Seungkyu

doi:10.1109/wacv.2018.00195

Cited by 3 publications

(4 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The depth map of each frame is obtained by projecting the 3D points onto the 2dimensional image space of each frame. The depth value at each pixel is calculated by interpolating the depth of projected 3D points falling within a patch of the fixed size described in Reference [24]. We simply perform a depth calculation for the points from overlaid multiple surfaces by selecting the closest group of points within the patch.…”

Section: Depth Extractionmentioning

confidence: 99%

“…NDD still provides a linear measurement of the discrepancy between the depth (observed distance) and 3D distance (predicted distance) assigning consistent confidence scores for identical points. Figure 8a and Equation (5) demonstrate how NDD is really calculated in method [24].…”

Section: Normalized Difference Of Depth-3d Distancementioning

confidence: 99%

“…However, in this method, the approximate NND distance is calculated from the ratio between q1 and q2 image space distance which is not compatible with the distance ratio in the 3D space. Instead of approximating the NDD distance from 2D space from Reference [24], we calculate NDD in 3D space which is illustrated in Figure 8b…”

Section: Q1+q2mentioning

confidence: 99%

“…In this paper, we propose a novel and robust video editing method under several challenging background situations like complex visual pattern, occlusion, overlaid object clutter, and depth variation observed from a moving camera which is the extension of previous work [24]. First, our method obtains 3D point cloud and camera translation matrix of video frames using a Structure From Motion (SfM) method.…”

Section: Introductionmentioning

confidence: 98%

See 3 more Smart Citations

Object-Wise Video Editing

Siddique

Lee

2021

Applied Sciences

Self Cite

View full text Add to dashboard Cite

Beyond time frame editing in video data, object level video editing is a challenging task; such as object removal in a video or viewpoint changes. These tasks involve dynamic object segmentation, novel view video synthesis and background inpainting. Background inpainting is a task of the reconstruction of unseen regions presented by object removal or viewpoint change. In this paper, we propose a video editing method including foreground object removal background inpainting and novel view video synthesis under challenging conditions such as complex visual pattern, occlusion, overlaid clutter and variation of depth in a moving camera. Our proposed method calculates a weighted confidence score on the basis of normalized difference between observed depth and predicted distance in 3D space. A set of potential points from epipolar lines from neighbor frames are collected, refined, and weighted to select a few number of highly qualified observations to fill the desired region of interest area in the current frame from video. Based on the background inpainting method, novel view video synthesis is conducted with arbitrary viewpoint. Our method is evaluated with both a public dataset and our own video clips and compared with multiple state of the art methods showing a superior performance.

show abstract

Section: Depth Extractionmentioning

confidence: 99%

Section: Normalized Difference Of Depth-3d Distancementioning

confidence: 99%

Section: Q1+q2mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 98%

See 2 more Smart Citations

Object-Wise Video Editing

Siddique

Lee

2021

Applied Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

Hybridized Cuckoo Search with Multi-Verse Optimization-Based Patch Matching and Deep Learning Concept for Enhancing Video Inpainting

Rao

Chakrapani²,

Kumar

2021

The Computer Journal

View full text Add to dashboard Cite

This paper aims to develop a novel deep learning concept to deal with video inpainting. Initially, motion tracking is performed, which is the process of determining motion vectors that describe the transformation from adjacent frames in a video sequence. Further, the regions or patches of each frame are categorized using the optimized recurrent neural network (RNN), in which the region is split into a smooth and structure region. It is performed using the texture feature called grey-level co-occurrence matrix. The filling of the smooth region is accomplished by replacing with the mean pixels of unmasked region, and the structure region is done by optimized patch matching approach based on scale-invariant feature transform (SIFT). The main objective optimized patch matching is based on the minimized Euclidean distance between the extracted SIFT features of the original patch and reference patch, and the certain patch is inpainted by the optimized patch. Here, the hybridization of two meta-heuristic algorithms like cuckoo search algorithm and multi-verse optimization (MVO) called Cuckoo Search-based MVO is used to optimize the RNN and patch matching. Finally, the experimental results verify the reliability of the proposed algorithm over existing algorithms.

show abstract

Automatic Occlusion Removal System using Optical Flow Method

Aneesha¹,

Helen²

2018

ijcse

View full text Add to dashboard Cite

Video Inpainting for Arbitrary Foreground Object Removal

Cited by 3 publications

References 13 publications

Object-Wise Video Editing

Object-Wise Video Editing

Hybridized Cuckoo Search with Multi-Verse Optimization-Based Patch Matching and Deep Learning Concept for Enhancing Video Inpainting

Automatic Occlusion Removal System using Optical Flow Method

Contact Info

Product

Resources

About