High-Accuracy and Quick Matting Based on Sample-Pair Refinement and Local Optimization

Abstract

SUMMARYBased on sample-pair refinement and local optimization, this paper proposes a high-accuracy and quick matting algorithm. First, in order to gather foreground/background samples effectively, we shoot rays in hybrid (gradient and uniform) directions. This strategy utilizes the prior knowledge to adjust the directions for effective searching. Second, we refine sample-pairs of pixels by taking into account neighbors'. Both high confidence sample-pairs and usable foreground/background components are utilized and thus more accurate and smoother matting results are achieved. Third, to reduce the computational cost of sample-pair selection in coarse matting, this paper proposes an adaptive sample clustering approach. Most redundant samples are eliminated adaptively, where the computational cost decreases significantly. Finally, we convert fine matting into a de-noising problem, which is optimized by minimizing the observation and state errors iteratively and locally. This leads to less space and time complexity compared with global optimization. Experiments demonstrate that we outperform other state-of-the-art methods in local matting both on accuracy and efficiency.

Confidence-aware guided image filter

“…All top performing methods conduct an initial per-pixel estimation of the alpha matte followed by a post-processing step to align the initial alpha matte to the structures (i.e. color gradients) visible in the image [6][7][8][9][10][11][12].…”

“…The confidence values lie in the interval (0..1] with an average confidence of about 0.4. For more details on how the Global Matting algorithm and alternative methods compute the confidence values, the interested reader is referred to [6][7][8][9][10][11][12].…”

A Comprehensive Survey on Sampling‐Based Image Matting

“…Global pattern : SRLO [HWS*13], WCT [SR12], Shared [GO10] and Global [HRR*11], which are shown from sparse to dense in Figure (a). In this pattern, each unknown pixel shoots several rays and acquires samples from the intersections between these rays and boundaries and .…”

“…Consequently, sampling‐based matting is currently showing better prospects and receiving increasing research attentions. In contrast, affinity‐based matting mainly plays a pre‐processing role based on sampling‐based one to smooth the final alpha mattes [WC07b, RRRAS08, RRG08, RRKG10, GO10, HWS*13, HRR*11, CM13, ZZX12, KEE15, JVCR16, FLZ16].…”

Accurate depth sensing by filtering IR cost on the guide of RGB information

“…The 3D information can be used in scene modeling [1,2], humancomputer interaction [3], movie making [4,5] and so on.…”