The advent of convolutional neural networks has led to remarkable progress in dense stereo labeling problem,achieving superior performance over the traditional methods.However,the ill-posed nature of stereo matching m...The advent of convolutional neural networks has led to remarkable progress in dense stereo labeling problem,achieving superior performance over the traditional methods.However,the ill-posed nature of stereo matching makes noise(outliers)in winner-takes-all(WTA)disparity maps inevitable.This paper presents a robust statistical approach to noise detection and refinement of WTA disparity maps.In the context of noise detection,the input noisy WTA disparity map is segmented into regular grid cells(regions)with the aim of leveraging Markov random field(MRF)to infer candidate disparity labels.However,there are two key problems:there can be large severe outliers in the regions;second,the regular partition process may produce regions with mixed disparity distributions.To overcome these problems,we optimize a robust objective function over the segmented disparity map.By obtaining the optimal solution of the objective function through a maximum a posteriori estimation in a probabilistic model,we are able to infer MRF candidate disparity labels.We then apply a soft-segmentation constraint on the estimated MRF candidate disparity labels to describe and detect outliers in the disparity map.Next,an edge-preserving statistical inference that leverages the joint statistics of the disparity map and its guidance reference image is used to select correct candidate disparity for each detected outlier.Finally,a weighted median filter is applied to remove small spikes and irregularities in the resulting disparity map.Rigorous and comprehensive experiments showed that the proposed method is distributionally robust and outlier resistant,and can effectively detect and correct outliers in disparity maps.Middlebury evaluation benchmark validated the competitive performance of the proposed method.展开更多
This paper presents a new robust global motion estimation method based on pre-analysis of the video content. The novel idea in the proposed method, compared to classical robust statistics-based estimation methods, is...This paper presents a new robust global motion estimation method based on pre-analysis of the video content. The novel idea in the proposed method, compared to classical robust statistics-based estimation methods, is to classify the video sequences into 3 classes based on the analysis of scene content before motion estimation. Different motion models and estimation methods are applied to different classes of image sequences. As a result, outliers can be identified and removed from the dominant motion estimate to solve the problem of inaccurate initial descending direction estimates associated with classical global motion estimation methods. The pre-analysis of scene content is based on the Spatial Temporal Gradient Scale (STGS) images derived from the original image sequences. The extra computation time for STGS-image-based pre-analysis of scene content is negligible compared to the overall speed and accuracy improvement achieved with the proposed method. Evaluations based on extensive experiments have shown that the proposed method significantly improves the speed of robust global motion estimation methods (saving about 50% of the execution time of classical methods).展开更多
基金the 2020 Guangdong International Cooperation Project(No.2019A050510007).
文摘The advent of convolutional neural networks has led to remarkable progress in dense stereo labeling problem,achieving superior performance over the traditional methods.However,the ill-posed nature of stereo matching makes noise(outliers)in winner-takes-all(WTA)disparity maps inevitable.This paper presents a robust statistical approach to noise detection and refinement of WTA disparity maps.In the context of noise detection,the input noisy WTA disparity map is segmented into regular grid cells(regions)with the aim of leveraging Markov random field(MRF)to infer candidate disparity labels.However,there are two key problems:there can be large severe outliers in the regions;second,the regular partition process may produce regions with mixed disparity distributions.To overcome these problems,we optimize a robust objective function over the segmented disparity map.By obtaining the optimal solution of the objective function through a maximum a posteriori estimation in a probabilistic model,we are able to infer MRF candidate disparity labels.We then apply a soft-segmentation constraint on the estimated MRF candidate disparity labels to describe and detect outliers in the disparity map.Next,an edge-preserving statistical inference that leverages the joint statistics of the disparity map and its guidance reference image is used to select correct candidate disparity for each detected outlier.Finally,a weighted median filter is applied to remove small spikes and irregularities in the resulting disparity map.Rigorous and comprehensive experiments showed that the proposed method is distributionally robust and outlier resistant,and can effectively detect and correct outliers in disparity maps.Middlebury evaluation benchmark validated the competitive performance of the proposed method.
基金the State High- Tech Developments Plan of China!(No.86 3- 30 6 - 0 3- 0 7)
文摘This paper presents a new robust global motion estimation method based on pre-analysis of the video content. The novel idea in the proposed method, compared to classical robust statistics-based estimation methods, is to classify the video sequences into 3 classes based on the analysis of scene content before motion estimation. Different motion models and estimation methods are applied to different classes of image sequences. As a result, outliers can be identified and removed from the dominant motion estimate to solve the problem of inaccurate initial descending direction estimates associated with classical global motion estimation methods. The pre-analysis of scene content is based on the Spatial Temporal Gradient Scale (STGS) images derived from the original image sequences. The extra computation time for STGS-image-based pre-analysis of scene content is negligible compared to the overall speed and accuracy improvement achieved with the proposed method. Evaluations based on extensive experiments have shown that the proposed method significantly improves the speed of robust global motion estimation methods (saving about 50% of the execution time of classical methods).