motion_1.0
所属分类：matlab编程
开发工具：matlab
文件大小：1024KB
下载次数：69
上传日期：2011-03-02 18:33:55
上 传 者：
维也纳
说明： Robust Motion Segmentation via Lossy Compression 运动目标分割
(Robust Motion Segmentation via Lossy Compression moving object segmentation)
文件列表:
motion_1.0\motion_1.0\datasets\1R2RC.mat (269317, 2006-06-09)
motion_1.0\motion_1.0\datasets\arm.mat (46950, 2006-06-23)
motion_1.0\motion_1.0\datasets\cars10.mat (184475, 2006-09-29)
motion_1.0\motion_1.0\datasets\oc1R2RC.mat (532060, 2007-11-14)
motion_1.0\motion_1.0\find_best_segmentation.m (4065, 2008-03-26)
motion_1.0\motion_1.0\helpers\.DS_Store (6148, 2008-03-26)
motion_1.0\motion_1.0\helpers\coding_length.m (1570, 2008-03-26)
motion_1.0\motion_1.0\helpers\coding_seg.m (5093, 2007-08-01)
motion_1.0\motion_1.0\helpers\compare_labels.m (1827, 2008-03-26)
motion_1.0\motion_1.0\helpers\distinct_labels.m (1519, 2009-11-05)
motion_1.0\motion_1.0\helpers\relabel_samples.m (2561, 2007-10-04)
motion_1.0\motion_1.0\helpers\repair_corrupted_data.m (2712, 2008-03-26)
motion_1.0\motion_1.0\helpers\repair_incomplete_data.m (3881, 2008-03-27)
motion_1.0\motion_1.0\helpers\total_coding_length.m (1085, 2008-03-26)
motion_1.0\motion_1.0\load_sequence.m (1474, 2008-03-26)
motion_1.0\motion_1.0\plot_error.m (2030, 2008-03-26)
motion_1.0\motion_1.0\process_sequence.m (5051, 2008-03-27)
motion_1.0\motion_1.0\try_sequence.m (3228, 2008-03-26)
motion_1.0\motion_1.0\datasets (0, 2011-03-02)
motion_1.0\motion_1.0\helpers (0, 2011-03-02)
motion_1.0\motion_1.0 (0, 2011-03-02)
motion_1.0 (0, 2011-03-02)
Robust Motion Segmentation via Lossy Compression v 1.0
We have included four example motion sequences to test our motion segmentation algorithm:
1R2RC: a checkerboard sequence,
arm: an articulated motion sequence,
cars10: a traffic sequence, and
oc1R2RC: a checkerboard sequence with missing entries
Before using our software, make sure that the 'helpers' directory is in the path:
> addpath 'helpers';
To use our L1-based methods for entry completion and error correction, the CVX package by Stephen Boyd (http://www.stanford.edu/~boyd/cvx/)
must be installed and in the MATLAB path.
** Trying a clean sequence (no incomplete or corrupted data) **
Here is code that will give you results for one clean sequence (The 'arm' sequence with projection down to R^5).
> epsilon = logspace(-5,3,101);
> [rawData, trueLabels] = load_sequence('arm');
> processedData = process_sequence(rawData, true);
> result = try_sequence('arm', processedData, epsilon);
> computedLabels = find_best_segmentation(result, processedData, 2, epsilon);
> err = compare_labels(trueLabels, computedLabels);
** Trying a sequence with incomplete data **
Here is code that will give you results for one incomplete sequence (The 'oc1R2RC' sequence with projection down to R^d, where d is the 'sparsity preserving' dimension).
> epsilon = logspace(-5,3,101);
> [rawData, trueLabels, mask] = load_sequence('oc1R2RC');
> processedData = process_sequence(rawData, 'sparse', 'incomplete');
> result = try_sequence('oc1R2RC', processedData, epsilon);
> computedLabels = find_best_segmentation(result, processedData, 3, epsilon);
> err = compare_labels(trueLabels, computedLabels);
** Trying a sequence with corrupted data **
Here is code that will give you results for one incomplete sequence (The 'oc1R2RC' sequence with projection down to R^d, where d is the 'sparsity preserving' dimension).
> epsilon = logspace(-5,3,101);
> [rawData, trueLabels] = load_sequence('oc1R2RC');
> processedData = process_sequence(rawData, 'sparse', 'corrupted');
> result = try_sequence('oc1R2RC', processedData, epsilon);
> computedLabels = find_best_segmentation(result, processedData, 3, epsilon);
> err = compare_labels(trueLabels, computedLabels);
近期下载者：
相关文件：
收藏者：