研究目的
To refine the depth map generated by the Adaptive Random Walk with Restart (ARWR) algorithm in order to obtain significant improvements in accuracy.
研究成果
The proposed post-processing technique increases the accuracy of the depth map computed by Adaptive Random Walk with Restart method by keeping the sharp edges and corners along with main structure of the reference image in the depth map. The comparison with the top 8 methods of the Middlebury benchmark and the ARWR without post-processing proved the performance quality of the proposed method.
研究不足
The processing time of the studied method is poor, but can be readily improved as much of this work was not optimized for fast computation.
1：Experimental Design and Method Selection:
The framework starts with the Adaptive Random Walk with Restart (ARWR) algorithm. To refine the depth map, a form of median solver/filter based on the concept of the mutual structure is introduced. This filter is further enhanced by a joint filter. A transformation in image domain is introduced to remove the artifacts that cause distortion in the image.
2：Sample Selection and Data Sources:
The proposed post-processing method is compared with the top eight algorithms in the Middlebury benchmark and with Google’s new depth map estimation method.
3：List of Experimental Equipment and Materials:
Not explicitly mentioned.
4：Experimental Procedures and Operational Workflow:
The process involves extracting the initial depth using the ARWR algorithm, applying mutual joint weighted median filter to fill the regions of occlusion or depth discontinuity, overwriting the structure of the RGB image on the depth map, and transferring the depth map to a signal to perform normalized interpolated convolution.
5：Data Analysis Methods:
The performance is evaluated based on metrics such as MSE, PSNR, SNR, SSIM, and DSSIM.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
