EPIC: Context Adaptive Lossless Light Field Compression using Epipolar Plane Images

Abstract

This paper proposes extensions of CALIC for lossless compression of light field (LF) images. The overall prediction process is improved by exploiting the linear structure of Epipolar Plane Images (EPI) in a slope based prediction scheme. The prediction is improved further by averaging predictions made using horizontal and verticals EPIs. Besides this, the difference in these predictions is included in the error energy function, and the texture context is redefined to improve the overall compression ratio. The results using the proposed method shows significant bitrate-savings in comparison to standard lossless coding schemes and offers significant reduction in computational complexity in comparison to the state-of-the-art compression schemes.

Epipolar Plane Image-Based Lossless and Near-Lossless Light Field Compression

“…In the literature, it can be noticed that there is a strong trade-off between coding efficiency and computational complexity when it comes to lossless LF compression. In our previous work for lossless LF compression [8], we have addressed this problem by modifying CALIC to efficiently exploit correlation present in the LF structure by introducing effective prediction and context formation techniques. We have introduced an EPI Slope-based predictor, which estimates the slope of the EPI line and subsequently predicts the intensity using 1-D quadratic interpolation.…”

“…4, due to the raster scan order of CALIC, only the pixels in the causal neighborhood are available, therefore, the gradients can not be estimated directly. Instead, by assuming that disparity maps are locally smooth, we estimate the gradient vector at X by blending the gradient vectors of the four neighboring regions (contrary to the set of 3 neighboring regions in [8]), as shown in Fig. 4.…”

“…In contrast to [8], we remove any outlying vector, out of the four estimates, by applying a constraint on the angle θ r of the gradient vector ∇ r . The gradient vector with the largest magnitude is selected as the reference vector ∇ ref .…”

“…Previously in [8], we averaged the estimated angles for the final estimate. Here, a robust estimate for the gradient vector ∇, perpendicular to the EPI line, is achieved through the weighted sum of the inlying gradient vectors,…”

“…The lossless LF compression schemes in the literature offer decent compression ratios at the cost of hours of encoding time [6] and/or complex computation [7]. For this reason, we have recently proposed an efficient lossless LF compression scheme, EPIC [8], which extends CALIC to exploit the LF structure for enhanced compression performance. In this paper, we extend this approach to gain further efficiency in compression ratio (5.3% reduction), processing time (71.3% reduction 1 ) and FIGURE 1: Raw lenslet image captured by the Lytro Illum camera.…”