BE-CALF: Bit-Depth Enhancement by Concatenating All Level Features of DNN

A Little Bit More: Bitplane-Wise Bit-Depth Recovery

“…Recently, deep learning-based approaches have gained in popularity [10], [11], [12], [13], [14], [15], [16]. Works by Hou and Qiu [14] and GG-DCNN [15] employ a U-Net style architecture to predict the HBD image given the LBD image.…”

“…BitNet [12] uses an encoder-decoder architecture with dilated convolutions and multi-scale feature integration. BE-CALF [11] employs a chain of convolutionaldeconvolutional layers with dense concatetations of all level features. BDEN [13] uses a two-stream architecture, one for flat and another for non-flat regions, with a local adaptive adjustment preprocessing step for the flat areas.…”

“…Several classical methods (e.g., [4], [5], [6], [7], [8], [9]) for bit-depth recovery exist in the literature. More recently, deep learning algorithms [10], [11], [12], [13], [14], [15], [16] have been proposed. The strategy employed by most deep learning methods (e.g., [10], [12], [13], [14], [15], [16]) is to train the network using the quantized low-bit-depth (LBD) image as input and have the network predict the high-bit-depth image as output.…”

“…The strategy employed by most deep learning methods (e.g., [10], [12], [13], [14], [15], [16]) is to train the network using the quantized low-bit-depth (LBD) image as input and have the network predict the high-bit-depth image as output. An alternative strategy recently proposed by Liu et al [11] is to train the network to predict the residual between the ground truth HBD image and the quantized LBD image. At test time, this output residual image is added to the quantized LBD image to produce the desired HBD image.…”

“…We adopt this latter formulation in our work. However, as opposed to [11], who estimate the residual in a single shot, we propose to recover the residual image corresponding to the lost bits one bitplane at a time.…”