Upsampling

About: Upsampling is a research topic. Over the lifetime, 2426 publications have been published within this topic receiving 57613 citations.

Learning Efficient Stereo Matching Network With Depth Discontinuity Aware Super-Resolution

Abstract: Deep convolutional neural networks (CNNs) have shown great potential to provide accurate depth estimation based on stereo images. Previous work has focused on developing robust stereo matching architectures, while little attention has been paid on improving the network efficiency. In this paper, we propose an efficient Siamese CNN architecture that combines the low resolution disparity estimation and the depth discontinuity aware super-resolution. Specifically, we propose to construct, filter and perform regression on a low resolution cost volume through the designed stereo matching backbone network. A fast depth discontinuity aware super-resolution subnetwork is proposed for upsampling the low resolution disparity map to the desired resolution. Under the guidance of the intensity edge features extracted from the left color image, depth edge residuals are hierarchically learned to refine the upsampled depth map. A delayed upsampling structure is designed to ensure that the computational complexity is proportional to the spatial size of the input disparity map. We also propose to supervise the first derivative loss of the predicted disparity map that makes the network adaptively aware of the depth discontinuity edges. Experiments show that the proposed stereo matching network achieves a comparable prediction accuracy and much faster running speed compared with state-of-the-art methods.

Vehicle Detection Based on Semantic-Context Enhancement for High-Resolution SAR Images in Complex Background

Abstract: Small-scale target detection (such as vehicles) in complex synthetic aperture radar (SAR) image scenes has always been a pain point for the advanced convolutional neural network (CNN)-based target detectors because of the downsampling operations and the local receptive field characteristics of CNNs. To tackle these limitations, a vehicle detector named SCEDet for the small-scale vehicles in SAR images is proposed to improve the detection performance in this letter. SCEDet mainly consists of two parts: subaperture semantic feature extraction and subaperture semantic-context enhancement (SCE) with SCE module. First, ResNet34 with subaperture decomposition is used to efficiently exploit the latent subaperture semantic features. Then, the SCE module is proposed to balance the multiscale semantic information as well as aggregate the global context information for vehicle detection with a small number of parameters and computation costs. The experimental results on the FARAD dataset (0.1 m $\times0.1$ m, Ka-band) demonstrate that both the detection performance and the speed are much better than other detection methods under the same hardware conditions.

Field-based upsampling method for performing zoom in a digital video system

Abstract: A video processing system that processes vertical column of pixels from individual fields is disclosed. The video processing system processes pixels from an even field independent of the pixels in the odd field, and vice versa. The video processing system preferably includes a system memory for storing fields of input video images and a vertical filter coupled to the system memory via a data bus. The field data is retrieved from the system memory by the vertical filter and processed as individual fields. The vertical filter preferably calculates a 2× enlargement of the input image, although the filter can be adapted to enlarge by different factors if desired. The enlargement process generally involves representing an input image with twice as many lines of pixels values as the initial image. The values that are used to represent the enlarged pixels are preferably weighted averages of the pixels from an input pixel field. The vertical filter calculates the weighted averages using coefficients that are based on the proximity of the resulting enlarged pixel values to the corresponding pixels from the input field.