Local Adaptive Image Filtering Based on Recursive Dilation Segmentation.

This paper introduces a simple but effective image filtering method, namely, local adaptive image filtering (LAIF), based on an image segmentation method, i.e., recursive dilation segmentation (RDS). The algorithm is motivated by the observation that for the pixel to be smoothed, only the similar pixels nearby are utilized to obtain the filtering result. Relying on this observation, similar pixels are partitioned by RDS before applying a locally adaptive filter to smooth the image. More specifically, by directly taking the spatial information between adjacent pixels into consideration in a recursive dilation way, RDS is firstly proposed to partition the guided image into several regions, so that the pixels belonging to the same segmentation region share a similar property. Then, guided by the iterative segmented results, the input image can be easily filtered via a local adaptive filtering technique, which smooths each pixel by selectively averaging its local similar pixels. It is worth mentioning that RDS makes full use of multiple integrated information including pixel intensity, hue information, and especially spatial adjacent information, leading to more robust filtering results. In addition, the application of LAIF in the remote sensing field has achieved outstanding results, specifically in areas such as image dehazing, denoising, enhancement, and edge preservation, among others. Experimental results show that the proposed LAIF can be successfully applied to various filtering-based tasks with favorable performance against state-of-the-art methods.

Robust RGB-T Tracking via Graph Attention-Based Bilinear Pooling.

RGB-T tracker possesses strong capability of fusing two different yet complementary target observations, thus providing a promising solution to fulfill all-weather tracking in intelligent transportation systems. Existing convolutional neural network (CNN)-based RGB-T tracking methods often consider the multisource-oriented deep feature fusion from global viewpoint, but fail to yield satisfactory performance when the target pair only contains partially useful information. To solve this problem, we propose a four-stream oriented Siamese network (FS-Siamese) for RGB-T tracking. The key innovation of our network structure lies in that we formulate multidomain multilayer feature map fusion as a multiple graph learning problem, based on which we develop a graph attention-based bilinear pooling module to explore the partial feature interaction between the RGB and the thermal targets. This can effectively avoid uninformed image blocks disturbing feature embedding fusion. To enhance the efficiency of the proposed Siamese network structure, we propose to adopt meta-learning to incorporate category information in the updating of bilinear pooling results, which can online enforce the exemplar and current target appearance obtaining similar sematic representation. Extensive experiments on grayscale-thermal object tracking (GTOT) and RGBT234 datasets demonstrate that the proposed method outperforms the state-of-the-art methods for the task of RGB-T tracking.

High-Resolution Representations Network for Single Image Dehazing.

Deep learning-based image dehazing methods have made great progress, but there are still many problems such as inaccurate model parameter estimation and preserving spatial information in the U-Net-based architecture. To address these problems, we propose an image dehazing network based on the high-resolution network, called DeHRNet. The high-resolution network originally used for human pose estimation. In this paper, we make a simple yet effective modification to the network and apply it to image dehazing. We add a new stage to the original network to make it better for image dehazing. The newly added stage collects the feature map representations of all branches of the network by up-sampling to enhance the high-resolution representations instead of only taking the feature maps of the high-resolution branches, which makes the restored clean images more natural. The final experimental results show that DeHRNet achieves superior performance over existing dehazing methods in synthesized and natural hazy images.

A Review of Remote Sensing Image Dehazing.

Remote sensing (RS) is one of the data collection technologies that help explore more earth surface information. However, RS data captured by satellite are susceptible to particles suspended during the imaging process, especially for data with visible light band. To make up for such deficiency, numerous dehazing work and efforts have been made recently, whose strategy is to directly restore single hazy data without the need for using any extra information. In this paper, we first classify the current available algorithm into three categories, i.e., image enhancement, physical dehazing, and data-driven. The advantages and disadvantages of each type of algorithm are then summarized in detail. Finally, the evaluation indicators used to rank the recovery performance and the application scenario of the RS data haze removal technique are discussed, respectively. In addition, some common deficiencies of current available methods and future research focus are elaborated.

IDE: Image Dehazing and Exposure Using an Enhanced Atmospheric Scattering Model.

Atmospheric scattering model (ASM) is one of the most widely used model to describe the imaging processing of hazy images. However, we found that ASM has an intrinsic limitation which leads to a dim effect in the recovered results. In this paper, by introducing a new parameter, i.e., light absorption coefficient, into ASM, an enhanced ASM (EASM) is attained, which can address the dim effect and better model outdoor hazy scenes. Relying on this EASM, a simple yet effective gray-world-assumption-based technique called IDE is then developed to enhance the visibility of hazy images. Experimental results show that IDE eliminates the dim effect and exhibits excellent dehazing performance. It is worth mentioning that IDE does not require any training process or extra information related to scene depth, which makes it very fast and robust. Moreover, the global stretch strategy used in IDE can effectively avoid some undesirable effects in recovery results, e.g., over-enhancement, over-saturation, and mist residue, etc. Comparison between the proposed IDE and other state-of-the-art techniques reveals the superiority of IDE in terms of both dehazing quality and efficiency over all the comparable techniques.

IDGCP: Image Dehazing Based on Gamma Correction Prior.

This paper introduces a novel and effective image prior, i.e., gamma correction prior (GCP), which leads to an efficient image dehazing method, i.e., IDGCP. A step-by-step procedure of the proposed IDGCP is as follows. First, an input hazy image is preprocessed by the proposed GCP, resulting in a homogeneous virtual transformation of the hazy image. Then, from the original input hazy image and its virtual transformation, the depth ratio is extracted based on atmospheric scattering theory. Finally, a "global-wise" strategy and a vision indicator are employed to recover the scene albedo, thus restoring the hazy image. Unlike other image dehazing methods, IDGCP is based on the "global-wise" strategy, and it only needs to determine one unknown constant without any refining process to attain a high-quality restoration, thereby leading to significantly reduced processing time and computation cost. Each step of IDGCP is tested experimentally to validate its robustness. Moreover, a series of experiments are conducted on a number of challenging images with IDGCP and other state-of-the-art technologies, demonstrating the superiority of IDGCP over the others in terms of restoration quality and implementation efficiency.

JsrNet: A Joint Sampling-Reconstruction Framework for Distributed Compressive Video Sensing.

Huge video data has posed great challenges on computing power and storage space, triggering the emergence of distributed compressive video sensing (DCVS). Hardware-friendly characteristics of this technique have consolidated its position as one of the most powerful architectures in source-limited scenarios, namely, wireless video sensor networks (WVSNs). Recently, deep convolutional neural networks (DCNNs) are successfully applied in DCVS because traditional optimization-based methods are computationally elaborate and hard to meet the requirements of real-time applications. In this paper, we propose a joint sampling-reconstruction framework for DCVS, named "JsrNet". JsrNet utilizes the whole group of frames as the reference to reconstruct each frame, regardless of key frames and non-key frames, while the existing frameworks only utilize key frames as the reference to reconstruct non-key frames. Moreover, different from the existing frameworks which only focus on exploiting complementary information between frames in joint reconstruction, JsrNet also applies this conception in joint sampling by adopting learnable convolutions to sample multiple frames jointly and simultaneously in an encoder. JsrNet fully exploits spatial-temporal correlation in both sampling and reconstruction, and achieves a competitive performance in both the quality of reconstruction and computational complexity, making it a promising candidate in source-limited, real-time scenarios.