To achieve robust and accurate unstructured road detection, a model‐based feature‐based combinatory method is proposed. This obtains the saliency map via graph‐based manifold ranking approach accurately segments region by fitting model. First, graph constructed with superpixel nodes distribution Then, seed query are robustly selected. Second, rankings foreground background queries conducted successively. improve robustness to false queries, authors propose using two different approaches...

To solve the homography estimation problem containing outliers and noise, a fast, robust, accurate method is proposed. In this method, are rejected based on differing characteristics of algebraic errors between inliers, estimated by minimising residual vector. The advantage in integrating outlier rejection into pipeline. computational complexity proposed not increased, random sample consensus algorithm needed to extract as was previously necessary. Since process an criterion without...

To achieve large-scale outdoor real-time localization, a semidirect visual odometry method for monocular camera is proposed. In this method, the performance of features from accelerated segment test (FAST) algorithm improved to detect corners with good tracking and distribution properties. A multiscale Lucas-Kanade (LK) approach presented build correspondence between map points robustly efficiently. Thereafter, system put forward long-distance localization ground vehicle. The main...

Road markings are important information of transport systems for drivers or intelligent vehicle. Efficient road feature extraction is pre-requisite to detection, recognition and visual localization. However, most previous lane extractors operating on conventional images, the methods omnidirectional images rarely considered in detection research. In this paper, three omni-directional presented based different characteristics markings. Those evaluated real world panoramic image dataset....

Personal thermal management clothing has garnered significant attention in recent years for offering innovative solutions to address extreme environments. By examining the “human-clothing-environment” system, we have analyzed contribution and developmental trajectory of personal clothing, considering diverse human needs application Local variations physiology necessitate localized comfort provided by clothing. Despite advancements, many garments still encounter challenges such as bulky size,...

Laser active imaging is widely used in many fields. The intensity image quality of laser affected by various degradations, such as speckle effect and noise. Most existing objective assessment (IQA) methods that consider only a single distorted are not suitable for an image. A multiscale full-reference IQA method presented. proposed based on improved gradient magnitude similarity deviation (GMSD) the nonsubsampled contourlet transform (NSCT) domain. reference images decomposed NSCT to emulate...

A novel method robust to outliers is proposed which integrates rejection within the homography estimation pipeline with a negligible computational overhead. With basis of algebraic criterion, has low level time an arbitrary large number key-point correspondences. Given set 2D-to-2D matches, we formulate pose problem as low-rank homogeneous system where solution lies on its 1D null space. Outlier correspondences are those rows linear perturb space and progressively detected by projecting them...

In view that the noise of image in monocular visual navigation, a Neigh-Shrink de-noising method based on Bayes-Shrink is proposed. First, uses adaptive threshold replace common as contraction Neigh-Shrink, and then eliminate high frequency sub-band obtained by system after non-subsample Contourlet transform (NSCT) decomposition. Experiment results show improved performance effects have been improved, with better anti-noise excellent edge protection, can effectively be used navigation.

In this paper, a fast algorithm is proposed to tackle the constrained total generalized variation (TGV) based image restoration problem. The proceeds by splitting: nonsmooth TGV model first decomposed into several subproblems easier solve; then linear gradient or proximity operators, including projections and shrinkages, of are individually called without inner iteration. highly parallel since most its steps can be executed simultaneously. Image experiments demonstrate that outperforms...

Purpose The purpose of this paper is to improve the robustness traditional Bhattacharyya metric for effect histogram quantization in histogram-based visual tracking. However, neglects correlation crossing-bin and not robust quantization. Design/methodology/approach In paper, authors propose a tracking method via similarity particle filter. Findings A matrix introduced into measuring reference candidate histogram, basic tasks measure such as maximum self triangle inequality are proven. use...

A fast algorithm is proposed to tackle the constrained total generalised variation (TGV)‐based image‐restoration and reconstruction problems. The proceeds by splitting: non‐smooth TGV model first decomposed into several sub‐problems easier solve, then linear gradient or proximity operators, including projections shrinkages, of are individually called without inner iteration. highly parallel since most its steps can be executed simultaneously. Image‐restoration experiments demonstrate that...

To Improve the robust performance of visual tracking in various kind scene, a novel method by harmony search and co-inference learning based on multi-cues was proposed. The candidate state achieved learning. Then memory vector corresponded biggest fitness function chosen as vector. Compared with color cue multi-cue integrating multiply fusion strategy, our algorithm has better robustness complex environments such illumination change seriously motion blur.

Time-sensitive targets strike is a hot and difficult topic in the research of attack system, however, acquisition reference images time-sensitive are high-cost data scarce. Aiming at problem, simulation method generating multi-view proposed. Firstly, imaging model aircraft viewpoints transformation established, coordinate deduced detail; Secondly, generation algorithm designed, implementation introduced detail. Simulation result shows that effective for with high value targets.