Human Visual Attention Mechanism-Inspired Point-and-Line Stereo Visual Odometry for Environments with Uneven Distributed Features

Visual odometry is critical in visual simultaneous localization and mapping for robot navigation.However,the pose estimation performance of most current visual odometry algorithms degrades in scenes with unevenly distributed features because dense features occupy excessive weight.Herein,a new human visual attention mechanism for point-and-line stereo visual odometry,which is called point-line-weight-mechanism visual odometry(PLWM-VO),is proposed to describe scene features in a global and balanced manner.A weight-adaptive model based on region partition and region growth is generated for the human visual attention mechanism,where sufficient attention is assigned to position-distinctive objects(sparse features in the environment).Furthermore,the sum of absolute differences algorithm is used to improve the accuracy of initialization for line features.Compared with the state-of-the-art method(ORB-VO),PLWM-VO show a 36.79%reduction in the absolute trajectory error on the Kitti and Euroc datasets.Although the time consumption of PLWM-VO is higher than that of ORB-VO,online test results indicate that PLWM-VO satisfies the real-time demand.The proposed algorithm not only significantly promotes the environmental adaptability of visual odometry,but also quantitatively demonstrates the superiority of the human visual attention mechanism.

Microstructure recognition of steels by machine learning based on visual attention mechanism

U-Net has achieved good performance with the small-scale datasets through skip connections to merge the features of the low-level layers and high-level layers and has been widely utilized in biomedical image segmentation as well as recent microstructure image segregation of the materials.Three representative visual attention mechanism modules,named as squeeze-and-excitation networks,convolutional block attention module,and extended calibration algorithm,were intro-duced into the traditional U-Net architecture to further improve the prediction accuracy.It is found that compared with the original U-Net architecture,the evaluation index of the improved U-Net architecture has been significantly improved for the microstructure segmentation of the steels with the ferrite/martensite composite microstructure and pearlite/ferrite composite microstructure and the complex martensite/austenite island/bainite microstructure,which demonstrates the advantages of the utilization of the visual attention mechanism in the microstructure segregation.The reasons for the accuracy improvement were discussed based on the feature maps analysis.

Snake-Hot-Eye-Assisted Multi-Process-Fusion Target Tracking Based on a Roll-Pitch Semi-strapdown Infrared Imaging Seeker

Aiming at intercepting large maneuvering targets precisely,the guidance law of advanced self-seeking missiles requires not only inertial line-of-sight(LOS)angular rate but also target maneuvering acceleration.Moreover,the semi-strapdown stabilization platform has lost the ability to measure the inertial LOS angular rate directly,which needs to be extracted by numerical calculation.The differential operation commonly used in traditional methods can magnify the measurement error of the sensor,resulting in insufficient calculation accuracy of the line-of-sight angular rate.By analyzing the mathematical relationship between the missile-target relative motion and the angle tracking system,a multi-process-fusion integrated filter model of relative motion and angle tracking is presented.To overcome the defect that the infrared seeker cannot directly measure the missile-target distance,following the snake-hot-eye visual mechanism,a visual bionic imaging guidance method of estimating the missile-target relative distance from the infrared images is proposed to improve the observability of the filter model.Finally,target-tracking simulations verify that the estimation accuracy of target acceleration is improved by four times.

A Parzen Window-based Approach to Detection of Infrared Dim Target Under Sea-sky Background

People attach great importance to high detection probability and low false alarm probability for infrared dim target detection. Consequently, a novel approach is proposed based on inverted local information entropy map and the improved region growing technique. The idea originates from the intrinsic property of natural image, the visual mechanism of flying insects and the information entropy theory. Besides qualitative analyses, other methods including the norms of local signal-to-background ratio, local signal-to-noise ratio, region non-uniformity, single-frame detection probability and single-frame false alarm probability are adopted to quantitatively evaluate the proposed approach. Both qualitative and quantitative comparisons confirm the validity and efficiency of the proposed approach.