This paper proposes a novel algorithm for lane detection based on inverse perspective transformation and Kalman filter. A simple method is presented to remove effects generate top-view image. does not need obtain the internal external parameters of camera. The Gaussian kernel function used convolute image highlight lines, then an iterative threshold segment searching applied in obtained from determine points their positions. Combining with feature voting mechanism, detected are fitted as...

Semantic segmentation of panoramic images plays a key role in many applications, such as security monitoring and autonomous driving. With the rapid development deep learning, some networks are developed to segment semantically. However, these don't have special modules correct image distortion according principle, which makes feature extraction unreasonable during convolution because distortion. This article proposes novel semantic network for outdoor scenes based on convolution. The...

The paper proposes a novel insulator-contour-detection-algorithm on power transmission lines images. will have very important academic and practical significance in helicopter or robot patrol system lines. physical characteristics of insulator are analyzed detail to obtain image features insulator, the Hough transform based gradient information is used detect ellipse image. After double thresholds processing size ellipse, linear fitting about center template matching hausdorff distance,...

Unmanned aerial vehicles are now playing an important role in water conservancy industry. This paper presents a visual navigation system that is performed with rotor-craft UAV. The main sensor stereo camera, which can output two frames of images at the same time from left and right sensors. sum absolute difference algorithm adopted for fast image matching, then disparity depth information computed. A simple efficient ground extraction method proposed locating obstacles. Finally, path...

According to color characteristics of insulator ,The paper is based on HSI model and Mean Shift algorithm in order segmentation images. It firstly introduces theory mean shift algorithm, then explains morphological processing with edge detection extract the images contour. Last using Hough transform obtain results. Experiment indicates that VC6.0 combined opencv simulation proposed could effectively provides basis for follow-up determination faults.

Comprehensive analysis of the tower in image feature information with Harris corner imprecise for complex image, we propose an improved algorithm, which use detection method to calculate number gray similar target pixel 8 field, analyze level distribution pixels within local range, and then lag suppression set threshold , if is objective function greater than threshold, change point was identified as final corner. The experimental results proved be effective determine area.

This paper aims to design a self-sensing cement mortar of optimised mechanical and sensing properties considering the water-cement ratio propose freeze–thaw damage evaluation parameter. An orthogonal experimental compressive strength test piezoresistive response was utilised find optimal proportion. Freeze–thaw conducted found method for characterising damage. It is that order influential factors on stress sensitivity CF (carbon fibre) content > water-ratio SF (steel content. A parameter Dk...

<title>Abstract</title> The union combining the stitching of normal images and unsupervised semantic segmentation stitched image is an important region, which crucial for autonomous driving, intelligent robots, vehicle detection. This paper designs method image. are by using designed Ribeiro D.. uses opened in github. contains distortion. distortion will make feature extraction unreasonable. form different from panoramic combined line images. Therefore, DCM proposed Xing Hu useless to...

This paper aims to develop a visualisation detection method of moisture accumulation in asphalt materials using electrical capacitance tomography technology. Firstly, the validity data measured by circular sensor for was analysed. Then, reconstructed and optimal imaging obtained. Finally, quality under different conditions evaluated. It is found that values specimens show good 'U' curve trend. The Maxwell normalisation model Landweber algorithm have been empirically validated their superior...

The union combining the stitching of normal images and unsupervised semantic segmentation stitched image is an important region, which crucial for autonomous driving, intelligent robots, vehicle detection. This paper designs method image. are by using designed Ribeiro D.. uses opened in github. contains distortion. distortion will make feature extraction unreasonable. form different from panoramic combined line images. Therefore, DCM proposed Xing Hu useless to extract features reasonable....

Vision-and-Language Navigation (VLN) is a challenging task that requires an agent to navigate through photorealistic environments following natural-language instructions. One main obstacle existing in VLN data scarcity, leading poor generalization performance over unseen environments. Tough argumentation promising way for scaling up the dataset, how generate both diverse and world-consistent remains problematic. To cope with this issue, we propose generation (WCGEN), efficacious...

Temperature-induced cracks during the construction of large concrete structures, such as water gates and bridges, caused by hydration heat, pose a serious threat to structural safety reliability. To address this, various countries have developed temperature control standards guidelines for mass providing design direction evaluation criteria. China United States (U.S.) are leaders in field with significant influence international projects. The study differences results between two countries’...

Occlusion and lack of visibility even in sparse crowd scenes make it difficult to track individual pedestrians correctly consistently, particularly a single view. We present novel pedestrian tracking approach that connects with reidentification locate maintain the identity certain people who may be occluded for long time. First, two models are constructed. One model tracks trains classifier, while other reidentifies interest from detection results trained classifier. Secondly, we design set...

The Sparsity Adaptive Matching Pursuit (SAMP) algorithm needs not to know a priori information of sparsity, which makes it has unique advantages compared other greedy algorithms. In the reconstruction process, step size can be changed with stage, increases its accuracy. Aiming at SAMP, is unreasonable, will only increase and decrease properly. If result one iteration poor, continue on larger step. This reduces quality algorithm. paper designs appropriately reduced, proposes SAMP Optimized...

Summary Block Matching Local SVD Operator Based Sparsity and TV Regularization (BMLSVDTV) method is a novel effective denoising scheme in the field of image processing, which integrates group sparsity global regularization unified framework works well attenuating random noise image. The able to recover structures with more details clearly by exploiting repetitiveness valid globally images, can suppress pseudo-Gibbs artifacts restoration. Because reflection seismic data generally has degree...

Abstract Semantic segmentation of panoramic images plays a crucial role in many applications, such as scene understanding, autonomous navigation, and community security. However, the traditional deep learning algorithms achieve lower accuracy for due to serious image distortion. This paper proposes new convolutional neural network segment outdoor scenes semantically. The proposed includes two branches: semantic sub-branch (SS-branch) feature enhancement (FE-branch). SS-branch contains parts:...

In the process of Installation and quality testing slewing ring , connection withinjuries for workers installing ball inconvenience we establish a bearing assembly inspection system based on virtual instrument .with labview interface goes to control stepper motor forward, reverse indexing stepping stopping in order stimulate rotate so that install ; Meanwhile, wo can go collect, display, store information generate reports, query historical data .

Stable Diffusion has established itself as a foundation model in generative AI artistic applications, receiving widespread research and application. Some recent fine-tuning methods have made it feasible for individuals to implant personalized concepts onto the basic with minimal computational costs on small datasets. However, these innovations also given rise issues like facial privacy forgery copyright infringement. In studies, researchers explored addition of imperceptible adversarial...