Connected Component Analysis Method Research Articles

Archaeological site segmentation of ancient city walls based on deep learning and LiDAR remote sensing

Ancient city walls, one of the most notable distinguishing features of Chinese ancient cities, are military defenses constructed of rammed earth. The ancient city walls have considerable study value because they served as the city's boundary and a symbol of power. However, as a result of natural erosion and human activities, many sites have been ruined. Existing optical remote sensing technologies, LiDAR point cloud processing algorithms, and deep learning methods are inadequate for the extraction and segmentation of ancient city wall sites. The novel semantic segmentation method for ancient city wall sites is described in this paper that extracts sites at the pixel level from LiDAR remote sensing data based on deep learning. To begin, the point cloud data collected by airborne laser scanning is processed into DEM data, and the distribution of ancient city walls in the study area is obtained through archaeological survey and expert interpretation. The dataset for deep learning semantic segmentation is then generated using image cropping and data augmentation techniques. Third, implement a U-Net semantic segmentation framework for microtopographic archaeological sites, and predict ancient city wall sites in the testing region after model training. Finally, the deep learning results are optimized using the connected component analysis method, and prediction mistakes such as holes and noise are removed. Taking Jinancheng, the capital city of the Chu kingdom, as an example, the proposed method process can identify and extract the ancient city wall sites at the pixel level, where the evaluation metrics reach 94.12% (Precision) and 81.38% (IoU). The experiment results are excellent due to improvement strategies in the dataset generation, model training, and post-processing steps. Thus, this study is significant for the current survey and protection of ancient city wall sites. The source code will be freely available at https://github.com/wshunli/Open-CHAI-CityWalls.

Connected Component Analysis of Dynamical Perturbation Contact Networks.

Describing protein dynamical networks through amino acid contacts is a powerful way to analyze complex biomolecular systems. However, due to the size of the systems, identifying the relevant features of protein-weighted graphs can be a difficult task. To address this issue, we present the connected component analysis (CCA) approach that allows for fast, robust, and unbiased analysis of dynamical perturbation contact networks (DPCNs). We first illustrate the CCA method as applied to a prototypical allosteric enzyme, the imidazoleglycerol phosphate synthase (IGPS) enzyme from Thermotoga maritima bacteria. This approach was shown to outperform the clustering methods applied to DPCNs, which could not capture the propagation of the allosteric signal within the protein graph. On the other hand, CCA reduced the DPCN size, providing connected components that nicely describe the allosteric propagation of the signal from the effector to the active sites of the protein. By applying the CCA to the IGPS enzyme in different conditions, i.e., at high temperature and from another organism (yeast IGPS), and to a different enzyme, i.e., a protein kinase, we demonstrated how CCA of DPCNs is an effective and transferable tool that facilitates the analysis of protein-weighted networks.

Robust Tracking and Clean Background Dense Reconstruction for RGB-D SLAM in a Dynamic Indoor Environment

This article proposes a two-stage simultaneous localization and mapping (SLAM) method based on using the red green blue-depth (RGB-D) camera in dynamic environments, which can not only improve tracking robustness and trajectory accuracy but also reconstruct a clean and dense static background model in dynamic environments. In the first stage, to accurately exclude the interference of features in the dynamic region from the tracking, the dynamic object mask is extracted by Mask-RCNN and optimized by using the connected component analysis method and a reference frame-based method. Then, the feature points, lines, and planes in the nondynamic object area are used to construct an optimization model to improve the tracking accuracy and robustness. After the tracking is completed, the mask is further optimized by the multiview projection method. In the second stage, to accurately obtain the pending area, which contains the dynamic object area and the newly added area in each frame, a method is proposed, which is based on a ray-casting algorithm and fully uses the result of the first stage. To extract the static region from the pending region, this paper designs divisible and indivisible regions process methods and the bounding box tracking method. Then, the extracted static regions are merged into the map using the truncated signed distance function method. Finally, the clean static background model is obtained. Our methods have been verified on public datasets and real scenes. The results show that the presented methods achieve comparable or better trajectory accuracy and the best robustness, and can construct a clean static background model in a dynamic scene.

Segmentation algorithm for overlap recognition of seedling lettuce and weeds based on SVM and image blocking

For the problem of a low recognition rate and shape feature failure caused by overlapping seedlings and weeds during the development of an intelligent lettuce weeding robot, a method to identify seedling lettuce and weeds based on an image block and support vector machine (SVM) is proposed, which realizes their precise identification and boundary segmentation. The a* channel is used to grayscale the collected image. The Otsu and morphological methods are selected to extract all the green targets in the image. The connected component analysis method is applied to label the green targets with regions of interest (ROIs), and those with pixel areas larger than the area threshold are normalized to 256 × 256 pixels. The image blocking technique is introduced to separately aliquot the normalized ROI, with block sizes of 16 × 16, 32 × 32, and 64 × 64 pixels. On this basis, the image sub-blocks are manually labeled, block by block, to extract three texture features: histogram of oriented gradient (HOG), local binary pattern (LBP), and gray-level co-occurrence matrix (GLCM). With the accuracy of fivefold cross-validation as the optimization objective, a genetic algorithm (GA) is used to optimize the SVM penalty and kernel parameters of 21 groups of research objects (one block size has three texture features, which are arbitrarily combined to form seven research objects, with a total of three block sizes). We compare the recognition performance of the SVM, RF, KNN, and GA-SVM classifiers in a single feature and a combination of fusion strategies through comparative analysis. When the block size is 32 × 32 pixels, the fusion of LBP and GLCM features under the GA-SVM classifier has the highest accuracy, and the optimal SVM model for the identification of lettuce and weeds in the seedling stage is obtained. For the misidentified image sub-blocks in optimization model recognition, an image block reconstruction method based on the comparison of the center point and eight-neighbor label value is proposed, and this is combined with the proportion of image blocks of two labels for comprehensive judgment. The center point label value is reconstructed to the improve recognition accuracy. Experimental results show that the average precision, recall, and F1 score of the proposed method are 0.9473, 0.9529, and 0.9498, respectively, and those of images without overlapping leaves can all reach 1, thus providing a theoretical basis for crop recognition and segmentation.

Motion feature based melt pool monitoring for selective laser melting process

Since various build defects in the Selective laser melting (SLM) process are found to be associated with the instability of melt pool, the melt pool monitoring is particularly important for the final product quality control. Previous studies focus on the geometric features to describe the changes of the size and shape of melt pool, which are not sufficient to characterize the dynamic variations of the melt pool during the build process monitoring. To solve this problem, a new motion feature is introduced to describe the moving melt pool. The melt pool and spatters are extracted by thresholds combined with the connected component analysis method. The distance between the centroid and the boundary of melt pool is calculated from the unfolded clockwise at a step angle, which constructs a high dimensional feature vector as the motion features. The k-means clustering algorithm is applied to cluster the motion features under varied process parameters, aiming at construct the link between the melt pool states and processing parameter for the quality control. The research results have shown that the extracted motion features can describe the variation of melt pool more accurately than the traditional geometric features, and they can distinguish the moving direction and melted states of over melting, partial melting and defects simultaneously. This research provides a new approach for intelligent online monitoring of the SLM process.

Chinese Traffic Panels Detection and Recognition From Street-Level Images

Traffic sign detection and recognition has been the active research topic due to its potential applications in intelligent transportation. However, detection and recognition of traffic panels containing much information, still remains to be a challenging problem. This paper proposes a method to detect and recognize traffic panels from street-level images in the urban scenes and to analyze the information on them. The traffic panels are detected based on histogram of oriented gradient and linear support vector machines. The text strings and symbols on traffic panels are segmented using connected component analysis method. Finally, the symbols on traffic panels are recognized by means of a model named bag of spatial visual words. Experimental results on images from Baidu Panorama Map prove the effectiveness of the proposed method.

Multilingual corpus construction based on printed and handwritten character separation

This paper proposes an effective method to extract printed and handwritten characters from multilingual document images to build corpus. To extract the characters from the document images, a connected component analysis method is used to remove the graphics. After that, multiple types of features and AdaBoost algorithm are introduced to classify printed and handwritten characters in a more versatile and robust way. Firstly, the content of the image is divided into several text patches which are then used to distinguish different languages. Secondly, we use the multiple types of features and AdaBoost algorithm to train the classifiers based on the segmented patches. Finally, we can separate printed and handwritten parts of new image set by the trained classifiers. The proposed method improves the precision of the extraction of written materials in text images of different languages. Experimental results demonstrate that the proposed method is more accurate in terms of precision and recall rate compared with the state-of the-art methods.

Obstacle Detection for Patrol Robot Using Bumblebee2 Stereo Vision System

This paper employs Bumblebee2 stereo vision system to detect the obstacles for patrol robot in substation environment. Firstly, with the selected points in the disparity image, the ground plane is calculated by the RANSAC (Random Sample Consensus) algorithm. And then, the local occupancy grid map is built for patrol robot, and the obstacles are detected through connected component analysis method. The actual test in substation environment verified the reliability of the system.