Mean-shift Segmentation Method Research Articles

A new strategy for weed detection in maize fields

Timely determination of weed distributions in fields is crucial for the precise spraying of herbicides. This facilitates weed control while saving costs and protecting the environment. Existing weed detection strategies often rely on the utilization of numerous weed samples to train detection models directly, which presents challenges in situations involving limited weed samples. To address this issue, a novel weed detection strategy was proposed in this study to identify weeds accurately in fields with varying coverage levels. For this purpose, red–green–blue (RGB) images of maize fields with different weed coverage levels were captured via a vertical take-off and landing fixed-wing unmanned aerial vehicle (UAV). The UAV images were first mosaicked, and a new weed detection strategy was developed and assessed. In this process, the MeanShift segmentation method, coupled with the local variance (LV) segmentation evaluation function and the Otsu automatic classification method, was initially employed to extract vegetation areas. The you-only-look-once (YOLO) v5n model was subsequently improved and used to detect maize plants. Finally, weed mapping was achieved by removing the identified maize plants from the vegetation through overlay analysis. The evaluation of the proposed method via an external dataset yielded favorable weed detection results, with an R2 value of 0.96 and a root mean square error (RMSE) value of 3.08 % under the different weed coverage levels. Specifically, in addition to adjusting the activation function and the nonmaximum suppression method, the impacts of integrating various attention modules at different positions on the performance of the YOLO v5n model for maize plant detection were analyzed. Improving the YOLO v5n model by incorporating the efficient channel attention (ECA) module into the backbone of the original model and utilizing the Hardswish activation function is recommended. Overall, this study offers support for precise weed control.

INDIVIDUAL TREE EXTRACTION FROM UAV LIDAR POINT CLOUDS BASED ON SELF-ADAPTIVE MEAN SHIFT SEGMENTATION

Abstract. Unman aerial vehicle (UAV) LiDAR has been widely used in the field of forestry. Individual tree extraction is a key step for forest inventory. Although many individual tree extraction methods have been proposed, the individual tree extraction accuracy is still low due to the complex forest environments. Moreover, many parameters in these methods generally need to be set. Thus, the degree of automation of the methods is generally low. To solve these problems, this paper proposed an automatic mean shift segmentation method, in which the kernel bandwidths can be calculated self-adaptively. Meanwhile, a hierarchy mean shift segmentation technique was proposed to extract individual tree gradually. A plot-level UAV LiDAR tree dataset was adopted for testing the performance of the proposed method. Experimental results showed that the proposed method can achieve better individual tree extraction result without any parameter setting. Compared with the traditional mean shift segmentation method, both the completeness and mean accuracy of the proposed method are higher.

A new method for 3D individual tree extraction using multispectral airborne LiDAR point clouds

Characterization of individual trees is essential for many applications in forest management and ecology. Previous studies relied on single tree detection from monochromatic wavelength airborne laser scanning (ALS) systems and they focused on the use of the geometric spatial information of the point clouds (i.e., X, Y, and Z coordinates). However, there is quite often a difficulty dealing with clumped trees when only the geometric spatial information is considered. The emergence of multispectral LiDAR sensors provides a new solution for individual tree structure acquisition. The aim of this paper is to investigate the performance of multispectral ALS data for delineating individual trees which are challenging by using the monochromatic wavelength ALS system. The proposed workflow utilizes the mean shift segmentation method on different feature spaces for crown isolation. In addition, both spatial domain and multispectral domain are used to refine the under-segmentation crown segments. Ten plots (2 sets of different structural complexity) located in the dense coniferous forest area in Tobermory, Ontario, Canada are selected as experiment data. Results show that the developed method correctly detects 88% and 82% of the dominant trees with and without multispectral information, respectively. Compared with segmentation using geometric spatial information solely, the main improvements are achieved for clumped tree segment with the distinguished multispectral features. This study demonstrates that multispectral airborne laser scanning data is more capable for individual tree delineation than monochromatic wavelength laser scanning data in dealing with forests with clumped crowns in dense forests.

Moving Object Detection With a Freely Moving Camera via Background Motion Subtraction

Detection of moving objects in a video captured by a freely moving camera is a challenging problem in computer vision. Most existing methods often assume that the background (BG) can be approximated by dominant single plane/multiple planes or impose significant geometric constraints on BG, or utilize a complex BG/foreground probabilistic model. Instead, we propose a computationally efficient algorithm that is able to detect moving objects accurately and robustly in a general 3D scene. This problem is formulated as a coarse-to-fine thresholding scheme on the particle trajectories in the video sequence. First, a coarse foreground (CFG) region is extracted by performing reduced singular value decomposition on multiple matrices that are built from bundles of particle trajectories. Next, the BG motion of pixels in the CFG region is reconstructed by a fast inpainting method. After subtracting the BG motion, the fine foreground is segmented out by an adaptive thresholding method that is capable of solving multiple-moving-objects scenarios. Finally, the detected foreground is further refined by the mean-shift segmentation method. Extensive simulations and a comparison with the state-of-the-art methods verify the effectiveness of the proposed method.

Reweighted mass center based object-oriented sparse subspace clustering for hyperspectral images

Considering the inevitable obstacles faced by the pixel-based clustering methods, such as salt-and-pepper noise, high computational complexity, and the lack of spatial information, a reweighted mass center based object-oriented sparse subspace clustering (RMC-OOSSC) algorithm for hyperspectral images (HSIs) is proposed. First, the mean-shift segmentation method is utilized to oversegment the HSI to obtain meaningful objects. Second, a distance reweighted mass center learning model is presented to extract the representative and discriminative features for each object. Third, assuming that all the objects are sampled from a union of subspaces, it is natural to apply the SSC algorithm to the HSI. Faced with the high correlation among the hyperspectral objects, a weighting scheme is adopted to ensure that the highly correlated objects are preferred in the procedure of sparse representation, to reduce the representation errors. Two widely used hyperspectral datasets were utilized to test the performance of the proposed RMC-OOSSC algorithm, obtaining high clustering accuracies (overall accuracy) of 71.98% and 89.57%, respectively. The experimental results show that the proposed method clearly improves the clustering performance with respect to the other state-of-the-art clustering methods, and it significantly reduces the computational time.

Object-Based 3-D Building Change Detection on Multitemporal Stereo Images

Due to the rapid process of urbanization, there is an increasing demand for detecting building changes over time using very high-resolution (VHR) images. Traditional two-dimensional (2-D) change detection methods are limited due to the image perspective variation and illumination discrepancies. One current trend for building detection combines the use of orthophotos and digital surface models (DSMs), because of its robustness against false changes, as well as its capability of providing volumetric information. In this paper, we propose an object-based three-dimensional (3-D) building change detection framework based on supervised classification, which makes use of the height, spectral, and shape information in a combined fashion with object-based analysis. The proposed method follows the following steps: First, a synergic mean-shift segmentation method is applied on the orthophoto with the constraints of the DSM, which derives segments with homogenous spectrum and height. In a second step, the segments are classified with a hybrid decision tree and SVM approach, and then the segments of the building class are merged as building objects for change detection. An initial change indicator (CI) is then computed for each building object concerning height and spectral information. Finally, an adaptive CI updating strategy based on segment overlapping is proposed and the traffic light system based on a dual threshold is used to identify the change status of each building as “change,” “no-change,” and “uncertain change”. The experimental results on scanned aerial stereo images have demonstrated that our proposed framework is able to achieve high-detection accuracy on images with limited spectral quality.

An Improved Mean Shift Segmentation Method of High-Resolution Remote Sensing Image Based on LBP and Canny Features

Mean shift algorithm is a robust approach toward feature space analysis, which has been wildly used for natural scene image and medical image segmentation. Due to fuzzy boundary and low accuracy of Mean shift segmentation method, this paper puts forward to an improved Mean shift segmentation method of high-resolution remote sensing image based on LBP and Canny features. The results show that this improved Mean shift segmentation access can enhance segmentation accuracy compared to the traditional Mean shift.

Census and Segmentation-Based Disparity Estimation Algorithm Using Region Merging

Disparity estimation is an ill-posed problem in computer vision. It is explored comprehensively due to its usefulness in many areas like 3D scene reconstruction, robot navigation, parts inspection, virtual reality and image-based rendering. In this paper, we propose a hybrid disparity generation algorithm which uses census based and segmentation based approaches. Census transform does not give good results in textureless areas, but is suitable for highly textured regions. While segment based stereo matching techniques gives good result in textureless regions. Coarse disparities obtained from census transform are combined with the region information extracted by mean shift segmentation method, so that a region matching can be applied by using affine transformation. Affine transformation is used to remove noise from each segment. Mean shift segmentation technique creates more than one segment of same object resulting into non-smoothness disparity. Region merging is applied to obtain refined smooth disparity map. Finally, multilateral filtering is applied on the disparity map estimated to preserve the information and to smooth the disparity map. The proposed algorithm generates good results compared to the classic census transform. Our proposed algorithm solves standard problems like occlusions, repetitive patterns, textureless regions, perspective distortion, specular reflection and noise. Experiments are performed on middlebury stereo test bed and the results demonstrate that the proposed algorithm achieves high accuracy, efficiency and robustness.

A Line-Based Adaptive-Weight Matching Algorithm Using Loopy Belief Propagation

In traditional adaptive-weight stereo matching, the rectangular shaped support region requires excess memory consumption and time. We propose a novel line-based stereo matching algorithm for obtaining a more accurate disparity map with low computation complexity. This algorithm can be divided into two steps: disparity map initialization and disparity map refinement. In the initialization step, a new adaptive-weight model based on the linear support region is put forward for cost aggregation. In this model, the neural network is used to evaluate the spatial proximity, and the mean-shift segmentation method is used to improve the accuracy of color similarity; the Birchfield pixel dissimilarity function and the census transform are adopted to establish the dissimilarity measurement function. Then the initial disparity map is obtained by loopy belief propagation. In the refinement step, the disparity map is optimized by iterative left-right consistency checking method and segmentation voting method. The parameter values involved in this algorithm are determined with many simulation experiments to further improve the matching effect. Simulation results indicate that this new matching method performs well on standard stereo benchmarks and running time of our algorithm is remarkably lower than that of algorithm with rectangle-shaped support region.

English

Colossal amount of research has been done in creating many different approaches and algorithms for medical image segmentation, but it is still complicated to evaluate all the images. However the problem remains challenging, with no general and unique solution in computer-aided diagnosis. This paper provides medical image segmentation based on Clustering for computer tomography images. In this paper, we consider a mean shift segmentation and medoid shift segmentation method. We validate the mean shift and medoid shift medical image segmentation approach with the parameters in terms of sensitivity, specificity and accuracy. The Real time dataset is used to evaluate the performance of the proposed method. The experimental result shows that the medoid shift segmentation method gives more accurate and robust segmentation results than mean shift segmentation method.

CAMSHIFT IMPROVEMENT WITH MEAN-SHIFT SEGMENTATION, REGION GROWING, AND SURF METHOD

CAMSHIFT algorithm has been widely used in object tracking. CAMSHIFT utilizescolor features as the model object. Thus, original CAMSHIFT may fail when the object color issimilar with the background color. In this study, we propose CAMSHIFT tracker combined withmean-shift segmentation, region growing, and SURF in order to improve the tracking accuracy.The mean-shift segmentation and region growing are applied in object localization phase to extractthe important parts of the object. Hue-distance, saturation, and value are used to calculate theBhattacharyya distance to judge whether the tracked object is lost. Once the object is judged lost,SURF is used to find the lost object, and CAMSHIFT can retrack the object. The Object trackingsystem is built with OpenCV. Some measurements of accuracy have done using frame-basedmetrics. We use datasets BoBoT (Bonn Benchmark on Tracking) to measure accuracy of thesystem. The results demonstrate that CAMSHIFT combined with mean-shift segmentation, regiongrowing, and SURF method has higher accuracy than the previous methods.

An approach based on simulated annealing to optimize the performance of extraction of the flower region using mean-shift segmentation

Flower identification and recognition are tedious and difficult tasks even for humans. Image segmentation based on automatic flower extraction is an essential step for computer-aided flower image recognition and retrieval processes. Furthermore, there is a challenge for segmentation of the object(s) from natural complex background in color images. In this study, a novel performance optimization approach for image segmentation, i.e. simulated annealing-based mean-shift segmentation (SAMS), is proposed and implemented. It is based on the simulated annealing solution of quadratic assignment problem model treated as an image segmentation process using feature-based mean-shift (MS) clustering on color images. The proposed approach is designed to realize a global and unsupervised (i.e., fully automatic) segmentation. It is a modified and optimized version of Backprojection-based mean-shift segmentation (BackMS) method. In conducted segmentation experiments, the performance results of SAMS approach are compared with the ones of BackMS method. Comparison of overall performance results and statistical analysis (i.e., Wilcoxon signed rank median test) show that SAMS approach improves the performance of BackMS method. It is measured as 49.33% when using object bounding boxes and as 51.33% when using object pixel regions.

Robust Object Detection in Military Infrared Image

The innovation of this paper is that we put forward a new algorithm of object detection form military infrared images with texture background according to the Mean-shift smooth and segmentation method combined with eight directions difference clustering. According to the texture characteristics of background, smoothing and clustering is carried out to extract the characteristics of object. The experimental results show that the algorithm is able to extract the object information form complex infrared texture background with better self-adapting and robustness. Future research particularly lies in raising the accuracy of object extracted.

Fusion of Panchromatic and Multispectral Remote Sensing Data Using an Improved Region Regression Model

A region-based regression model is presented to fuse a pair of panchromatic (PAN) and multispectral image (MS) in order to get a pansharpened image with high spectral fidelity. Firstly, the images from two different sensors are segmented into isolated regions respectively by the mean shift segmentation method with embedded edge confidence. Then, three segmentation maps got from two images are used as reference of the following segment-based modeling process. The regression model is established independently for each isolated region by considering the linear relationship between corresponding PAN and MS image pixel values in the region. Finally, a modulation-based fusion technique is employed to get the fusion result. Two pairs of images are used for evaluating the method in both qualitative and quantitative ways. The method shows the superior result.

Ambiguous Region Matching and Boundary Pixel Optimization for Stereo Computation

In this paper, we present a stereo matching algorithm based on planar surface hypothesis. It improves the results of low texture regions and mixed pixels on object boundaries. First, regions are segmented by applying the mean-shift segmentation method. Then we propose a coarse-to-fine algorithm to increase the reliable correspondences in low texture regions. Third, the Belief Propagation algorithm is used to optimize disparity plane labeling. Finally, for a mixed pixel, we utilize the results of the depth plane and the local region of it to regulate its disparity. Experimental results using the Middlebury stereo test show that the performance of our method is high.

An object-oriented daytime land-fog-detection approach based on the mean-shift and full lambda-schedule algorithms using EOS/MODIS data

A new algorithm is presented for land-fog detection using daytime imagery from the Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) data. MODIS images constitute an ideal data source for fog detection due to their outstanding spatial and spectral resolution. In this article, a parameter named the Normalized Difference Fog Index (NDFI) is proposed, based on analysing the spectral character of fog and cloud by utilizing the Streamer radiative-transfer model and MODIS data. A mean-shift segmentation method is used to preliminary segment the NDFI image, and a full lambda-schedule algorithm is then iteratively applied to merge adjacent segments based on the combination of spectral and spatial information. Then, some properties (e.g. mean value of brightness temperature) are calculated for each segment, and each object is identified as either fog or not. The algorithm's performance is evaluated against ground-based measurements over China in winter, and the algorithm is proved to be effective in detecting fog accurately based on three cases.

Texture Based Infrared Military Target Extraction

The innovation of this paper is that it forwards a new algorithm of target extraction form military infrared images with texture background according to the Mean-shift smooth and segmentation method combined with eight directions difference clustering. According to the texture characteristics of background image, smoothing and clustering are both carried out to extract the characteristics of target. The method is relatively simple making it easy for practical applications. The experimental results show that the algorithm is able to extract the target information form complex military infrared texture background with better self-adapting and robustness.

The Registration Detection System of Rotary Screen Printing Machine Based on the Machine Vision Technology

With the development of the computer and the digital image processing technology, application of machine vision technology to make online detection research has been gradually developed, so the machine vision technology has been introduced into the registration detection of rotary screen printing, in order to enhance the accuracy and speed of the registration error’s detection, that is a very significant research topic. Firstly, general planning about the rotary screen printing machine registration detection system has been designed. In addition, texture noise of printing picture has been analyzed, which has been collected by smart camera. Finally, based on mean shift arithmetic for color image segmentation has been put forward. Experiments show that the algorithm based on mean shift segmentation method for color image segmentation textile printing can achieve good results, moreover that can accurately carry out the fabric chromatic region segmentation, which for the registration error detection system design on rotary screen printing machine lays the foundation.