Under-segmentation Error Research Articles

Weighted‐graph‐based supervoxel segmentation of 3D point clouds in complex urban environment

A new method for supervoxel segmentation based on a weighted graph in complex urban environment is proposed. The proposed algorithm is designed to minimise the cost of the shortest path in the weighted graph with consideration of characteristics of the sensor for complex urban environment. We collected point clouds in a complex urban environment to evaluate the proposed method. The performance of the proposed algorithm was evaluated using boundary recall and under-segmentation errors.

BUILDING CHANGE DETECTION FROM LIDAR POINT CLOUD DATA BASED ON CONNECTED COMPONENT ANALYSIS

Abstract. Building data are one of the important data types in a topographic database. Building change detection after a period of time is necessary for many applications, such as identification of informal settlements. Based on the detected changes, the database has to be updated to ensure its usefulness. This paper proposes an improved building detection technique, which is a prerequisite for many building change detection techniques. The improved technique examines the gap between neighbouring buildings in the building mask in order to avoid under segmentation errors. Then, a new building change detection technique from LIDAR point cloud data is proposed. Buildings which are totally new or demolished are directly added to the change detection output. However, for demolished or extended building parts, a connected component analysis algorithm is applied and for each connected component its area, width and height are estimated in order to ascertain if it can be considered as a demolished or new building part. Finally, a graphical user interface (GUI) has been developed to update detected changes to the existing building map. Experimental results show that the improved building detection technique can offer not only higher performance in terms of completeness and correctness, but also a lower number of undersegmentation errors as compared to its original counterpart. The proposed change detection technique produces no omission errors and thus it can be exploited for enhanced automated building information updating within a topographic database. Using the developed GUI, the user can quickly examine each suggested change and indicate his/her decision with a minimum number of mouse clicks.

Efficient Segmentation and Classification of Mammogram Images with Fuzzy Filtering

Background: Worldwide and across India breast cancer is the most common cause of cancer deaths in women. Early or timely detection leads to decrease in the mortality rate. Hence, classification of patients based on the size of the tumor/abnormal masses and less treatment cost must be high priority. Methods/Statistical Analysis: In this paper mammogram images are being acquired from real time and standard databases for imaging the suspected patients. The main purposes of the suggested methods are to diagnose the cancer using fuzzy rules with minimum phases in implementation. Important factors were drawn from the images for subsequent investigation and analysis with the help of Fuzzy Enhanced Mammogram Segmentation scheme. The paper presents two methods and is implemented in (i.e. FEM1 and FEM2) Mat lab programming environment. Results: The images examined were marked by qualified Radiologist and extracted the images using Photoshop tool. The proposed methodologies were evaluated for real images and Mammographic Image Analysis Society (MIAS) database images consists of 320 images for 160 patients each of 1024x1024 resolutions based gray level images. Based on the results it is found that the CDR for FEM1 is 87% whereas FEM2 demonstrates only 77% and also takes 6.25 times lesser execution time. Radiologists need more precise and reduced processing time making the outcome of FEM1 method more practicable. For the evaluation of performance, statistical properties like Similarity Index (SI), Correct Detection Ratio (CDR), and Under Segmentation Error (USE) are computed. The paper presents computations of segmentation efficiency, enhancement performance and comparative analysis between the method 1 and method 2 in terms of segmentation efficiency and CPU processing time. Finally Support Vector Method is used to classify whether the mammogram under test is normal or abnormal. Conclusion: FEM1 outperforms other similar methods. The proposed work provides faster, accurate results and more useful for the diagnosis and classify the abnormal tumors or masses at a cheaper cost.

Video segmentation with L0 gradient minimization

Video segmentation is an important preprocessing step for many computer vision and graphics tasks. Its main goal is to group the voxels in the video volume with similar appearance and motion into spatio-temporally consistent supervoxels. In this paper, we formulate video segmentation as an L0 gradient minimization problem, so that the spatio-temporal coherence can be effectively enforced through a gradient sparsity pursuit way. In our method, the appearance and motion descriptor space is first built for over-segmented image patches of each video frame. Then the L0 gradient minimization is performed in the descriptor space, for both spatial and temporal dimensions. To solve the non-convex L0 norm minimization problem, we extend the fused coordinate descent algorithm from 2D image grids to 3D video volume. We conduct quantitative evaluation of our method in a public video segmentation benchmark LIBSVX. The experimental results demonstrate our superior performance to state-of-the-arts in segmentation accuracy and undersegmentation error, and comparable performance in boundary recall and explained variation.

An Improved SLIC Superpixels using Reciprocal nearest Neighbor Clustering

An improved SLIC method using uniform segmentation and reciprocal nearest neighbor (RNN) clustering is presented in this paper. This approach is made of two steps. First, image is segmented to a large number of regular homogeneous and small regions which are similar to cell. Second, instead of the original image pixels, small regions segmented are regarded as input of RNN clustering. A new similarity criterion is decided by regional diversity of average value normalized and variance. Regional constraint filter limited the large size of superpixel guarantees the uniformity and compactness of superpixel. Finally, the regions in a small range of distance are merged by RNN clustering. Results of experiment on BSDS 500 dataset of natural images show the proposed method has advantages of high boundary recall and low under-segmentation error over SLIC on small numbers superpixel.

Refining High Spatial Resolution Remote Sensing Image Segmentation for Man-made Objects through aCollinear and Ipsilateral Neighborhood Model

AbstractMan-made objects, such as buildings and roads, which areimportant targets for information extraction from high spatial resolution (HSR) remote sensing images, often feature straight boundaries. This study employs this knowledge on HSR image segmentation by embedding a straight-line constraint in regionbased image segmentation. A new concept called collinear and ipsilateral neighborhood is proposed and applied to hardboundary constraint-based image segmentation for accuracy improvement. In the experimental areas, the method accuracy measured by recall ratio r increases from 0.036 to 0.048 (on the average) after the refinement, with significantly smaller decreases in precision p that are all less than 0.006. In sum, the proposed technique effectively reduces over-segmentation errors and maintains the same level of under-segmentation error ratio, particularly in man-made areas. It facilitates subsequent objectbased image analyses, including feature extraction, object recognition, and classification.

Fast and Scalable Mesh Superfacets

AbstractIn the field of computer vision, the introduction of a low‐level preprocessing step to oversegment images into superpixels – relatively small regions whose boundaries agree with those of the semantic entities in the scene – has enabled advances in segmentation by reducing the number of elements to be labeled from hundreds of thousands, or millions, to a just few hundred. While some recent works in mesh processing have used an analogous oversegmentation, they were not intended to be general and have relied on graph cut techniques that do not scale to current mesh sizes. Here, we present an iterative superfacet algorithm and introduce adaptations of undersegmentation error and compactness, which are well‐motivated and principled metrics from the vision community. We demonstrate that our approach produces results comparable to those of the normalized cuts algorithm when evaluated on the Princeton Segmentation Benchmark, while requiring orders of magnitude less time and memory and easily scaling to, and enabling the processing of, much larger meshes.

Transcript mapping for handwritten Chinese documents by integrating character recognition model and geometric context

Creating document image datasets with ground-truths of regions, text lines and characters is a prerequisite for document analysis research. However, ground-truthing large datasets is not only laborious and time consuming but also prone to errors due to the difficulty of character segmentation and the large variability of character shape, size and position. This paper describes an effective recognition-based annotation approach for ground-truthing handwritten Chinese documents. Under the Bayesian framework, the alignment of text line images with text transcript, which is the crucial step of annotation, is formulated as an optimization problem by incorporating geometric context of characters and character recognition model. We evaluated the alignment performance on a Chinese handwriting database CASIA-HWDB, which contains nearly four million character samples of 7356 classes and 5091 pages of unconstrained handwritten texts. The experimental results demonstrate the superiority of recognition-based text line alignment and the benefit of integrating geometric context. On a test set of 1015 handwritten pages (10,449 text lines), the proposed approach achieved character level alignment accuracy 92.32% when involving under-segmentation errors and 99.04% when excluding under-segmentation errors. The tool based on the proposed approach has been practically used for labeling handwritten Chinese documents.

VCells: Simple and Efficient Superpixels Using Edge-Weighted Centroidal Voronoi Tessellations

VCells, the proposed Edge-Weighted Centroidal Voronoi Tessellations (EWCVTs)-based algorithm, is used to generate superpixels, i.e., an oversegmentation of an image. For a wide range of images, the new algorithm is capable of generating roughly uniform subregions and nicely preserving local image boundaries. The undersegmentation error is effectively limited in a controllable manner. Moreover, VCells is very efficient with core computational cost at O(K√n(c)·N) in which K, n(c), and N are the number of iterations, superpixels, and pixels, respectively. Extensive qualitative discussions are provided, together with the high-quality segmentation results of VCells on a wide range of complex images. The simplicity and efficiency of our model are demonstrated by complexity analysis, time, and accuracy evaluations.

Assessing object-based classification: advantages and limitations

The advantages of object-based classification over the traditional pixel-based approach are well documented. However, the potential limitations of object-based classification remain less explored. In this letter, we assess the advantages and limitations of an object-based approach to remote sensing image classification relative to a pixel-based approach. We first quantified the negative impacts of under-segmentation errors on the potential accuracy of object-based classification by developing a new segmentation accuracy measure. Then we evaluated the advantages and limitations of object-based classification by quantifying their overall effects relative to pixel-based classification, with respect to their classification units and features at multiple segmentation scales. The results based on a QuickBird satellite image indicate that (1) segmentation accuracies decrease with increasing segmentation scales and the negative impacts of under-segmentation errors become significantly large at large scales and (2) there are both advantages and limitations in using object-based classification, and their trade-off determines the overall effect of object-based classification, which is dependent on the segmentation scales.

Improved Automatic Detection and Segmentation of Cell Nuclei in Histopathology Images

Automatic segmentation of cell nuclei is an essential step in image cytometry and histometry. Despite substantial progress, there is a need to improve accuracy, speed, level of automation, and adaptability to new applications. This paper presents a robust and accurate novel method for segmenting cell nuclei using a combination of ideas. The image foreground is extracted automatically using a graph-cuts-based binarization. Next, nuclear seed points are detected by a novel method combining multiscale Laplacian-of-Gaussian filtering constrained by distance-map-based adaptive scale selection. These points are used to perform an initial segmentation that is refined using a second graph-cuts-based algorithm incorporating the method of alpha expansions and graph coloring to reduce computational complexity. Nuclear segmentation results were manually validated over 25 representative images (15 in vitro images and 10 in vivo images, containing more than 7400 nuclei) drawn from diverse cancer histopathology studies, and four types of segmentation errors were investigated. The overall accuracy of the proposed segmentation algorithm exceeded 86%. The accuracy was found to exceed 94% when only over- and undersegmentation errors were considered. The confounding image characteristics that led to most detection/segmentation errors were high cell density, high degree of clustering, poor image contrast and noisy background, damaged/irregular nuclei, and poor edge information. We present an efficient semiautomated approach to editing automated segmentation results that requires two mouse clicks per operation.