Renbo Xia

An Optimal Initialization Technique for Improving the Segmentation Performance of Chan-Vese Model

In level set method, initialization mode not only influences evidently the implementation efficiency but also relates directly to the final results. The paper presents an new initialization scheme for improving the segmentation performance of Chan-Vese model. The proposed initialization scheme consists of two stages. The first stage computes rough edges by using canny edge detection operator. The second stage removes noise edges and redundant edges by a morphological filter, and generates closed contours by iteratively connecting edge points according to a local cost function. In comparison with the primal Chan-Vese model, experimental data show that the Chan-Vese model equipped with our initialization scheme provides superior segmentation results and takes less computational cost.

Recognition of Airport Runways in FLIR Images Based on Knowledge

Airport runway recognition technology would play an important role in developing intelligent weapon systems in the future. In this letter, a method of automatically finding runways in forward looking infrared (FLIR) images is proposed based on the knowledge of vision. First, the line segments in the images are extracted by a fast line segment detector (LSD) and an improved line segment linking method. Then, the regions of interest (ROI) of runways are detected using some constraint rules based on the direction, gradient, and width of line segment pairs. Afterward, an ROI length backtracking technique based on texture distribution is presented to retrieve the complete ROI. Finally, using runway regional self-similarity and contextual information, several decision criteria are formulated to accurately recognize the runway. Experimental results on the FLIR images with different imaging ranges show that the proposed algorithm is robust and has a good real-time performance.

A computing method for accurate slice contours based on an STL model

The STL model has become a standard in rapid prototyping (RP) technology, but slice errors are high because of tessellation. To improve the precision of slice contours, a novel slicing method is presented to compute accurate contour curves based on an STL model. First, the normal vector of a triangle facet to be sliced is calculated. An interpolation curve between two vertices is constructed in a normal section to approximate the original curve in the normal section in terms of the position of the two vertices and the tangent vector at the two vertices. The exact points of the slice contour are obtained by calculating the intersection points between the interpolation curve and the slice plane. Finally, accurate smooth contours are achieved by fitting the intersection points with a cubic B-spline curve. The experimental results indicate that the presented slicing method improves the precision of cross contours.

NURBS curve interpolation algorithm based on tool radius compensation method

This paper focuses on developing an algorithm that can generate toolpaths in NURBS form for smooth, high speed and accurate machining. The initial toolpaths are obtained by tool radius compensation method which is based on the workpiece boundary offsetting. According to different lengths and the continuous short block (CSB) criterion, these offset linear segments can be regarded as CSBs or long straight segments. Junctions are located where the curvature value is greater than the preset curvature threshold value or where it changes abruptly, or at the two end points of any long straight segment. During machining, the NURBS fitting module first looks ahead several CSBs and converts them into parametric curves in real time. During the conversion, continuities of the position, slope or even curvature at the transition of the parametric curves and unfitted line segments can be guaranteed. Then the acceleration/deceleration feedrate-planning scheme is proposed to determine the transition feedrate at the junction between the fitted curves and unfitted long straight segments, and the corner feedrate within the fitted curve. Simulations and experiments show that the proposed algorithm can significantly improve machining accuracy and reduce cutting time to satisfy today’s high-speed and high-accurate machining requirements.

Application of genetic algorithm in rapid prototyping

Part-building orientation (PBO) and scanning direction of path planning (SDPP) are two tasks of process plan in rapid prototyping technology (RPT). Through investigating the geometric issues of STL model and process planning of RPT. This paper establishes optimizing model based on the considerations of staircase effect, support area and production time. And then, through analyzing the hatching characteristic of polygonal contours, approximately optimization model of direction of scanning vectors is established. The best part-building orientation is obtained by solving the general satisfactory degree function employing genetic algorithm and the optimal scanning direction is also solved by genetic algorithm. Two cases of experiment show that GA can effectively solve not only the determination problem of part-building orientation but also the optimization problem of scanning direction in RP.

Robust Algorithm for Detecting the Maximum Inscribed Circle

In this paper, we propose a new robust algorithm for the detection of the maximum inscribed circles (MIC) in images. We first use a vector distance transformation (VDT) strategy to create a distance field. Then, we globally search the maximal value in distance field to extract the medial axes. Finally, we give a procedure to determine the center and radius of MIC. Our experimental results show indeed that new algorithm is capable of detecting the MIC with excellent accuracy and high efficiency under various image conditions.

Double spiral tool-path generation and linking method for complex pocket machining

ABSTRACT Complex pockets with one or more islands have been widely used in industrial and manufacturing production. In this paper, a new double spiral tool-path generation and linking method are proposed for complex pockets with islands which can be used for high-speed machining (HSM) is used. Taking into account the path interval, step length and other processing parameters, precise milling can be achieved without cutter lifting and retraction motions to guarantee machining accuracy and reduce processing time. The method has been implemented in several simulations and validated successfully through the actual machining of a complicated pocket. The results indicate that this method is superior to other existing machining methods, and it can achieve HSM of complicated shaped pockets based on parametric surface.

Fully automatic matching of circular markers for camera calibration

Camera calibration is one of critical steps in computer vision, also an exhaustive process because substantial human computer interactions are frequently required to deal with the matching problem. In this paper, an automatic matching method of markers for camera calibration is presented based on the local architecture characteristics of a new planar circle pattern, which can be applied to a wide range of spoiled images including those of distortion, noise, and blur. Firstly, a robust ellipse detector is developed to extract ellipses from the image. After the centroids of the ellipses are triangulated into triangle network by Delaunay method to find k annular neighborhood of each ellipse, a cost function is defined to locate orientation ellipses for the homography matrix between the image and the calibration pattern. Then the correspondences between the marker ellipses and the marker circles are established by a homography method followed by a point sets registration strategy. Finally, the method is tested with real and transformed images to show its accuracy and robustness, which is entirely unsupervised and easily embedded into the exiting algorithms.

Automatic extrinsic calibration of display-camera system using an annular mirror

Our work focuses on the extrinsic calibration of a display-camera system where the camera has no direct view of the display. An annular planar mirror is used to reflect the display so that the camera can capture the reflection images. The position of the mirror can be obtained after outer circle is detected. The position parameters are determined uniquely by putting two orthogonal lines in the background and optimized by re-projecting the inner circle to the image plane. The display pixels are encoded using gray code and its imaged position in the mirror can be obtained by solving the PnP problem. Finally, the real extrinsic parameters between the camera and the display are obtained according to the mirror imaging principle. Our approach is simple and fully automatic without any manual intervention. Both simulation and real experiments validate our approach.

An automatic registration algorithm of infrared and visible images based on hybrid image features

In this paper, a new method is presented to match a pair of visible and infrared images of same scene based on hybrid visual features including line segments and interest points. First, improved Harris corner extraction method and line segment detector method is used to extract feature points and segments. Then, a novel descriptor integrating the information of line segments and interest points is proposed. Finally, the nearest neighbor algorithm is utilized to match the descriptors, and the RANSAC(Random Sample Consensus) algorithm is employed to rule out the wrong match pairs. The performances are evaluated by extensive experiments on real images. The results show that the proposed algorithm can take advantage of similar structures between the multimodal images to realize automatic matching efficiently.