The extraction of contours using deformable models, such as snakes, is a problem of great interest in computervision, particular in areas of medical imaging and tracking. Snakes have been widely studied, and many met...
详细信息
ISBN:
(纸本)9780769531533
The extraction of contours using deformable models, such as snakes, is a problem of great interest in computervision, particular in areas of medical imaging and tracking. Snakes have been widely studied, and many methods are available. In most cases, the snake converges towards the optimal contour by minimizing a sum of internal (prior) and external (image measurement) energy terms. This approach is elegant, but frequently mis-converges in the presence of noise or complex contours. To address these limitations, a novel discrete snake is proposed which treats the two energy terms separately. Essentially, the proposed method is a deterministic iterative statistical data fusion approach, in which the visual boundaries of the object are extracted, ignoring any prior, employing a Hidden Markov Model(HMM) and Viterbi search, and then applying importance sampling to the boundary points, on which the shape prior is asserted. The proposed implementation is straightforward and achieves dramatic speed and accuracy improvement. Compared to four other published methods and across six different images (two original, four published), the proposed method is demonstrated to be, on average, 7 times faster with a 45 percent reduction in the mean square error Only the proposed method was able to successfully segment the desired object in each test image.
A technique to improve the positional accuracy of mobile ground-based LIDAR systems is proposed. Terrapoint's TITAN (TM) system scans the same objects at different times, so by aligning scans, any drift over time ...
详细信息
ISBN:
(纸本)9780769531533
A technique to improve the positional accuracy of mobile ground-based LIDAR systems is proposed. Terrapoint's TITAN (TM) system scans the same objects at different times, so by aligning scans, any drift over time can be estimated This paper describes a simple way of tessellating the scanned data into segments based on the vehicle's path. Principal Components Analysis is then used to estimate how well pairs of segments will align when registered with an Iterative Closest Point algorithm. The results show that this analysis does indeed find segments which are likely to register well. Finally a more formal method to analyze the results is proposed to better determine the quality of the registration so that it can be used to improve the position estimate for the LIDAR system.
This paper is focused on an adaptive vision system for the guidance of a robot to intercept a non-cooperative target satellite with unknown dynamics parameters. A Kalman filter is developed to reliably estimate the st...
详细信息
ISBN:
(纸本)9781424420575
This paper is focused on an adaptive vision system for the guidance of a robot to intercept a non-cooperative target satellite with unknown dynamics parameters. A Kalman filter is developed to reliably estimate the states of the object as well as all of its inertial parameters-namely, the moment-of-inertia ratios, the center-of-mass location, and the orientation of the principle axes-from vision information. The estimates are then used to optimally plan the motion of the manipulator. The optimization performance index includes the time of travel and the weighted norms of the end-effector velocity and acceleration, and it is subject to the conditions that the robot end-effector and the satellite gasping point arrive at the rendezvous point with the same velocity and that the interception occurs within the robot reach. The variational method is used to find the optimal path, which turns out to be the solution of a fourth-order differential equation. Subsequently, a closed-form solution is obtained. The solution to the optimal terminal-time problem is also obtained from the Hamiltonian of the entire system. Experiments are conducted by using a robotic arm to move a satellite mockup according to orbital mechanics and measuring the satellite pose by a laser camera system. The results demonstrate a successful grasping even though the inertial parameters are not known by the control system.
Discerning depth from stereopsis is difficult because the quality of un-cooled sensors is not sufficient for generating dense depth maps. We show how to produce sparse disparity maps from un-calibrated infrared stereo...
详细信息
ISBN:
(纸本)9780769531533
Discerning depth from stereopsis is difficult because the quality of un-cooled sensors is not sufficient for generating dense depth maps. We show how to produce sparse disparity maps from un-calibrated infrared stereo images which can be interpolated to produce a dense/semi-dense depth field. In our proposed technique, the sparse disparity map is produced by a robust features-based stereo matching method capable of dealing with the problems of infrared images, such as low resolution and high noise. Initially, a set of stable features are extracted from stereo pairs using the phase congruency model, which contrary to the gradient-based feature detectors, provides features that are invariant to geometric transformations. Then, a set of Log-Gabor wavelet coefficients at different orientations and frequencies is used to analyze and describe the extracted features for matching. The resulting sparse disparity map is then refined by triangular and epipolar geometrical constraints. In densifying the sparse map, a watershed transformation is applied to divide the image into several segments, where the disparity inside each segment is assumed to vary smoothly. The surface of each segment is then reconstructed independently by fitting a spline to its known disparities. Results indicate strong correlation with ground truth. The marginal results from the watershed segmentation on IR is chiefly responsible for the errors in the renonstructed depth map.
Shadows are physical phenomena observed in most natural scenes. They can cause many problems in computervision performance. The paper addresses the problem of shadow detection and removal from solo image of natural s...
详细信息
ISBN:
(纸本)9783540885160
Shadows are physical phenomena observed in most natural scenes. They can cause many problems in computervision performance. The paper addresses the problem of shadow detection and removal from solo image of natural scenes. Our method is based on Retinex theory which is an image enhancement and illumination compensation model of the lightness and color perception of human vision. The approach proposed here does not use any special prior knowledge and assumptions. The shadow extraction algorithm originates from a simple idea that the human-vision-based Retinex has the natural ability to enhance the shadow region of an image no matter it is penumbrae or umbrae. The penumbrae and umbrae regions will be highlighted if we compare the Retinex-enhanced images with original images. Then through adding smooth light forcibly to shadow edges and introducing shadow edge masks, we reduce the effects of shadow edges in the Retinex enhancement processing. Experiment results validate the approach.
Pivotal techniques of cotton-harvestrobots were summarized, including image segmentation, features generation, artificial classifiers and performances evaluation. Solutions based on machine vision and pattern recogni...
详细信息
ISBN:
(纸本)9780387772523
Pivotal techniques of cotton-harvestrobots were summarized, including image segmentation, features generation, artificial classifiers and performances evaluation. Solutions based on machine vision and pattern recognition were analyzed to distinguish ripe from under-ripe/over-ripe cottons, and rank cottons according to government grading standards.
vision navigation and location is a main function in the vision system of intelligent agricultural mobile robot, and the edge feature of images is an important feature for vision navigation and location. According to ...
详细信息
ISBN:
(纸本)9780387772523
vision navigation and location is a main function in the vision system of intelligent agricultural mobile robot, and the edge feature of images is an important feature for vision navigation and location. According to the characteristic of cropland scenery, a compactly supported dyadic antisymmetric wavelet with respect to origin is brought forward to detect edges of cropland image. A set of filters were given that can be used to construct the edge detecting wavelet. The edge features are provided by determining the local maxim of wavelet coefficient at dyadic scale of the image. After the computer simulation was carried out on the cropland image, the continuous and smooth edge image can be got. The edges of cropland scenery are extracted accurately. The experiment result reveals that the method is efficient and practicable.
暂无评论