|
|
Robot navigation on the chessboard
Mikulík, Andrej ; Holub, Viliam (advisor) ; Ježek, Pavel (referee)
O ffered thesis investigates localization of mobile robot. Localization is based on computer vision. Projected algorithms recognize square-patterned pad on which robot moves from recorded image. Inclination angle and position of the robot is then analyzed from image. Proposed algorithm was discussed according to tests on real robot.
|
|
| |
|
|
Methods for precise local affine frame constructions on MSERs
Mikulík, Andrej
Feature detection and matching is a fundamental problem in many applications in computer vision. We propose a novel approach that improves repeatability and precision of Local Affine Frames (LAFs) constructed on discretized contours detected by Maximally Stable Extremal Regions (MSERs) detector. Proposed method reconstructs a discretized contour of extremal region by taking into account the intensity function in local neighborhood of the contour points. Additionally we propose a new method for detection of local curvature extrema, based on the refined contour. The extensive experimental evaluation on publicly available datasets showed higher number of correspondences and higher inlier ratio in more than 80\% of the image pairs. Since the procesing time of the contour refinement is negligible, there is no reason not to include the proposed algorithms as a standard extension of MSER detector. Powered by TCPDF (www.tcpdf.org)
|
|
|
Large-Scale Content-Based Sub-Image Search
Mikulík, Andrej ; Matas, Jiří (advisor)
In this work the problems of specific object and image retrieval including the more challenging sub-image are studied. Given a query image of a specific object a retrieval engine returns relevant images of the same object from a database. The thesis focuses on the bag-of-words approach which is one of the most effective content-based approach especially when the specific object covers only a part of the picture, can be occluded or only partially visible. The thesis improves a number of components of the standard bag-of-words retrieval approach. A novel similarity measure for bag-of-words type large scale image retrieval is pre- sented. The similarity function is learned in an unsupervised manner, requires no extra space over the standard bag-of-words method and is more discriminative than both L2- based soft assignment and Hamming embedding. The novel similarity function achieves mean average precision that is superior to any result published in the literature on the standard datasets and protocols. We study the effect of a fine quantization and very large vocabularies (up to 64 mil- lion words) and show that the performance of specific object retrieval increases with the size of the vocabulary. This observation is in contradiction with previously published results. We further demonstrate that the large...
|
|
|
Large-Scale Content-Based Sub-Image Search
Mikulík, Andrej ; Matas, Jiří (advisor)
In this work the problems of specific object and image retrieval including the more challenging sub-image are studied. Given a query image of a specific object a retrieval engine returns relevant images of the same object from a database. The thesis focuses on the bag-of-words approach which is one of the most effective content-based approach especially when the specific object covers only a part of the picture, can be occluded or only partially visible. The thesis improves a number of components of the standard bag-of-words retrieval approach. A novel similarity measure for bag-of-words type large scale image retrieval is pre- sented. The similarity function is learned in an unsupervised manner, requires no extra space over the standard bag-of-words method and is more discriminative than both L2- based soft assignment and Hamming embedding. The novel similarity function achieves mean average precision that is superior to any result published in the literature on the standard datasets and protocols. We study the effect of a fine quantization and very large vocabularies (up to 64 mil- lion words) and show that the performance of specific object retrieval increases with the size of the vocabulary. This observation is in contradiction with previously published results. We further demonstrate that the large...
|
|
|
Methods for precise local affine frame constructions on MSERs
Mikulík, Andrej
Feature detection and matching is a fundamental problem in many applications in computer vision. We propose a novel approach that improves repeatability and precision of Local Affine Frames (LAFs) constructed on discretized contours detected by Maximally Stable Extremal Regions (MSERs) detector. Proposed method reconstructs a discretized contour of extremal region by taking into account the intensity function in local neighborhood of the contour points. Additionally we propose a new method for detection of local curvature extrema, based on the refined contour. The extensive experimental evaluation on publicly available datasets showed higher number of correspondences and higher inlier ratio in more than 80\% of the image pairs. Since the procesing time of the contour refinement is negligible, there is no reason not to include the proposed algorithms as a standard extension of MSER detector. Powered by TCPDF (www.tcpdf.org)
|
|
| |
|
|
Robot navigation on the chessboard
Mikulík, Andrej ; Ježek, Pavel (referee) ; Holub, Viliam (advisor)
O ffered thesis investigates localization of mobile robot. Localization is based on computer vision. Projected algorithms recognize square-patterned pad on which robot moves from recorded image. Inclination angle and position of the robot is then analyzed from image. Proposed algorithm was discussed according to tests on real robot.
|
guest ::
