Wyniki wyszukiwania - BazTech

1

Comparison of selected methods of characteristic point detection in satellite images

Kulawik J.

Przegląd Elektrotechniczny

|

2018

|

R. 94, nr 1

139--143

EN

The article presents two methods of detecting objects in images of the surface of the earth from the air. The search was performed using local characteristic features, i.e. key points. In the first method, the corner detection was supplied using the Harris & Stephens algorithm. The descriptors were built for detection key points by the FREAK algorithm. In the second method the blobs were provided by the SURF algorithm. The descriptors were built by the SURF algorithm. After the usage of the above methods, a comparison was made. The obtained results were shown on the example images.

PL

W artykule przedstawiono dwa przykłady detekcji obiektów w zdjęciach powierzchni ziemi z powietrza. Wyszukiwanie wykonano przy użyciu cech charakterystycznych. W pierwszym przykładzie dokonano detekcji narożników przy użyciu algorytmu Harris & Stephens. Następnie zbudowano deskryptory do znalezionych punktów kluczowych w oparciu o algorytm FREAK. W drugim przykładzie zastosowano metodę SURF do odnalezienia plamek i zbudowania ich deskryptorów. Po użyciu powyższych metod dokonano porównania. Uzyskane wyniki zaprezentowano na przykładowych zdjęciach.

2

Key-point Detection based Fast CU Decision for HEVC Intra Encoding

Xu Z., Min B., Cheung R. C. C.

International Journal of Electronics and Telecommunications

|

2018

|

Vol. 64, No. 3

321--327

EN

As the most recent video coding standard, High Efficiency Video Coding (HEVC) adopts various novel techniques, including a quad-tree based coding unit (CU) structure and additional angular modes used for intra encoding. These new techniques achieve a notable improvement in coding efficiency at the penalty of significant computational complexity increase. Thus, a fast HEVC coding algorithm is highly desirable. In this paper, we propose a fast intra CU decision algorithm for HEVC to reduce the coding complexity, mainly based on a key-point detection. A CU block is considered to have multiple gradients and is early split if corner points are detected inside the block. On the other hand, a CU block without corner points is treated to be terminated when its RD cost is also small according to statistics of the previous frames. The proposed fast algorithm achieves over 62% encoding time reduction with 3.66%, 2.82%, and 2.53% BD-Rate loss for Y, U, and V components, averagely. The experimental results show that the proposed method is efficient to fast decide CU size in HEVC intra coding, even though only static parameters are applied to all test sequences.

3

Comparison of selected methods for detecting a reference element using key points in static images

Kulawik J.

Image Processing & Communications

|

2017

|

Vol. 22, no. 2

15--20

EN

The article presents a possible way to detect key points. The tests were carried out by the case of detection of a reference object in static images. For comparative purposes, Chris Harris & Mike Stephens [12] and Speeded-Up Robust Features (SURF) detectors [2, 3] were used. The descriptors were built based on the Fast Retina Key point (FREAK) [1, 17] and SURF algorithms [2, 3]. Six different configurations of key point detection methods with the above descriptors were implemented. The obtained results have been presented on exemplary images and in the table. They show that this type of detection of an element of interest can be successful and should be developed.

4

Fusion of door and corner features for scene recognition

Chacon-Murguia M. I., Sandoval-Rodriguez R., Guerrero-Saucedo C. P.

Journal of Automation Mobile Robotics and Intelligent Systems

|

2011

|

Vol. 5, No. 1

68-76

EN

Scene recognition is a paramount task for autonomous systems that navigate in open scenarios. In order to achie ve high scene recognition performance it is necesary to use correct information. Therefore, data fusion is beco ming a paramount point in the design of scene recognition systems. This paper presents a scenery recognition system using a neural network hierarchical approach. The system is based on information fusion in indoor scenarios. The system extracts relevant information with respect to color and landmarks. Color information is related mainly to localization of doors. Landmarks are related to corner de tection. The corner detection method proposed in the pa per based on corner detection windows has 99% detection of real corners and 13.43% of false positives. The hierar chical neural systems consist on two levels. The first level is built with one neural network and the second level with two. The hierarchical neural system, based on feed for ward architectures, presents 90% of correct recognition in the first level in training, and 95% in validation. The first ANN in the second level shows 90.90% of correct recogni tion during training, and 87.5% in validation. The second ANN has a performance of 93.75% and 91.66% during training and validation, respectively. The total perfor mance of the systems was 86.6% during training, and 90% in validation.

5

Analysis of operators for detection of corners set in automatic image matching

Zawieska D.

Archiwum Fotogrametrii, Kartografii i Teledetekcji

|

2011

|

Vol. 22

423--436

EN

Reconstruction of three dimensional models of objects from images has been a long lasting research topic in photogrammetry and computer vision. The demand for 3D models is continuously increasing in such fields as cultural heritage, computer graphics, robotics and many others. The number and types of features of a 3D model are highly dependent on the use of the models, and can be very variable in terms of accuracy and time for their creation. In last years, both computer vision and photogrammetric communities have approached the reconstruction problems by using different methods to solve the same tasks, such as camera calibration, orientation, object reconstruction and modelling. The terminology which is used for addressing the particular task in both disciplines is sometimes diverse. On the other hand, the integration of methods and algorithms coming from them can be used to improve both. The image based modelling of an object has been defined as a complete process that starts with image acquisition and ends with an interactive 3D virtual model. The photogrammetric approach to create 3D models involves the followings steps: image pre-processing, camera calibration, orientation of images network, image scanning for point detection, surface measurement and point triangulation, blunder detection and statistical filtering, mesh generation and texturing, visualization and analysis. Currently there is no single software package available that allows for each of those steps to be executed within the same environment. For high accuracy of 3D objects reconstruction operators are required as a preliminary step in the surface measurement process, to find the features that serve as suitable points when matching across multiple images. Operators are the algorithms which detect the features of interest in an image, such as corners, edges or regions. This paper reports on the first phase of research on the generation of high accuracy 3D model measurement and modelling, focusing upon the application of different operators for accurate feature point extraction. The implementation of those operators is discussed and performance of differen operators is analysed. The optimal operator for high accuracy close range object reconstruction is then highlighted. This research has facilitated a development of the feature extraction and image measurement process that will be central to the development of an automatic procedure for high accuracy point cloud generation in multi image networks where robust orientation and 3D point determination will facilitate surface measurement and modelling within a single software system.

6

Robust Candidate Pruning Approach Based on the PSO-SVM for Fast Corner Detection with Noise Tolerance in Gray-Level Images

Liu M-T., Yu P-T.

Fundamenta Informaticae

|

2009

|

Vol. 95, nr 4

491-510

EN

This paper presents a fast two-stage corner detector with noise tolerance. In the first stage, a novel candidate pruning approach based on PSO-SVM is proposed to select candidatecorner pixels which have great potential to be corners. In the second stage the Harris corner detector is employed to recognize real corners among the candidate-corner pixels. The parameters and feature selection of SVM classifier is optimized by using particle swarm optimization (PSO). The method takes advantage of the minimum structure risk of SVM and the quickly globally optimizing ability of PSO. Generally speaking, corners are considered as the junction of edges. Thus, edge pixels with a high gradient in more than one direction should be selected as candidate corners. Meanwhile, impulse noise often corrupts digital images while images are transmitted over an unreliable channel or are captured using a camera with faulty sensors. Noise-corrupted pixels usually cause serious false detection problems in most corner detectors. The proposed PSO-SVM candidate pruning approach detects noisy pixels and excludes them from being candidate corners to enhance the noise tolerance of the corner detector. Through the well-selection of candidate corners, the proposed candidate pruning approach can 1) enhance the noise tolerance capability, and 2) reduce the computational effort of the corner detectors.