Adam L. Kaczmarek

Interactive Query Expansion With the Use of Clustering-by-Directions Algorithm

This paper concerns clustering-by-directions algorithm. The algorithm introduces a novel approach to interactive query expansion. It is designed to support users of search engines in forming Web search queries. When a user executes a query, the algorithm shows potential directions in which the search can be continued. This paper describes the algorithm, and it presents an enhancement which reduces the computational complexity of the algorithm. Moreover, in this paper, a new type of interface is introduced. It is based on a tag cloud, in which terms are located in a radial arrangement. This paper also presents the new experimental results and the evaluation of the algorithm.

Clustering by directions algorithm to narrow search queries

This article describes a novel algorithm called clustering by directions. This algorithm is designed to create a list of words which serve as suggestions for users of search engines how to modify their current search query. The algorithm focuses on preparing various suggestions. It concerns modifying queries by adding more words to them. A list of suggested words is presented to the user in a form of a tag cloud. Experiments which were performed with the algorithm show its usefulness.

Improving depth maps of plants by using a set of five cameras

Abstract. Obtaining high-quality depth maps and disparity maps with the use of a stereo camera is a challenging task for some kinds of objects. The quality of these maps can be improved by taking advantage of a larger number of cameras. The research on the usage of a set of five cameras to obtain disparity maps is presented. The set consists of a central camera and four side cameras. An algorithm for making disparity maps called multiple similar areas (MSA) is introduced. The algorithm was specially designed for the set of five cameras. Experiments were performed with the MSA algorithm and the stereo matching algorithm based on the sum of sum of squared differences (sum of SSD, SSSD) measure. Moreover, the following measures were included in the experiments: sum of absolute differences (SAD), zero-mean SAD (ZSAD), zero-mean SSD (ZSSD), locally scaled SAD (LSAD), locally scaled SSD (LSSD), normalized cross correlation (NCC), and zero-mean NCC (ZNCC). Algorithms presented were applied to images of plants. Making depth maps of plants is difficult because parts of leaves are similar to each other. The potential usability of the described algorithms is especially high in agricultural applications such as robotic fruit harvesting.

Stereo vision with Equal Baseline Multiple Camera Set (EBMCS) for obtaining depth maps of plants

The multi-camera set is applied for obtaining distances to plants.The method for merging disparity maps from different cameras is proposed.The proposed vision system is over 20% more accurate than a stereo camera.The system is designed for autonomous harvesting robots. This paper presents a method of improving the estimation of distances between an autonomous harvesting robot and plants with ripe fruits by using the vision system based on five cameras. The system is called Equal Baseline Multiple Camera Set (EBMCS). EBMCS has some features of a camera matrix and a camera array. EBMCS is regarded as a set of stereo cameras for estimating distances by obtaining disparity maps and depth maps. This paper introduces Exceptions Excluding Merging Method (EEMM) which makes it possible to improve the quality of disparity maps by integrating maps acquired from individual stereo cameras included in EBMCS. The method was tested with eight different stereo matching algorithms including Efficient Large-scale Stereo Matching (ELAS), algorithms implemented in the OpenCV library and algorithms available in Middlebury Stereo Vision Page. Experiments were performed on input data sets which contained images of strawberry plants, cherry trees and redcurrant plants. The bad matching pixels (BMP) metric was used for measuring the error rate in disparity maps used in the distance estimation. The results of experiments showed that, on average, the EEMM merging method used with EBMCS consisting of five cameras reduces the error rate of the distance estimation by 26.55% in comparison to results obtained from stereoscopy based on a single stereo camera. The best results were acquired by using with five cameras a stereo matching algorithm based on a graph cut.

Exploring contexts of use of cultural objects in virtual museums

This paper presents a system which facilitates discovering knowledge about cultural objects. The system is based on semantic modeling of a virtual museum which consists of cultural objects placed in a virtual 3D space. The article describes an extension to the concept of cultural objects which includes information on the use of these objects. This extension enables to place objects in an appropriate context in a virtual museum. Therefore it is possible to restore the true appearance of antique interiors and other spaces. On the basis of the information sorted in the system new facts concerning the objects can be discovered. This system is an IT tool designed to facilitate humanistic research.

Using Open Source Libraries for Obtaining 3D Scans of Building Interiors

This paper describes methods for making 3D scans of building interiors. The main application of these methods is the development of first-person view (FPV) perspective games or creating virtual museums. 3D scans can be made with the use of different equipment such as Light Detection and Ranging (LIDAR), time of flight (TOF) cameras and structural light 3D scanners. However, the paper focuses on using stereo cameras for obtaining 3D scans, because of its low cost. It is a significant factor for small and medium size game development studios. The paper considers both the method based on photogrammetry and stereophotogrammetry. In photogrammetry the Structure from Motion technology is used for making 3D scan on the basis of images of an object taken from different locations. Photogrammetry uses stereo vision algorithms for acquiring depth maps and point clouds representing distances between a stereo camera and objects. The paper analyses implementations of these technologies available in programing libraries OpenCV, openMVG and openMVS. The paper shows that the algorithm for Structure from Motion provided in openMVG can be successfully applied for obtaining point clouds from pair of images despite this algorithm is intended for use with a greater number of input images.

Influence of Aggregating Window Size on Disparity Maps Obtained from Equal Baseline Multiple Camera Set (EBMCS)

This paper is concerned with obtaining disparity maps on the basis of images from Equal Baseline Multiple Camera Set (EBMCS). EBMCS consists of a central camera and side cameras. Algorithms for obtaining disparity maps with the use of EBMCS take advantage of aggregating windows similarly to stereo matching algorithms for a stereo camera, a camera matrix or a camera array. The paper analyzes the influence of aggregating window size on the quality of disparity maps. Experiments presented in this paper include Sum of Sum of Squared Differences (SSSD) and Sum of Sum of Absolute Differences (SSAD) matching cost functions. Results show that for EBMCS with five cameras the highest quality of disparity maps is obtained when the size of the aggregating window is on average over 55 % smaller than the size of the most effective window for a pair of cameras.

Information Retrieval with the Use of Music Clustering by Directions Algorithm

This paper introduces the Music Clustering by Directions (MCBD) algorithm. The algorithm is designed to support users of query by humming systems in formulating queries. This kind of systems makes it possible to retrieve songs and tunes on the basis of a melody recorded by the user. The Music Clustering by Directions algorithm is a kind of an interactive query expansion method. On the basis of query, the algorithm provides suggestions that may be used to enhance the query. The MCBD algorithm is bases on the Clustering by Directions (CBD) that was designed to support users of web search engines in information retrieval. The MCBD algorithm is used for retrieval of music documents similarly as the CBD algorithm is used with text documents. This paper also introduces a new kind of interface designed to perform interactive query expansion in query by humming systems. The interface is a kind of tag cloud applied to music information retrieval systems.