Damir Filko

Emotion Recognition System by a Neural Network Based Facial Expression Analysis

Human-computer interfaces are getting more complex every day with the purpose of easing the use of computers and enhancing the overall user experience. Since research has shown that a majority of human interaction comes from non-verbal communication, user emotion detection is one of the directions that can be taken to enhance the overall user experience. This paper proposes a system for human emotion recognition by analyzing key facial regions using principal component analysis and neural networks. The proposed system has been trained and tested on the FEEDTUM database where it achieved a relatively high average score of correct recognition and therefore showed promise for future development.

Place recognition based on matching of planar surfaces and line segments

This paper considers the potential of using three-dimensional (3D) planar surfaces and line segments detected in depth images for place recognition. A place recognition method is presented that is based on matching sets of surface and line features extracted from depth images provided by a 3D camera to features of the same type contained in a previously created environment model. The considered environment model consists of a set of local models representing particular locations in the modeled environment. Each local model consists of planar surface segments and line segments representing the edges of objects in the environment. The presented method is designed for indoor and urban environments. A computationally efficient pose hypothesis generation approach is proposed that ranks the features according to their potential contribution to the pose information, thereby reducing the time needed for obtaining accurate pose estimation. Furthermore, a robust probabilistic method for selecting the best pose hypothesis is proposed that allows matching of partially overlapping point clouds with gross outliers. The proposed approach is experimentally tested on a benchmark dataset containing depth images acquired in the indoor environment with changes in lighting conditions and the presence of moving objects. A comparison of the proposed method to FAB-MAP and DLoopDetector is reported.

Fast Pose Tracking Based on Ranked 3D Planar Patch Correspondences

A fast robot pose tracking algorithm based on planar segments extracted from range images is presented. A range image obtained from a 3D sensor is transformed to a 2.5D triangle mesh from which planar segments are extracted. Using information provided by each planar segment based on its size and orientation, a directed search hypothesis generation algorithm using a tree structure is presented. The presented approach is experimentally evaluated using 3D data obtained by a Kinect sensor mounted on a mobile robot. Results indicate that the proposed method is much faster than similar previously proposed methods.

Wound detection and reconstruction using RGB-D camera

The advent of inexpensive RGB-D sensors pioneered by the original Kinect sensor, has paved the way for a lot of innovations in computer and robot vision applications. In this article, we propose a system which uses the new Kinect 2 sensor in a medical application for the purpose of detection and 3D reconstruction of chronic wounds. Wound detection is based on a per block classification of wound tissue using color histograms and the nearest neighbor approach. The 3D reconstruction is similar to KinectFusion where ICP is used for determining the rigid body transformation, color enhanced TSDF is applied for scene fusion, while the marching cubes algorithm is used for creating a surface mesh. The entire system is implemented in CUDA which enables real-time operation. The end result of the developed system is a precise 3D colored model which can be used for determining a correct therapy and treatment of chronic wounds.

Evaluation of color and texture descriptors for matching of planar surfaces in global localization scheme

This paper presents a systematic study about the applicability of color/texture descriptors in a global localization system based on planar surface segments. Two comprehensive experiments regarding matching of planar surface segments and robot pose hypothesis evaluation were conducted. The experiments show that using color/texture descriptors to prune potential surface pairs in the initial correspondence phase and to provide additional information in the hypothesis evaluation phase of a feature-based localization scheme can result in considerable speedup of the localization process and help distinguish between geometrically similar places. An experimental benchmark which enables researchers to evaluate the performance of color and texture descriptors in the context of mobile robot localization based on planar surface segments is presented. Indoor global localization system based on planar segments with visual descriptors.Applicability of 6 color and 3 texture descriptors is systematically analyzed.Performance increase in initial correspondence and pose hypothesis evaluation phases.Evaluation benchmark for visual descriptors in global localization is proposed.

Detection, Reconstruction and Segmentation of Chronic Wounds Using Kinect v2 Sensor

The advent of inexpensive RGB-D sensors pioneered by the original Kinect sensor, has paved the way for a lot of innovations in computer and robot vision applications. In this article, we propose a system which uses the new Kinect v2 sensor in a medical application for the purpose of detection, 3D reconstruction and segmentation of chronic wounds. Wound detection is based on a per block classification of wound tissue using colour histograms and nearest neighbour approach. The 3D reconstruction is similar to KinectFusion where ICP is used for determining rigid body transformation. Colour enhanced TSDF is applied for scene fusion, while the Marching cubes algorithm is used for creating the surface mesh. The wound contour is extracted by a segmentation procedure which is driven by geometrical and visual properties of the surface. Apart from the segmentation procedure, the entire system is implemented in CUDA which enables real-time operation. The end result of the developed system is a precise 3D coloured model of the segmented wound, and its measurable properties including perimeter, area and volume, which can be used for determining a correct therapy and treatment of chronic wounds. All experiments were conducted on a medical wound care model.

Wound measurement by RGB-D camera

The robot and computer vision community has seen a lot of novelties developed in the past few years as a result of the appearance of cheap RGB-D sensors spearheaded by the Kinect sensor. In this paper, the feasibility of using an RGB-D camera in detecting, segmenting, reconstructing and measuring chronic wounds in 3D is explored. The wound is detected by implementing nearest-neighbor approach on color histograms generated from the image. The proposed wound segmentation procedure extracts the wound contour using visual and geometrical information of the surface. A procedure comparable to KinectFusion is used for the 3D reconstruction of the wound. In order to achieve real-time performance, the whole system is realized in CUDA. The resulting system provides an accurate colored 3D model of the segmented wound and enables the user to determine the volume, area and perimeter of the wound, thereby aiding in the selection of a suitable therapy. The developed system is experimentally evaluated using the Saymour II wound care model by VATA Inc.

Low cost robot arm with visual guided positioning

Low cost robotic solutions are of great importance for improvement and development of robotics. In this paper, two visually guided low cost robot arms are proposed. The proposed system performs automatic hand-eye calibration and, after the calibration, positions its end effector above the object of interest using visual servoing based on off the shelf marker tracker. The presented experiments demonstrate positioning accuracy of the proposed setup.

Segmentation of depth images into objects based on local and global convexity

An approach for object detection in depth images based on local and global convexity is presented. The approach consists of three steps: image segmentation into planar patches, greedy planar patch aggregation based on local convexity and segment grouping based on global convexity. The proposed approach improves upon existing similar methods, which use convexity as a cue for object detection, by detecting convex objects represented by multiple spatially separated image regions as well as hollow convex objects. The presented method is experimentally evaluated using a publicly available benchmark dataset and compared to two state-of-the art approaches. The experimental analysis demonstrates improvement achieved by high-level segment grouping based on global convexity.