Janez Zaletelj

Viewer authentication for personalized iTV services

The proliferation of interactive digital TV services will depend on their complexity and ease of use. The investigation of simple, intuitive human-machine interfaces is therefore crucial for their success. Personalized iTV services require consumer authentication and consumer input, which is traditionally done by using the remote control. In this paper, we investigate alternative ways to collect consumer input based on computer vision methods. The goal of the proposed system is to enable consumer authentication and automatic collection of consumer feedback on the TV program. The image processing system on the STB continually monitors the TVviewer(s). The face detection is employed to detect and track the number of people watching the TV program. Face recognition is employed to identify specific TV viewers, which is necessary in order to enable personalized user interface and personalized services. Eye detection and gaze recognition is used to continually track the interest level of the specific viewer towards the TV content, which is important feedback information.

Vision-based human-computer interface using hand gestures

The proliferation of interactive digital TV applications will depend on their complexity and ease of use. The investigation of simple, intuitive human-machine interfaces is therefore crucial for their success. While the operation of most home devices is today based on using remote control, in this paper we investigate an alternative solution using a visual input of the userss gestures. The proposed system recognises point and click gestures of the users hand, which is enough to control simple iconic user interfaces of the interactive TV applications. Position of the users hand is tracked using Kalman filtering, and shape analysis is performed to detect the pose of the forefinger. A finger bending gesture is recognized and transformed into a click action on the screen. Simulations show that the system performs well in uncontrolled home environment under different lighting conditions and cluttered backgrounds and thus could be used in future smart home devices.

Registration of satellite images based on road network map

This paper presents a novel approach to fully automatic satellite image registration to a local earth coordinate system based on the digital map of the road network. Automatic satellite image co-registration methods typically establish correspondences between points on the target and registered image using local image features such as SIFT, however they are not effective for multi-temporal and multi-sensor registration. As the road network is a prominent feature on the satellite images and is visible in most circumstances, we propose to map it directly to the digital road network topographic map, which is readily available in GIS databases. Our approach starts by detection of road segments in a satellite image, producing a road mask image. Small road mask patches are matched against rasterized road network map to find a set of probable patch locations. Following optimization of translation parameters, each satellite image tile is coarsely located on a road map. In a second fine optimization step, smaller road patches locations are further optimized. The algorithm produces a set of tie points, which include coordinates of satellite image pixels within local metric coordinate system. Experiments were performed on a set of 5 RapidEye images of Slovenia from different periods of the year. The results show that the average error of the points is below 1 pixel for all images, and that the reliability in terms of outlier pixels is very high.

A novel system for interactive live TV

For the production of interactive live TV formats, new content and new productions workflows are necessary. We explain how future content of a parallel multi-stream production of live events may be created from a design and a technical perspective. It is argued, that the content should be arranged by dramaturgical principles taking into account the meaning of the base material. To support the production process a new approach for content recommendation is described which uses semantic annotation from audio-visual material and user feedback.

Reliable Subpixel Ground Control Point Estimation Algorithm Using Vector Roads

This letter proposes a novel algorithm for reliable and accurate ground control point (GCP) estimation, which uses vector road network data as a reference. Roads are detected in satellite images, and road patches are used as a template, which is matched against the vector road data by using a nonlinear criterion based on the distance transform. The proposed approach is compared with the standard correlation-based template matching in terms of the reliability and accuracy of localization on a test set of five RapidEye images with 380 reference control points. The experiments demonstrate superior reliability with respect to the size of the search window and the variable image acquisition conditions. While the subpixel accuracy is achieved by the cross-correlation method over the search area up to 20% of the template size, the proposed method achieves comparable accuracy over a search area that is ten times larger than the template image.

Optimization and Tracking of Polygon Vertices for Shape Coding

The efficiency of shape coding is an important problem in low-bitrate object-based video compression. Lossy contour coding methods typically include contour approximation by polygons or splines, spatial and/or temporal prediction of vertices, and entropy coding the of prediction error. In conventional contour coding schemes, however, the coding gain in the interframe mode is typically small. This indicates that the temporal redundancy is not successfully removed. The paper addresses the issue of temporal shape decorrelation by proposing the Kalman filtering-based approach to vertex tracking and prediction. A temporal vertex association procedure is proposed effectively minimizing bit rate in each frame. The prediction error is coded using adaptive arithmetic encoding. Vertex optimization is employed to reduce the shape reconstruction error.

Video object segmentation based on edge tracking

This paper addresses the problem of semi-automatic video object segmentation by tracking image edges, which compose the object boundary. Given an outline of the object in the first frame of the sequence, we analyze its curvature function and determine an optimal B-spline shape model. Affine motion model parameters are estimated and used for propagating the shape model to the next frame, where it serves as an initialization of curve evolution. An evolution criterion integrates edge and local intensity information by using intensity and color models of boundary neighborhood regions, which allows successful tracking of object edges in cases of object deformations and minor occlusion.

Curvature analysis approach to shape coding using B-splines

This work presents an algorithm for efficient shape coding using cubic B-splines. In the framework of object-based layered coding of image sequences, shape information is essential for content-based access to video objects, and its efficient encoding needs to be investigated. We present a rate and distortion controlled algorithm for vide object shape approximation by variable number of cubic B-spline segments and motion compensated inter-frame coding of B-spline control points. Rate-distortion efficiency of the proposed algorithm is compared to MPEG-4 context arithmetic encoding and two stage motion compensated chain coding.

User-centred app design for speech sound disorders interventions with tablet computers

Traditional speech and language pathology practice (SLPP) faces challenges delivering effective and timely therapy due to long waiting lists, the need for regular practice outside the clinic and a lack of children’s motivation to engage in persistent practice. Technology has untapped potential to address these issues and improve SLPP. This paper describes the design of a tablet app for delivering technology-enhanced therapy for children with speech sound disorders and investigates the impact of the use of apps on SLPP. The initial design was informed by a nation-wide survey of speech-language pathologists (SLPs). The quantitative analysis disclosed that even though SLPs positively perceive mobile technology, they do not currently fully exploit it in their practice due to a lack of apps in their native language and the limited usefulness of apps in foreign languages. Using a user-centred design process, a multidisciplinary team created three prototypes and a final version of an app that has been tested in real therapeutic sessions during everyday practice and informed by feedback from SLPs and children. The observation analysis is presented based on an adaptation of Koole’s FRAME model. The qualitative findings indicate that SLPs identify mobile apps as enabling greater mobility, allowing new therapeutic approaches, creating possibilities for practice outside the therapeutic setting and increasing children’s motivation, supporting greater persistence to practise in the context of the therapy.

Video-based learners’ observed attention estimates for lecture learning gain evaluation

A significant problem in the field of higher education is maintaining learners’ attention during lectures, which is known to significantly affect their learning outcomes. Attention management is commonly associated with the individual ability of a lecturer to track and respond to the common behaviour of an auditorium; it lacks a detailed estimation of the intra-variability of individual learners’ attention during the course of the lecture. This paper suggests an objective and non-intrusive evaluation of learners’ attention against learning outcomes by introducing an observed attention estimate (OAE). The procedure uses human annotations based on visual cues with a supporting video recording/playback system and a web-based annotation system. This proposed procedure enables us to estimate the attention level of individual learners as observed by human annotators for given time intervals associated with specific concepts covered by the lecture. As part of the procedure, we use an inventory-based lecture gain evaluation and representation based on a novel learning gain matrix. This procedure allows for a detailed analysis of the lecture time flow with regard to learners’ attention. We have verified the applicability of the procedure on a small-scale case study.