Jukka Saarinen

Sequence Prediction With Sparse Distributed Hyperdimensional Coding Applied to the Analysis of Mobile Phone Use Patterns

Modeling and prediction of temporal sequences is central to many signal processing and machine learning applications. Prediction based on sequence history is typically performed using parametric models, such as fixed-order Markov chains (n-grams), approximations of high-order Markov processes, such as mixed-order Markov models or mixtures of lagged bigram models, or with other machine learning techniques. This paper presents a method for sequence prediction based on sparse hyperdimensional coding of the sequence structure and describes how higher order temporal structures can be utilized in sparse coding in a balanced manner. The method is purely incremental, allowing real-time online learning and prediction with limited computational resources. Experiments with prediction of mobile phone use patterns, including the prediction of the next launched application, the next GPS location of the user, and the next artist played with the phone media player, reveal that the proposed method is able to capture the relevant variable-order structure from the sequences. In comparison with the n-grams and the mixed-order Markov models, the sparse hyperdimensional predictor clearly outperforms its peers in terms of unweighted average recall and achieves an equal level of weighted average recall as the mixed-order Markov chain but without the batch training of the mixed-order model.

A method of fingerprint image enhancement based on second directional derivatives

We present an approach to fingerprint image enhancement that relies on detecting the fingerprint ridges based on the sign of the second directional derivative of the digital image. A facet model is used in order to approximate the derivatives at each image pixel based on the intensity values of pixels located in a certain neighborhood. The size of this neighborhood determines the scale of the image details that are preserved. We develop a selection criterion for the neighborhood size that aims to preserve minutiae details and remove smaller details from the enhanced image. Experimental results demonstrate the ability of the proposed approach to preserve a large percentage of the genuine minutiae in the enhanced image.

Open Source, Open Innovation and Intellectual Property Rights – A Lightning Talk

Open innovation projects are fast paced aiming at producing a quick proof of concept of an innovative software product. This need for speedy results makes the use of open source components as a basis for the work appealing. Open source brings with it an inherent risk of license conflicts that may become an issue when aiming to develope an innovative demo into an actual product. In this study, the first results of investigating the knowledge the participants of innovation projects have on intellectual property are presented. The effect this may have on the project results is also discussed.

Document image binarization using the camera device in mobile phones

This paper proposes an adaptive binarization method for the document image binarization acquired by an imaging phone. The used algorithm determines the local thresholds with the information from the global trend as well as the local details. As a consequence, the proposed method is good not only for preserving the fine details of the character structure, but also for alleviating noise. The effectiveness of the proposed method is illustrated using the results obtained with a Nokia imaging phone.

Applying Open Source Practices and Principles in Open Innovation: The Case of the Demola Platform

In numerous fields, businesses have to rely on rapid development and release cycles. Variant new ideas and concepts can emerge through open innovation as the participants are not limited to the company scope. This makes open innovation an increasingly appealing option for the industry. One such open innovation platform, Demola, allows university students to work on real life industrial cases of their own interest. We have identified similarities with its way of operation to open source software development and find that it offers a viable motivational, organizational and collaborative solution to open innovation.

A novel quantization scheme for the noise-like component in waveform interpolation speech coding

This paper presents a novel and efficient coding scheme for the noise-like component in waveform interpolation speech coding. The techniques employed in the task are described in detail and the advantages gained by using them are discussed. These techniques are shown to enable perceptually transparent coding of the magnitude spectra of the noise-like component with only ten bits per frame and to offer speech quality improvements at higher bit rates. The most significant features of the proposed scheme are smoothed representation and matrix quantization of the magnitude spectra. The proposed coding techniques can be used for enhancement of speech coders based on the waveform interpolation approach.

Pothole detection and tracking in car video sequence

In this paper, we propose a low complexity method for detection and tracking of potholes in video sequences taken by a camera placed inside a moving car. The region of interest for the detection of the potholes is selected as the image area where the road is observed with the highest resolution. A threshold-based algorithm generates a set of candidate regions. For each region the following features are extracted: its size, the regularity of the intensity surface, contrast with respect to background model, and the regions contour length and shape. The candidate regions are labeled as putative potholes by a decision tree according to these features, eliminating the false positives due to shadows of wayside objects. The putative potholes that are successfully tracked in consecutive frames are finally declared potholes. Experimental results with real video sequences show a good detection precision.

Lossy-to-lossless progressive coding of depth-map images using competing constant and planar models

In this paper we propose an extension of our lossy-to-lossless progressive coding method by placing the planar model in a competition with the piecewise constant model during the region reconstruction stage of the algorithm. A sequence of lossy images is generated using an hierarchical segmentation, of the initial image, based on region merging. The progressive coding method is able to compress this sequence of images by encoding the elements that represent the differences between two consecutive images. The method is splitting some regions from the current image segmentation using an encoded set of contours, and it is defining a set of new regions, which are reconstructed using either the piecewise constant model or the planar model. An efficient solution is proposed for encoding the model parameters in a progressive way. Results show an improvement of 3 - 4 dB compared to the baseline method based only on constant regions, and for a wide range it achieves almost similar results with the non-progressive methods.