Local Features in Image and Video Processing - Object Class Matching and Video Shot Detection

The usage of digital content, such as video clips and images, has increased dramatically during the last decade. Local image features have been applied increasingly in various image and video retrieval applications. This thesis evaluates local features and applies them to image and video processing tasks. The results of the study show that 1) the performance of different local feature detector and descriptor methods vary significantly in object class matching, 2) local features can be applied in image alignment with superior results against the state-of-the-art, 3) the local feature based shot boundary detection method produces promising results, and 4) the local feature based hierarchical video summarization method shows promising new new research direction. In conclusion, this thesis presents the local features as a powerful tool in many applications and the imminent future work should concentrate on improving the quality of the local features.

Making SHERPA/FACT Local

Presentation at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

From Local Practice to Linked Open Data: Rethinking Metadata in Hydra

Presentation at Open Repositories 2014, Helsinki, Finland, June 9-13, 2014

Local Binary Patterns in Focal-Plane Processing. Analysis and Applications

Feature extraction is the part of pattern recognition, where the sensor data is transformed into a more suitable form for the machine to interpret. The purpose of this step is also to reduce the amount of information passed to the next stages of the system, and to preserve the essential information in the view of discriminating the data into different classes. For instance, in the case of image analysis the actual image intensities are vulnerable to various environmental effects, such as lighting changes and the feature extraction can be used as means for detecting features, which are invariant to certain types of illumination changes. Finally, classification tries to make decisions based on the previously transformed data. The main focus of this thesis is on developing new methods for the embedded feature extraction based on local non-parametric image descriptors. Also, feature analysis is carried out for the selected image features. Low-level Local Binary Pattern (LBP) based features are in a main role in the analysis. In the embedded domain, the pattern recognition system must usually meet strict performance constraints, such as high speed, compact size and low power consumption. The characteristics of the final system can be seen as a trade-off between these metrics, which is largely affected by the decisions made during the implementation phase. The implementation alternatives of the LBP based feature extraction are explored in the embedded domain in the context of focal-plane vision processors. In particular, the thesis demonstrates the LBP extraction with MIPA4k massively parallel focal-plane processor IC. Also higher level processing is incorporated to this framework, by means of a framework for implementing a single chip face recognition system. Furthermore, a new method for determining optical flow based on LBPs, designed in particular to the embedded domain is presented. Inspired by some of the principles observed through the feature analysis of the Local Binary Patterns, an extension to the well known non-parametric rank transform is proposed, and its performance is evaluated in face recognition experiments with a standard dataset. Finally, an a priori model where the LBPs are seen as combinations of n-tuples is also presented

Plant-Herbivore Interaction in a Fragmented Landscape: Local Adaptation and Inbreeding

Reciprocal selection between interacting species is a major driver of biodiversity at both the genetic and the species level. This reciprocal selection, or coevolution, has led to the diversification of two highly diverse and abundant groups of organisms, flowering plants and their insect herbivores. In heterogeneous environments, the outcome of coevolved species interactions is influenced by the surrounding community and/or the abiotic environment. The process of adaptation allows species to adapt to their local conditions and to local populations of interacting species. However, adaptation can be disrupted or slowed down by an absence of genetic variation or by increased inbreeding, together with the following inbreeding depression, both of which are common in small and isolated populations that occur in fragmented environments. I studied the interaction between a long-lived plant Vincetoxicum hirundinaria and its specialist herbivore Abrostola asclepiadis in the southwestern archipelago of Finland. I focused on mutual local adaptation of plants and herbivores, which is a demonstration of reciprocal selection between species, a prerequisite for coevolution. I then proceeded to investigate the processes that could potentially hamper local adaptation, or species interaction in general, when the population size is small. I did this by examining how inbreeding of both plants and herbivores affects traits that are important for interaction, as well as among-population variation in the effects of inbreeding. In addition to bi-parental inbreeding, in plants inbreeding can arise from self-fertilization which has important implications for mating system evolution. I found that local adaptation of the plant to its herbivores varied among populations. Local adaptation of the herbivore varied among populations and years, being weaker in populations that were most connected. Inbreeding caused inbreeding depression in both plants and herbivores. In some populations inbreeding depression in herbivore biomass was stronger in herbivores feeding on inbred plants than in those feeding on outbred ones. For plants it was the other way around: inbreeding depression in anti-herbivore resistance decreased when the herbivores were inbred. Underlying some of the among-population variation in the effects of inbreeding is variation in plant phenolic compounds. However, variation in the modification of phenolic compounds in the digestive tract of the herbivore did not explain the inbreeding depression in herbivore biomass. Finally, adult herbivores had a preference for outbred host plants for egg deposition, and herbivore inbreeding had a positive effect on egg survival when the eggs were exposed to predators and parasitoids. These results suggest that plants and herbivores indeed exert reciprocal selection, as demonstrated by the significant local adaptation of V. hirundinaria and A. asclepiadis to one another. The most significant cause of disruption of the local adaptation of herbivore populations was population connectivity, and thus probably gene flow. In plants local adaptation tended to increase with increasing genetic variation. Whether or not inbreeding depression occurred varied according to the life-history stage of the herbivore and/or the plant trait in question. In addition, the effects of inbreeding strongly depended on the population. Taken together, inbreeding modified plant-herbivore interactions at several different levels, and can thus affect the strength of reciprocal selection between species. Thus inbreeding has the potential to affect the outcome of coevolution.

TAKING ADVANTAGE OF THE LOCAL ADVANTAGE? - Investment value of the analyst local advantage in the Finnish stock market in 2006–2014

An investor can either conduct independent analysis or rely on the analyses of others. Stock analysts provide markets with expectations regarding particular securities. However, analysts have different capabilities and resources, of which investors are seldom cognizant. The local advantage refers to the advantage stemming from cultural or geographical proximity to securities analyzed. The research has confirmed that local agents are generally more accurate or produce excess returns. This thesis tests the investment value of the local advantage regarding Finnish stocks via target price data. The empirical section investigates the local advantage from several aspects. It is discovered that local analysts were more focused on certain sectors generally located close to consumer markets. Market reactions to target price revisions were generally insignificant with the exception to local positive target prices. Both local and foreign target prices were overly optimistic and exhibited signs of herding. Neither group could be identified as a leader or follower of new information. Additionally, foreign price change expectations were more in line with the quantitative models and ideas such as beta or return mean reversion. The locals were more accurate than foreign analysts in 5 out of 9 sectors and vice versa in one. These sectors were somewhat in line with coverage decisions and buttressed the idea of local advantage stemming from proximity to markets, not to headquarters. The accuracy advantage was dependent on sample years and on the measure used. Local analysts ranked magnitudes of price changes more accurately in optimistic and foreign analysts in pessimistic target prices. Directional accuracy of both groups was under 50% and target prices held no linear predictive power. Investment value of target prices were tested by forming mean-variance efficient portfolios. Parallel to differing accuracies in the levels of expectations foreign portfolio performed better when short sales were allowed and local better when disallowed. Both local and non-local portfolios performed worse than a passive index fund, albeit not statistically significantly. This was in line with previously reported low overall accuracy and different accuracy profiles. Refraining from estimating individual stock returns altogether produced statistically significantly higher Sharpe ratios compared to local or foreign portfolios. The proposed method of testing the investment value of target prices of different groups suffered from some inconsistencies. Nevertheless, these results are of interest to investors seeking the advice of security analysts.