**3.7 Handwriting recognition**

Handwritings constitute another important part of the collected data. The objective of this work package is to develop computer vision and pattern recognition methods to identify and recognize the large volumes of unconstrained handwritten text to assist the experts.

Methods will be developed not only for writer identification which is a very important task in forensic applications but also for automatic recognition of the text found in the environment such as notes and letters written by the subject or the scene text in the photos taken, such as the shop labels and billboards.

Image enhancement techniques will be carefully applied to reduce the noise without deforming the original data. Alternative to character-based systems, wordbased systems will be developed to read the text 'in the wild' using methods inspired from object detection and recognition literature. Generic image descriptors such as salient points, gradient histograms and line pairs will be considered for describing words. Efficient and effective similarity measures will be developed to match handwritten text in large volumes in a fast manner without losing any important data.

• Representation of words

Words in handwritten text will be described with advanced visual features used in generic object recognition. Novel descriptors based on contours, salient points and shapes will be generated.

• Similarity-based word matching

Development of 'query by example,'subword, keyword and phrase-based retrieval schemes using conventional and structural similarity-based methods capable of part and whole similarity matching.

The task management mechanism that can handle the interaction and task division between the model-based module and the similarity-based module will be

The developed novelty detection unit will be integrated into the CBR unit and

To provide a clear picture of the current legal framework regulating the process of gathering, processing, analyzing and integrating multimedia data for security and judicial purposes at the European (EU Directives and Regulations) and national level. To specialize the investigation on legal issues related to the use of sensible data for forensic purposes by the analysis of case studies and EU (ECJ) and national

To provide a framework of standards, quality indicators and approaches for the

To support partners in sorting out potential ethical questions, by supporting the consulting process of the advisory body in solving questions specific to ethics, issues concerning sensible data processing and, generally, rules limiting the use of per-

• This task aims at providing a deep survey of the legal sources at the national and European level. The survey will outline the state of the art in EU law, namely, the first Data Protection Directive (Directive 95/46/EC), and will investigate the important role played by the European Court of Justice (ECJ) in its interpretation. The emphasis is on the principle of proportionality—the key concept in the ECJ's judgments—that requires that every specific instance of processing of personal data to be necessary for its concrete purpose. The socalled proportionality test has three components, which involve an assessment

of a measure's suitability, necessity and proportionality strictu sensu. References to the test could be useful for evaluating the legal effects of the implementation of tools developed by the project. National legislation of the eight countries participating in the Consortium (FR, DE, IT, BG, GR, SP, TK and NL) will be collected, analyzed and compared, to provide a complete picture of the level of harmonization among European countries. The impact of

• The second step makes specific reference to the new European rules under

• **Evaluation of digital evidence:** this task aims at explaining the concept of legal validity in the light of digital evidence and at pointing out the criteria

discussion (proposal for a Directive-COM-2012-2110 and Sec-2012-2072 final) and its accompanying documents (Impact Assessment by Commission staff working paper to the Regulation of the European Parliament and of the Council on the protection of individuals with regard to the processing of personal data and on the free movement of such data and to the Directive 72). The implementation process of the new rules will be monitored, and their impact (especially the interaction between regulation and Directive), will be analyzed. Actual changes of the regulative framework in case of the enactment of the new legislation during the

'proportionality test' in national judiciary will be tested.

project life will be explained to partners by training activities.

preservation and validity assessment of digital evidence for forensic purposes.

sonal data extracted by massive data processing techniques:

developed, implemented and evaluated.

*DOI: http://dx.doi.org/10.5772/intechopen.92548*

*Novel Methods for Forensic Multimedia Data Analysis: Part II*

• Integration into the CBR unit

tested.

**3.9 Legal aspects**

judgments.

**149**

• Writer identification

Documents will be preprocessed for enhancement and noise removal. Exemplarbased word and subword matching will be used to identify and verify the writers. Experts will be provided with a set of results and will be asked for feedback to be used in an active learning scheme.

• Automatic text recognition

Development of both character and word-based recognition schemes. Manual effort for labeling data during training will be reduced by learning the relationships between available handwritten and printed text pairs. Scene text will also be recognized.
