Hichem Karray

A comprehensive method for Arabic video text detection, localization, extraction and recognition

With the rapid growth of the number of TV channels, the internet and online information services, more and more information becomes available and accessible. The digitization enhances preservation of records and makes the access to documents easier. However, when the quantity of documents become important the digitalization is not enough to ensure an efficient access. Indeed, we need to extract semantic information to help users to find what we need quickly. The text included in video sequences is highly needed for indexing and searching system. However, this text is difficult to detect and recognize because of the variability of its size, low resolution characters and the complexity of the backgrounds. To resolve these shortcomings, we propose a two task system: As a first step, we extract the textual information from video sequences and second, we recognize this text. Our system is tested on a diverse database composed of several Arabic news broadcast. The obtained results are encouraging and prove the qualities of our approach.

A New System for Event Detection from Video Surveillance Sequences

In this paper, we present an overview of a hybrid approach for event detection from video surveillance sequences that has been developed within the REGIMVid project. This system can be used to index and search the video sequence by the visual content. The platform provides moving object segmentation and tracking, High-level feature extraction and video event detection.We describe the architecture of the system as well as providing an overview of the descriptors supported to date. We then demonstrate the usefulness of the toolbox in the context of feature extraction, events learning and detection in large collection of video surveillance dataset.

Indexing Video Summaries for Quick Video Browsing

Multimedia data are used in many fields. The problem is how to manipulate the large quantity of these data. One of the proposed solutions is an intelligent video summarization system. Summarizing a video consists in providing another version that contains pertinent and important items. The most popular type of summary is the pictorial summary. We propose in this chapter to index pictorial summaries in order to accelerate the browsing operation of video archives. The chapter presents a conception of a digital video archive that offers three access levels making easier the search for video sequences. The first access level offers to the user a full access for the whole archive. The second access level allows to the user to browse video archive by consulting video summaries. We contribute by adding a third access level that accelerates the archive browsing by adding an indexing subsystem, which operates on video summaries. We propose to index video summaries to accelerate the research of desired sequences. We treat the case of news broadcast video

NF-SAVO: Neuro-Fuzzy system for Arabic Video OCR

In this paper we propose a robust approach for text extraction and recognition from video clips which is called Neuro-Fuzzy system for Arabic Video OCR. In Arabic video text recognition, a number of noise components provide the text relatively more complicated to separate from the background. Further, the characters can be moving or presented in a diversity of colors, sizes and fonts that are not uniform. Added to this, is the fact that the background is usually moving making text extraction a more intricate process. Video include two kinds of text, scene text and artificial text. Scene text is usually text that becomes part of the scene itself as it is recorded at the time of filming the scene. But artificial text is produced separately and away from the scene and is laid over it at a later stage or during the post processing time. The emergence of artificial text is consequently vigilantly directed. This type of text carries with it important information that helps in video referencing, indexing and retrieval.

Toward an interactive device for quick news story browsing

In this paper, we present a new design for an interactive information service based on on-line recognition of the handwriting and quick news stories browsing. A person communicates with server PC using PDA and Bluetooth headset technology in order to consult same key frame that represent a summaries of video news. The result of the server research will by returned to the PDA.

Regim, research group on intelligent machines, tunisia, at TRECVID 2008, BBC rushes summarization

In this paper, we describe our system used to summarize BBC rushes, the TRECVID database. Our summarization process starts with shot boundary detection. Then we filter obtained shots to retain only useful ones. After that we try to localize from every retained shot the important parts (sub-shots). Finally, we select some of them to formulate the skim. The selection of sub-shots must respond to many criteria as redundancy removing, covering all important events of the original video sequence and not exceeding the upper duration. Genetic algorithms are naturally suited for doing incremental selection. We use it to do the selection of relevant subs-shots. We consider the summarization process as an optimization problem which takes into consideration all evoked criteria. The obtained results are encouraging.

AVOCR: Arabic video OCR

Nowadays the majority of TV channels are performing digitization of their archival materials. The digitization enhances preservation of records and makes the access to documents easier. However, when the quantity of documents become important the digitization is not enough to insure an efficient access. Indeed, we need to extract semantic information to help users to find what we need quickly. The text included in video sequences is an important needful for indexing and searching system. However, this text is difficult to detect and recognize because of the variability of it size, their low resolution characters and the complexity of the backgrounds. To solve these problems, we propose a system performing in two main tasks: first we extract the textual information from the video sequence and second we recognize this text.

Content-Based Arabic Speech Similarity Search and Emotion Detection

System performance of speech emotion recognition is prejudiced by many factors, including the quality of speech samples, features and classifiers. The present study proposed a specific architecture of an automatic recognition system of Arabic speech emotions, recognized as: Anger, Happiness, Neutral, Sadness and Fear. To perform this system we select two approaches classifications. KNN (k-nearest neighbor’s algorithm) and SVMs support vector machines multiclass. Good performances are carried out: The rate of cross-validation is over 92 % and the rate of recognition accuracy is about 93 %. The intelligent system outputs contribute to the addition of phonemes in Arabic speech, but it should be noted that it is related not only to the language but also to culture and environment. Corpus is performed without delimiter syntax and language, because our goal is to extract emotions not speech recognition. Even the silent sounds are processed as dynamic signal. To accomplish the emotions, we had created a REGIM_TES database as support and we developed a similarity module to compare features vectors to settle on emotions. The subjective evaluation of emotions is carried out by performing the subjective listening tests with listeners candid. Twelve listeners (6 women and 6 men) with ages ranging from 18 to 42 years participated in the experiment and each listener was presented with 550 expressions. We conclude that mistakes of the recognition system (objective) are significantly lower than those of human (subjective) evaluators. Finally, the emotion recognition system, allow research to detect a person’s behavior and speech abnormalities.

Ant-based clustering algorithm for magnetic resonance breast image segmentation

This article introduces an improved version of the ant-clustering approach for image segmentation. An application of breast cancer magnetic resonance breas imaging has been chosen and the improved ant-based clustering approach has been applied to see their ability and accuracy to isolate the region of interest in the MRI images. The aim of the proposed ant-based clustering is to identify target objects through an The experimental results obtained, show that the modified ant-based clustering is superior to the classical ant-based clustering and the overall accuracy offered by the improved approach confirm that the effectiveness and performance is 98% in average.

Toward Interactive TV

Televisions and spatial channels have invaded wealthy and poor families. With the growing number of TV channels, the TV viewers become unable to watch what they prefer easily. They will be obliged to zap between thousands of channels permanently to have the right program. With the apparition of DTT (Digital Terrestrial Television), it will be possible for televisions of next generations to receive more than one channel at the same time. Convinced by the importance of the problem, we are working on the concept of the Intelligent Television (AITV). We propose to add to televisions having the ability to receive more than one channel at the same time (this type of television is commercialized and its price will be popular in the next few years) an intelligent layer which can be configured according to the preference of the viewer. If a viewer is watching a program on given a channel and the television detect a program corresponding to his preference, the watched program will be interrupted and it will be switched by the detected program. In that case all viewers will guarantee that no program or subject or scenes will escape them. This may be applied to sports events, to news broadcasts, to documentary … The idea of this work is to make the television personalized, adapted for each user and able to analyze all the components of the video stream (Auditory analysis, image analysis, and natural language analysis) to filter for every user the suitable programs (see Figure 1). • Automatically extracting knowledge from the TV content via combination of text mining, visual semantics analysis, and audio semantics analysis;. • Selectively recording all the TV content which may be preferred and desired by consumers. • Searching, retrieving and managing the recorded TV content via a simple remote control set. • Automatically upgrading itself by deleting the obsolete content and adding new content. (Interactive TV) will also provide new technologies to help the content producers to ensure that the content produced will reach the right audience, and thus the efficiency and effectiveness of their content production will be significantly improved. To experiment our system, we will use the dataset of TRECVID Workshops (NIST).