Lochandaka Ranathunga

Performance evaluation of the combination of Compacted Dither Pattern Codes with Bhattacharyya classifier in video visual concept depiction

High dimensionality and multi-feature combinations can have negative effect on visual concept classification. In our research, we formulated a new compacted form which is Compacted Dither Pattern Code (CDPC) as a chromatic syntactic feature for visual feature extraction. The effectiveness of CDPC with Bhattacharyya classifier for irregular shapes based visual concepts depiction is reported in this paper. The proposed technique can reduce feature space and computational complexity while maintaining visual data mining and retrieval accuracy in high standard. Our system was empowered with Bhattacharyya classifier which has improved efficiency by considering one numeric value which is the Bhattacharyya coefficient. Experiments were conducted on various combinations and compared with different visual descriptors and classifiers. The first experiment illustrates the comparison of the CDPC based results with well known feature space reduction classes. The second and third experiments demonstrate the effectiveness of our approach with multiple perspectives of performance measures including various concepts.

Analysis of Video Content in Multi Codec Formats with Compacted Dither Coding

The digital video content analysis and retrieval is a hot research topic in this era as videos act as an information source. However the content analysis and retrieval is not as simple as meaningful retrieval of information in a normal data system. The visual information of video data lies in high chromatic depth and density. At the same time these data uses different codecs for storage and interpretation. The differences of video codec coding system properties cause for chromatic data differences of video files. This paper elicits the difficulties of dealing with multi codec level video visual classification and it proposes an approach to handle visual features and higher level color attributes of video data even though it is in multi-codecs nature. This approach uses novel compacted dither pattern code extractor for visual classification. This extractor reduces the complexity of colour visual feature extraction and classification. Further, this approach has been tested with various lossy and loss-less video formats to generate a probabilistic coding mechanism. The results of this approach show that it can be further enhanced to calibrate a video visual classification knowledge base.

Invariant properties of a locally salient dither pattern with a spatial-chromatic histogram

Compacted Dither Pattern Code (CDPC) is a recently found feature which is successful in irregular shapes based visual depiction. Locally salient dither pattern feature is an attempt to expand the capability of CDPC for both regular and irregular shape based visual depiction. This paper presents an analysis of rotational and scale invariance property of locally salient dither pattern feature with a two dimensional spatial-chromatic histogram, which expands the applicability of the visual feature. Experiments were conducted to exhibit rotational and scale invariance of the feature. These experiments were conducted by combining linear Support Vector Machine (SVM) classifier to the new feature. The experimental results revealed that the locally salient dither pattern feature with the spatial-chromatic histogram is rotationally and scale invariant.

A segmentation method for extraction of main arteries from Coronary Cine-Angiograms

Coronary Cine-Angiogram (CCA) based subjective assessment of vascular malfunction is a preliminary diagnostic method in Cardiac clinical procedures. Even though there are many other medical image modalities available, improving the CCA method to objectively detect and assess the stenosis is a cost effective approach in Cardiac clinical procedures. Segmentation of Coronary Arteries (CA) is a basic and challenging area in such an endeavour. Hence, in this study we proposed a segmentation method to extract the major areas of CA based on Frangis vessel enhancement filter and region growing segmentation method called flood fill. Experimental results of our proposed segmentation method have clearly proven its ability to extract the main CA almost completely. Moreover this proposed segmentation method possesses 93.73% average segmentation accuracy. Further, it detects the vessel path lines of the segmented frames using a thinning algorithm. The results obtained from this proposed segmentation method can be further enhanced to determine the functional severity of the CA and this study lays a foundation to improve the Coronary Angiogram image modality to do objective diagnosis of stenosis in future.

Reduction of Motion Disturbances in Coronary Cineangiograms through Template Matching

The coronary cineangiogram (CCA) is an invasive medical image modality which is used to determine the stenosis in Coronary Arteries. The motion artifacts occurring due to the heart pulse makes great disturbance to visualize a stable contrast agent flow within the vessel structure of CCA and it negatively affects to quantify the stenosis based on the functional significance within the arterial flow. This paper describes an application of template matching to reconstruct the CCA by reducing the global motion artifacts. The Normalized Correlation Coefficient (NCC) method has been used for the template matching because, it reports the lowest false matching occurrences. Further, the NCC technique has 99.5% accuracy and demonstrates its ability to maintain the visual correlation of the internal blood flow among the frames. Producing Motion eliminated CCA to maintain the visual correlation of arterial flow is an improvement of angiography technique which can be useful for advanced processing.

Conventional video shot segmentation to semantic shot segmentation

Video shot segmentation is a preliminary process used in video content analysis which requires for content description. Conventional shot segmentation techniques nourished with statistical approaches depend on chromatic distributions of video frames. However, these conventional approaches lagging behind in performing semantic shot segmentation where more intuitive and perceptive for human understanding and search. This paper presents a novel mechanism which capable of perform semantic shot segmentation without additional computational and description cost for semantic video content depiction process. This mechanism bypasses the traditional video shot segmentation which is normally performs at the beginning of video content depiction and instead it perform meaningful shot boundary identification at the end based on the results of video visual concept classification. This is a positive starting point to perform structuring of video content using semantic units of shots rather than abrupt discontinued shots. Further, this paper introduces a search mechanism empowered by semantic shot segmentation.

Reduction of syntactic video data clustering complexity in processing with compacted dither coding

The growing consumption of the digital video information is significant in this era. The digital video analysis and retrieval is not as simple as analysis and retrieval of information in normal data system. The visual information of video data lies in very complex nature with its high chromatic depth and density. The extraction of visual features from noisy and complex video data has a hierarchy of different sub systems from video file to chromatic attributes. This paper introduces a novel approach to reduce the video visual feature analyzing complexity and the higher level colour complexity of video data. It comes with simple vector quantization mechanism, high rate performance approach for classification of digital video visual features. Further this approach has tested with various video formats to generate probabilistic coding mechanism. The results of this approach show that it can be further enhanced with video graphical knowledge to guide the visual feature clustering with trained knowledge base.

A feature clustering approach based on Histogram of Oriented Optical Flow and superpixels

Visual feature clustering is one of the cost-effective approaches to segment objects in videos. However, the assumptions made for developing the existing algorithms prevent them from being used in situations like segmenting an unknown number of static and moving objects under heavy camera movements. This paper addresses the problem by introducing a clustering approach based on superpixels and short-term Histogram of Oriented Optical Flow (HOOF). Salient Dither Pattern Feature (SDPF) is used as the visual feature to track the flow and Simple Linear Iterative Clustering (SLIC) is used for obtaining the superpixels. This new clustering approach is based on merging superpixels by comparing short term local HOOF and a color cue to form high-level semantic segments. The new approach was compared with one of the latest feature clustering approaches based on K-Means in eight-dimensional space and the results revealed that the new approach is better by means of consistency, completeness, and spatial accuracy. Further, the new approach completely solved the problem of not knowing the number of objects in a scene.

Can pedagogical evaluation process of web based E-Learning be automated?

E-Learning aims to improve the quality of learning through the use of modern information and communication technologies. Designing E-Learning course materials is the most critical task because its entire educational system depends on the way content is presented. For that reason the necessity of evaluating the web based E-Learning content is arising. The evaluation of E-Learning content is beneficial for learner as well as instructional designer because it facilitates learner to choose the best learning material, and helps designers to improve the quality of their learning materials. This paper describes the importance of automating the pedagogical evaluating process of web based E-Learning contents and the theoretical framework used in the evaluation process. The aim of this research is to improve the quality of the content in e-Learning websites using the feedback given by the automated pedagogical evaluation process.

Compacted Dither Pattern Codes versus Principal Component Analysis in video visual depiction

Requirement of reduction of feature space of visual descriptors gets attention due to negative effects of high dimensional feature space. This paper reports the performance of Compacted Dither Pattern Code (CDPC) over Principal Component Analysis (PCA) based compact colour descriptor. There are several competitive advantages of CDPC in feature extraction and classification stages when compared to PCA feature vectors. The embedded texel properties, spatial colour arrangements, high compactness, and robust feature representation of CDPC have proven its performances in our experimental study. Visual description experiments were conducted for ten irregular shapes based visual concepts in videos with three setups namely CDPC with Bhattacharyya classifier, PCA with Support Vector Machine (SVM) classifier and PCA with Bhattacharyya classifier. The experimental results were presented based on three common performance measures. The results depict that CDPC with Bhattacharyya classifier provides a good generalized performance for irregular shape based visual description as compared to the other two experimental setups.