Zenilton Kleber G. do Patrocínio

A Simple Hierarchical Clustering Method for Improving Flame Pixel Classification

In this paper, we propose a new approach for color image simplification in order to improve flame pixel classification. The fire detection performance depends critically on the performance of the flame pixel classifier. Color image simplification is the process of simplifying an image in order to decrease the number of colors while preserving, as much as possible, shapes. In this work, a hierarchical clustering method in a given color space is used to map the original colors into a smaller set of representative ones, allowing the use of a simple heuristic rule for classifying the clusters related to candidate flame colors. Using reverse mapping, we identify possible flame colors in the image. Main contributions of our work are the application of a simple hierarchical clustering method to color simplification, that decreases the number of possible flame colors, and a filtering methodology to reduce the influence of outliers. Several color spaces and distance measures were used to evaluate the proposed method. Experimental results demonstrate that color simplification is essential to successfully employ heuristic classification of flame colors.

A Graph-Based Hierarchical Image Segmentation Method Based on a Statistical Merging Predicate

Hierarchical image segmentation provides a set of image segmentations at different detail levels in which coaser details levels can be produced by simple merges of regions from segmentations at finer detail levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy. In addition, for image segmentation, the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph-based image segmentation relying on a statistical region merging. Furthermore, we study how the inclusion of hierarchical property have influenced the computation of quality measures in the original method. Quantitative and qualitative assessments of the method on two image databases show efficiency and ease of use of our method.

Decreasing the Number of Features for Improving Human Action Classification

Action classification in videos has been a very active field of research over the past years. Human action classification is a research field with application to various areas such as video indexing, surveillance, human-computer interfaces, among others. In this paper, we propose a strategy based on decreasing the number of features in order to improve accuracy in the human action classification task. Thus, to classify human action, we firstly compute a video segmentation for simplifying the visual information, in the following, we use a mid-level representation for representing the feature vectors which are finally classified. Experimental results demonstrate that our approach has improved the quality of human action classification in comparison to the baseline while using 60% less features.

Hierarchical Image Segmentation Relying on a Likelihood Ratio Test

Hierarchical image segmentation provides a set of image segmentations at different detail levels in which coarser details levels can be produced by simple merges of regions from segmentations at finer detail levels. However, many image segmentation algorithms relying on similarity measures lead to no hierarchy. One of interesting similarity measures is a likelihood ratio, in which each region is modelled by a Gaussian distribution to approximate the cue distributions. In this work, we propose a hierarchical graph-based image segmentation inspired by this likelihood ratio test. Furthermore, we study how the inclusion of hierarchical property have influenced the computation of quality measures in the original method. Quantitative and qualitative assessments of the method on three well known image databases show efficiency.

Improving the Quality of Color Image Segmentation Using Genetic Algorithm

Color image segmentation is the process of grouping regions according to some criterium. In this work, we cope with this problem using a graph-based approach based on removal of minimum spanning tree edges, however the tuning of parameters is a difficult task. To better identify the set of parameters which optimizes the error producing good segmentations, we propose the use of genetic algorithm in order to establish the best set of parameters. According to test experiments, our proposed method presents better results when compared to other approaches from the literature.

A two-step video subsequence identification based on bipartite graph matching

Subsequence identification consists in identifying real positions of a specific video clip in a video stream together with the operations that may be used to transform the former into a subsequence from the latter. In order to cope with this problem, we propose a two-step method. First, a clip filtering strategy based on the identification of dense segments is used, in order to decrease the number of video clip candidates. Then, for each dense segment, a graph matching approach is applied to identify video subsequences similar to the query video. Our main contribution is the use of a simple and efficient distance to solve subsequence identification problem along with the definition of a hit function that identifies precisely which operations were used in query transformation. Experimental results demonstrate good performance for our method (90% recall with 93% precision).

Human Action Classification Using an Extended BoW Formalism

In human action classification task, a video must be classified into a pre-determined class. To cope with this problem, we propose a mid-level representation which extends the Bag-of-Words formalism in order to better described the low-level features, exploring distance-to-codeword histograms. The main contribution of this article is the assembly of low-level features by a mid-level representation enriched with information about distances between descriptors and codewords. The proposed representation takes into account volumes of hyper-regions obtained from hyperspheres centered at codewords. Experimental results demonstrated that our strategy either has improved the classification rates more than 6% with respect to the compared mid-level representation for UCF Sports, or it is a competitive one, for KTH and UCF-11.

Gameplay Genre Video Classification by Using Mid-Level Video Representation

As video gameplay recording and streaming is becoming very popular on the Internet, there is an increasing need for automatic classification solutions to help service providers with indexing the huge amount of content and users with finding relevant content. The automatic classification of gameplay videos into specific genres is not a trivial task due to their high content diversity. This paper address the problem of classifying video gameplay recordings into different genres by using mid-level video representation based on the BossaNova descriptor. The paper also proposes a public dataset called GameGenre containing 700 gameplay videos groped into 7 genres. The results from experimental testing show up to 89% classification accuracy when the gameplay videos are described by BossaNova descriptor using BinBoost as low-level image descriptor.

Kernel Combination Through Genetic Programming for Image Classification

Support vector machine is a supervised learning technique which uses kernels to perform nonlinear separations of data. In this work, we propose a combination of kernels through genetic programming in which the individual fitness is obtained by a K-NN classifier using a kernel-based distance measure. Experiments have shown that our method KGP-K is much faster than other methods during training, but it is still able to generate individuals (i.e., kernels) with competitive performance (in terms of accuracy) to the ones that were produced by other methods. KGP-K produces reasonable kernels to use in the SVM with no knowledge about the distribution of data, even if they could be more complex than the ones generated by other methods and, therefore, they need more time during tests.

Hierarchical Combination of Semantic Visual Words for Image Classification and Clustering

Image classification and image clustering are two important tasks related to image analysis. In this work a two-level hierarchical model for both tasks using a hierarchical combination of image descriptors is presented. The construction of a latent semantic representation for images is also presented and its impact on the results of both tasks for the two-level hierarchical model is evaluated. Experiments have shown the superior performance attained by the hierarchical combination of descriptors when compared to the simple concatenation of them or to the use of single descriptors. The hierarchical combination of a latent semantic representation has presented results similar to the other hierarchical combinations, using only a small fraction of the time and space needed by others, which is interesting specially for those with restrictions of computer power and/or storage space.