Héctor A. Montes-Venegas

Color-based texture image segmentation for vehicle detection

This paper describes a novel method for detecting vehicles on a highway using two visual features: color and texture. Our method consists of a segmentation process computed on the L*u*v* color space and a texture feature extraction procedure based on Dual-Tree Complex Wavelet Transform. We also apply a denoising process using morphological operations to build a background model and make possible the detection of the vehicles in the scene.

A Modification of the TPVD Algorithm for Data Embedding

Pixel-Value-Differencing PVD methods for data hiding have the advantage of a high payload. These algorithms have however the problem of overflow/underflow pixels, thus a location map for those pixels usually ignored when embedding the message is necessary. In this paper, we modified the Tri-way Pixel-Value Differencing method that removes the need of the location map and fix the problem. Our proposal replaces the table of ranges to estimate the amount of information to be embedded by a function based on the floor and ceil functions. As for the problem of overflow/underflow pixels we tackle it by means of a linear transformation. The linear transformation is based on the floor function, so information is lost therefore a location map to compensate for this data lost is necessary to recover the embedded message. The inclusion of the map in the algorithm is also discussed. The technique uses two steganographic methods, namely, the tri-way method to store the message and a reversible steganographic method to store the map needed to invert the linear function in order to recover the encoded message.

Estimating the relevance on Communication Lines Based on the Number of Edge Covers

Abstract Counting the number of edge covers on graphs, denoted as the #Edge_Covers problem, is a #P-complete problem. Knowing the number of edge covers is useful for estimating the relevance of the lines in a communication network, which is an important measure in the reliability analysis of a network. In this paper, we present efficient algorithms for solving the #Edge_Covers problem on the most common network topologies, namely, Bus, Stars, Trees and Rings. We show that if the topology of the network G does not contain intersecting cycles (any pair of cycles with common edges), then the number of edge covers can be computed in linear time on the size of G.

An Optimization Approach to the TWPVD Method for Digital Image Steganography

In Digital Image Steganography, Pixel-Value Differencing methods commonly use the difference between two consecutive pixel values to determine the amount of data bits that can be inserted in every pixel pair. The advantage of these methods is the overall amount of data that an image can carry. However, these algorithms frequently either overflow or underflow the pixel values resulting in an incorrect output image. To circumvent this issue, either a number of extra steps are added to adjust those values, or simply the pixels are deemed unusable and they are ignored. In this paper, we adopt the Tri-way Pixel-Value Differencing method and find an optimal pixel value for each computed pixel block such that their difference holds the maximum input data and neither underflow or overflow pixels exist.

Measuring rainbow trout by using simple statistics

Traditionally, a manual method is used to classify the rainbow trout in small farms, which generally cause stress and physical damage to the fish. Additionally, this manual classification is not always accurate, as farmers only visually check whether the trout is fry, fingerling, or table-fish size. In this article, we robustly evaluate our simple statistical system to measure rainbow trout in farms. For this research, we have designed and implemented a novel prototype that includes canalization, illumination, and vision components to take a 2D downward-view image of the trout. After that, this image is processed to get the trout’s contour, which is used to estimate the fish’s length by adjusting the best regression curve to this contour. Finally, the trout’s size is defined by the minimum Mahalanobis distance to training data. We have evaluated our experimental results as a binary classification problem and the best precision scores are 95.93%, 93.21%, and 96.25% when classifying fry, fingerling, and table-fish trout, respectively. These experimental results are computed by using our state-of-the-art database of 1800 rainbow trout images, which was collected with our physical prototype for this publication.

Improved Image Dominant Plane Extraction for Robot Navigation

The dominant plane is a planar area that occupies the largest portion of an image captured by a camera. In this paper, we present an improved method for extracting the dominant plane from optical flow and use the resultant image to steer a mobile robot along a corridor route. In this route the dominant plane is the free space where the robot can navigate without colliding with the obstacles cluttering the environment. Our experimental results illustrate significant performance improvements over those published in previous works.

A New Optimization Strategy for Solving the Fall-Off Boundary Value Problem in Pixel-Value Differencing Steganography

In Digital Image Steganography, Pixel-Value Differencing (PVD) methods use the difference between neighboring pixel values to determine the amount of data bits to be inserted. The main advantage of these methods is the size of input data that an image can hold. However, the fall-off boundary problem and the fall in error problem are persistent in many PVD steganographic methods. This results in an incorrect output image. To fix these issues, usually the pixel values are either somehow adjusted or simply not considered to carry part of the input data. In this paper, we enhance the Tri-way Pixel-Value Differencing method by finding an optimal pixel value for each pixel pair such that it carries the maximum input data possible without ignoring any pair and without yielding incorrect pixel values.

Vehicle tracking based on multiple hypotheses

This paper describes a vehicle tracking method that uses texture, color, size, distance and trajectory as modeling features. Before the tracking task starts, a representation to detect the target vehicles is constructed. Two methods are used to perform vehicle detection. The first method uses color, texture and a background model to detect the vehicle regions. The second one uses texture and lightness differences between the current frame and a previously modeled background. An experimental comparison of the two vehicle detection methods is performed both qualitatively and quantitatively in order to choose the most suitable one. Vehicle tracking is then achieved through a multiple hypotheses tracking method that integrates size, color, distance and trajectory in a single similarity vector by using a hierarchical analysis.

An Evolutionary Hybrid Strategy for Solving the Rural Postman Problem

The Rural Postman Problem (RPP) is a classical arc routing problem proven to be NP-Hard whith applications in many contexts of practical interest. A common strategy for solving RPP instances is to first determine a suitable graph transformation in order to either reduce the dimensionality of the search space or to produce a single cursus graph to derive an Eulerian circuit in polynomial time. In this paper we present a simple but effective hybrid heuristic for the RPP that uses such a graph transformation and finds solutions using a Genetic Algorithm for global search and a local search algorithm to compute optimal traversal directions of a solution tour. Our approach is tested on a set of instances than have been already used in previously published work.

Detecting obstacle-free regions for visual robot navigation by inferring scene horizons

We present preliminary results of an algorithm for detecting obstacle-free regions in indoor environments using both color and texture information for visual robot navigation. By modeling color information in the L∗u∗v∗ color space, a color-based segmentation is performed to find similar regions. This segmentation yields a set of regions that are joined together into single areas using texture information. Texture-based similarity measures between segmented regions are computed using ANOVA (Analysis of Variance), which produces the final obstacle-free regions. These regions are then used as visual information to drive a robot towards the scene horizon, i.e. the boundary between visible free space and unexplored areas. The performance of our algorithm was tested on several sequences of images taken from office corridors.