Raul Pinto Elias

Edge Preserving Lossy Image Compression with Wavelets and Contourlets

In this paper an edge preserving lossy image coder is presented. An edge image is obtained from the original with a digital image processing module using four different filters: Canny, Sobel, Roberts and Prewitt, then the original image is domain transformed with wavelets or contourlets, and a pixel mapping from original domain to transformed is done. For the compression, the edges points and the approximation image (which determines the compression factor) are selected; finally the image is decompressed in order to observe the reconstruction quality and edge preserving. The stages of the compression system are: a) Image selection, b) Digital image processing, c) Feature vector extraction, d) Domain Transformation, e) Pixel mapping, f) Image compression and g) Image decompression. Additionally, the results obtained from comparisons of error measures between original and decompressed images are shown and finally conclusions about the coder are presented.

Feature Preserving Image Compression: A Survey

With the increase of the use of Internet and wireless mobile devices, the digital information needs to be send and received efficiently in low bit rates in order to exploit the bandwidth. At low bit rates it is almost impossible to generate errors or artifacts in images. In order to solve that problem, researchers are trying to design and build lossy image coders which can preserve important features of images. With this approach the features of an image that are very important to perception and recognition are preserved even at low bit rates. In this paper a revision of some works that propose feature preserving image compression (FPIC) algorithms are presented. Finally a new methodology to obtain FPIC is presented

Edging out the competition: Lossy image coding with wavelets and contourlets

Image compression offers a good representation of images while using the least quantity of bits. Several lossy image coders are designed without considering the image nature. The image important information (e.g., edges) can be discarded at the coding quantization stage; that information is needed for image understanding and recognition. The possibility of saving storage space and preserving image important information in a joint way becomes imperative in areas such as medicine, mobile devices, and pattern recognition systems. This article addresses the design of an edge-preserving lossy image coder by means of wavelets and contourlets. The results obtained demonstrate the superior performance of the proposed coder against the traditional edge-preserving coders. The proposed coder ensures that the edges of an image are always preserved even at very low bit rates and the obtained decompressed images can be successfully used for future pattern recognition tasks.

Identification of Relevant Features based on the Variation of Chrono-Valued Descriptors

In this paper the problem of feature selection in chronological information type is dealt. Objects are described by several tuples chained chrono-valued variables. We work with a universe of data consisting of sets of objects separated into classes wherein each object is described by t tuples of variables and also each tuple contains n variables. In general we can speak of two types of problems where data points are used and where no specific data are used. In the first case the data obtained from static objects, such as images; in the second case the data are obtained from dynamic objects, such as videos. Found research in which they treat information that care about the order of the data but with objects that are represented by a single tuple of variables, that is, are objects that represent the same phenomenon. This research is of great importance because it is intended to work with information that has two main features. The first feature is that objects are described by several tuples chained chrono-valued variables, that is, each object is represented in matrix form. The second feature is that a phenomenon described by the same object but at different points in time.

Selective Search Method for Object Localization and Detection using Wavelets and Hierarchical Segmentations

This article proposes a selective search method for object localization in natural images by applying image multi-segmentation, image scaling, and heuristics. The method increases the number of generated windows that delimitate the area of an object with an accuracy superior to 50%. Over-segmentation is applied on original size images in order to locate small objects, and it is also applied over scaled images because these can still be over-segmented. This process produces less regions on areas with many textures. The over-segmentation was applied using the CIE Luv color model, and using the H and the I channels of the HSI model. The proposed method is category independent and allows the location of objects with heterogeneous characteristics by using heuristics and hierarchical segmentation. The proposed method produces 9, 366 windows per image covering 96.78% of the objects in the PASCAL VOC 2007 test image collection, increasing in 0.8% the localization results reported in the state of the art.