Soft Data Fusion in Image Processing

Abstract

Abstract. Data fusion is a long tenn of research in image processing that is becoming more and more relevant owing to the complementary developments of computer and sensory technologies. Although operator research related to soft-computing, specially in the field of fuzzy systems, has evolved considerably during this last two decades, implemented frameworks of data fusion for image processing take seldom into consideration this kind of operators. Most of pattern recognition systems with image fusion are still based in basic operators, e.g. minimum or product. The purpose of the here presented tutorial is to analyze this fact, present some of the fuzzy aggregation operators in the context of data fusion for image processing and show some applications where the usage of the fuzzy integral, one of these operators, increased the perfonnance of image processing systems considering data fusion.

Color Seal Extraction from Documents: Robustness through Soft Data Fusion

“…Soft data fusion is a conceptual framework presented in [13]. This section gives a general overview on it.…”

Image enhancement through intelligent localized fusion operators in the automated visual inspection of highly reflective surfaces

“…(7), and f 3 , Eq. (8). The evolution of these fitness functions for the best individual of the population in each generation is analyzed for three values of crossover probability and four of mutation probability.…”

“…The theoretical framework of soft data fusion [8,9] relate different fusion operators by taking the flexibility of the operation result into account. In this sense traditionally used fusion operators can be considered as hard.…”

“…In the framework presented in [5] a multi-dimensional set of images taken under different illumination condi- tions are fused with a so-called Intelligent Localized Fusion Operator (ILFO), which is a further development of the fuzzy integral [7] within a theoretical framework denoted as soft data fusion [8,9]. This theoretical framework focus its attention on the role of the operator binding the data from the different information sources in multi-sensory systems.…”

“…(b) Effect of the tolerance variation (e) on the final results of the automated pre-processing system. Fuzzy measures are constructed after a GA with the fitness function f 3(8) and parameterized according to…”