Interpretation of tactile data from an FSR pressure pad transducer using image processing techniques

The thesis investigates the use of an array of force sensing resistors (FSRs) as a tactile imaging device. A complete imaging system using the FSR array has been developed as a combination of hardware and software. The data acquisition circuit, implemented in hardware, uses a combination of analog and digital circuitry to scan each element of the array. The raw data is then processed using software techniques making use of image processing algorithms.;Objects with simple circular and rectangular footprints are used as test objects. Algorithms have been devised to locate and identify the objects, placed within the array, by finding its dimensions and the position of its centroids. Two methods have been developed to locate the objects, namely, a heuristic method and a model based method. The heuristic method uses a priori knowledge about the object footprint, which is incorporated using morphological filtering. In some cases, subpixel image processing techniques are used to artificially increase the size of the array due to the large dimensions of each element and the small number of total elements.;The second method to locate and indentify objects uses Bayesian image restoration techniques. The images are modeled as a Markov Random Field (MRF) with the probability distribution governed by the Gibbs distributions. The discontinuities at the edges in the images are modeled using the line field. No other preprocessing is carried out on the raw image. The results on real data suggest that this technique can also be exploited to locate and identify the objects. (Abstract shortened by UMI.).