Showing papers by "Thanarat H. Chalidabhongse published in 2018"

Vision-Based Bus Route Number Reader for Visually Impaired Travelers

Abstract: In this paper, we present vision-based bus route number reader by using machine learning and image processing techniques. This prototype system is designed for people with low vision to recognize bus route number from natural scene images. Our proposed system consists of three stages: Bus panel detection, image enhancement, and bus number recognition. The bus panel detection is done using faster RCNN. The pre-processing is applied to enhance bus panel images. Next, the bus route number is recognized using Google Cloud Vision API. Finally, the recognized bus route number is translated to synthesized audio. The prototype system can be applied to develop a mobile application or embedded system at the bus stop to aid people with low vision when taking the public bus transportation. Experimental results presented in precision, recall and f-measure in bus panel detection stage and obtain 67.9 percent accuracy of bus number recognition.

A speedup scheme for MRF stereo using local label hierarchy

Abstract: Random field formulation has proven to be a powerful framework for solving stereo correspondence problems because of its ability to intuitively incorporate global smoothness constraint with local matching costs. However, solving such problems for cases where large number of pixel variables and possible disparity labels are common can be impractical as the computational complexity grows fast with the number of labels. We proposed a speedup scheme using local label hierarchy in which we exploit characteristics of stereo vision problem to obtain a hierarchical energy minimization technique. In doing so, we give definitions and notations for local label hierarchy as well as approaches for label-wise grouping. We also generalize the definition of energy function to include sets of labels and present heuristics for assigning group potentials. Our approach builds different “local” hierarchy for each variable using information from the energy function which enables us to achieve better performance when compared to using the same hierarchy for every variable. The added processing steps have significantly less theoretical computational complexity than the overall process. Our method was tested with different combinations of cost functions, and our experiment has shown that our heuristics can assist in speeding up the computation time while providing comparable energy and error.