Título del libro: Proceedings Of The 2010 International Conference On Artificial Intelligence, Icai 2010
Título del capítulo: Robot skill acquisition: A new method for learning and fast object recognition
JUAN MARIO PEÑA CABRERA; ROMAN VICTORIANO OSORIO COMPARAN; HUMBERTO GOMEZ NARANJO;
Autores externos:
Idioma:
Año de publicación:
2010
Palabras clave:
Artificial; Assembly; Manufacturing; Neural; Robotics; Vision
Resumen:
Machines that perceive their environments and perform required tasks have an obvious usefulness for diverse application areas such as planetary space exploration, automated medical applications, and industrial assembly and inspection. They can assist in many tasks that are routine, tedious and even dangerous for human to perform. The ability to control an industrial or any other process which requires such a capability, relies heavily on the quality of the sensing devices that input information as well as the input codification. The industrial environment today offers many potential applications for image processing, some important applications are: inspection of items involved in manufacturing processes with robot manipulation of raw materials, partly manufactured or completed manufactured products, these applications require as well identification of items involving item type, position and orientation for automatic packing and assembly processes. Other factors such as lighting and optical systems are very important, there is no point in producing a highly accurate and intelligent computer vision system if there is no possible means of obtaining dynamic range out of the image transducer and a robust and efficient vector descriptor of the objects to be used in real time with adaptive lighting capabilities to obtain an even distribution of light within the picture area due to the impossibility of stabilizing the ambient light of the assembly environment. In this sense, the vision system has to analyse the distribution of light on the scene each time a picture is taken by a camera and sets the appropriate sensitivity and threshold values to achieve an object-background separation stage to produce binary images of the interested objects within a region of interest. The gray-levels in a digital image can be histogrammed, and a histogram of gray-levels can be defined as the plot of the gray-scale versus the frequency of occurrence of each gray-level in a digital image and is useful for automatic threshold adjustments when using threshold selection techniques for image segmentation. Segmentation of images using thresholding is carried out by way of a threshold operator "T", different methods for determining the value of "T" can be used as: a constant value, a value dependent upon the average of
Entidades citadas de la UNAM:
ISBN:
9781601321480
Editorial:
Tipo de producto:
Conference Paper
