| Summary: | 碩士 === 國立臺灣科技大學 === 資訊管理系 === 105 === Word art is a kind of stylization art that represents the content of image through characters. More common types can be divided into ASCII Art, NPR Packing and Collage. But the aforementioned techniques do not represent the outline of the image contour by the shape structure of the characters themselves. Therefore, this paper proposes a system that can automatically recognize the shape of characters to collage image contours. A user can input an original image or simple contour images, and by some method in image pre-processing to obtain the main contour lines of the image. Then through the proposed image matching model to identify the outline of the character that can represent the image contour. Next we use selection and combination mechanisms to find the candidate character that can represent contour. In order to enhance the image matching speed, we use the GPU multi-thread parallel processing frameworks, and compare with those using single-core and multi-core ones. Finally, we show the results that are automatically generated by our system, as well as the limitations of the system and future prospects.
|
|---|
