Localization of the inspected chip image is one of the key problems with machine vision aided surface mount devices (SMD) and other micro-electronic equipments. This paper presents a new edge-directed subpixel edge lo...Localization of the inspected chip image is one of the key problems with machine vision aided surface mount devices (SMD) and other micro-electronic equipments. This paper presents a new edge-directed subpixel edge localization method. The image is divided into two regions, edge and non-edge, using edge detection to emphasize the edge feature. Since the edges of the chip image are straight, they have straight-line characteristics locally and globally. First, the line segments of the straight edge are located to subpixel precision, according to their local straight properties, in a 3×3 neighborhood of the edge region. Second, the subpixel midpoints of the line segments are computed. Finally, the straight edge is fitted using the midpoints and the least square method, according to its global straight property in the entire edge region. In this way, the edge is located to subpixel precision. While fitting the edge, the irregular points are eliminated by the angles of the line segments to improve the precision. We can also distinguish different edges and their intersections using the angles of the line segments and distances between the edge points, then give the vectorial result of the image edge with high precision.展开更多
Copyright and its international complications have presented a significant barrier to the Universal Digital Library （UDL）＇s mission to digitize all the published works of mankind and make them available throughout ...Copyright and its international complications have presented a significant barrier to the Universal Digital Library （UDL）＇s mission to digitize all the published works of mankind and make them available throughout the world. We discuss the effect of existing copyright treaties and various proposals, such as compulsory licensing and the public lending fight that would allow access to copyrighted works without requiring permission of their owners. We argue that these schemes are ineffective for purposes of the UDL. Instead, making use of the international consensus that copyright does not protect facts, information or processes, we propose to scan works digitally to extract their intellectual content, and then generate by machine synthetic works that capture this content, and then translate the generated works automatically into multiple languages and distribute them free of copyright restriction.展开更多
文摘Localization of the inspected chip image is one of the key problems with machine vision aided surface mount devices (SMD) and other micro-electronic equipments. This paper presents a new edge-directed subpixel edge localization method. The image is divided into two regions, edge and non-edge, using edge detection to emphasize the edge feature. Since the edges of the chip image are straight, they have straight-line characteristics locally and globally. First, the line segments of the straight edge are located to subpixel precision, according to their local straight properties, in a 3×3 neighborhood of the edge region. Second, the subpixel midpoints of the line segments are computed. Finally, the straight edge is fitted using the midpoints and the least square method, according to its global straight property in the entire edge region. In this way, the edge is located to subpixel precision. While fitting the edge, the irregular points are eliminated by the angles of the line segments to improve the precision. We can also distinguish different edges and their intersections using the angles of the line segments and distances between the edge points, then give the vectorial result of the image edge with high precision.
文摘Copyright and its international complications have presented a significant barrier to the Universal Digital Library （UDL）＇s mission to digitize all the published works of mankind and make them available throughout the world. We discuss the effect of existing copyright treaties and various proposals, such as compulsory licensing and the public lending fight that would allow access to copyrighted works without requiring permission of their owners. We argue that these schemes are ineffective for purposes of the UDL. Instead, making use of the international consensus that copyright does not protect facts, information or processes, we propose to scan works digitally to extract their intellectual content, and then generate by machine synthetic works that capture this content, and then translate the generated works automatically into multiple languages and distribute them free of copyright restriction.
