摘要 |
Methods and apparatus, including computer program products, implementing techniques for identifying one or more objects within an image. The techniques include receiving an image that includes at least one object; identifying a plurality of edge pixels in the image, an edge pixel being a pixel that borders two contrasting areas of the image, the plurality of edge pixels including both inner edge pixels and outer edge pixels; and finding a substantially connected component correlated with each object, the substantially connected component comprising a set of the edge pixels that are connected by traversing substantially only edge pixels.
|