| 摘要 |
A system for automatic thresholding video data to educe and detect objects of interest from both high and low contrast imagery is disclosured here, which makes an object screening system adaptive to changing scene intensity and contrast levels automatically is disclosed. The real-time adaptive image interval extraction device extracts the edges of an object and simultaneously generates a bright signal for intensity larger than a given threshold. The interval, over which an object may exist, is derived from a logical combination of edge and bright signals. The video data is smoothed by a two-dimensional, weighted average, low pass or smoothing filter and passed through two filters, an edge filter and a bright filter. The edge filter determines the edge intensity magnitude associated between sucessive picture elements or pixels in the scan line of video data. An edge threshold is determined automatically from the average edge intensity of each scan line of data. When the data edge becomes greater than the edge threshold, a logical edge output is produced. In the same manner, the data is also passed through a bright filter which emphasizes data above a given background threshold. The background and threshold is updated on a line by line basis. This produces a logical bright signal when the intensity is above the automatically determined threshold.
|
