【摘要】 首先对灰度化后的图像进行小波重构以及二值形态学处理,然后运用多尺度边缘检测加强边缘,最后通过投影定位出字幕区域。仿真实验表明,此方法在复杂多变的视频背景下,仍然有比较好的效果。更多还原

【Abstract】 First the gray image is reconstructed based on wavelet and processed with binary morphology.Then the image edge is strengthened by multi-scale edge detection,and the caption regional is located by projection method.The simulation experiments show that the method can locate caption effectively in the complicated background.更多还原