【摘要】 在视频或图像中包含丰富的文字信息,这些文字信息与视频或图像的内容密切相关,因此可以作为视频片断或者图像的内容标识和索引。以视频维吾尔文为例,首先用Canny算子将彩色图像转换为灰度化的边缘图像,其次对图像中的文字进行定位,再对图像进行边缘点彩色分割,然后进行文字图像行区域和列区域的检测,最后基于候选文本区的特征从不同复杂度的彩色图像中提取文本信息,实验结果表明了算法的有效性。更多还原

【Abstract】 There are plenty of character information in video or images,which is closely related to the content of video clips and images.So,the character information is often served as the annotation and indexing clues of the corresponding video clips or images.For the Uigur news videos,uses Canny operator the colour image to be shifted for gradation image first,carries out allocation on characters in image secondly,colour carves up the border being in progress again to the image,the image carrying out characters does area and enumerates the area detecting and then,the colour never with complicated degree extracts version information finally owing to waiting for the characteristic choosing version area in the image.There experimental results illustrate the effectiveness of the proposed algorithm.更多还原