ANIMATION RETRIEVAL DEVICE

摘要

PROBLEM TO BE SOLVED: To improve the detection accuracy of captions and also to fast detect the captions by counting the number of caption pixels in every block and extracting the blocks having the count value larger than the prescribed threshold as the caption blocks. SOLUTION: A caption pixel count means 102 defines the pixels having their luminance value larger than the 2nd threshold as the caption pixels and counts these pixels in every block for the frame image that is inputted via an input terminal 101. A caption block extraction means 103 produces a binary image where a block having its count value larger than the 1st threshold is defined as 1 with other blocks defined as 0 respectively. A caption frame extraction means 104 detects the section of a frame including the captions by means of the binary image and extracts the start frame of the detected section as a caption frame. An image display means 106 shows a list of images of caption frames stored in an image information storage means 105 and then reproduces the animation image out of the frame corresponding to the caption frame that is selected from the displayed list.