Text enhancement in digital video

Text information extraction in images and video: a survey

“…Li et al [64] presented a text tracking approach that is suitable for several circumstances, including scrolling, captions, text printed on an athlete's jersey, etc. They used the sum of the squared di erence for a pure translational motion model, based on multi-resolution matching, to reduce the computational complexity.…”

Fast and robust text detection in images and video frames

“…Text should be clearly extracted from its background to obtain a good recognition result for the characters. Temporal information can be used to enhance the quality of video text as in [26][27][28]. Special technique should be investigated to segment the characters from their background before putting them into an OCR software in the future work as in [29][30][31].…”

Adaptive Fuzzy Text Segmentation in Images with Complex Backgrounds Using Color and Texture

“…The excerpt shown in Figure 1 consisting of three images extracted from the test sets used in this paper illustrates various complex backgrounds where text is embedded. The proposed method overcomes the difficulties for finding the optimal global/local threshold [12,22,11,21] or the request for different training samples in [15,4] by applying unsupervised fuzzy clustering. In the following, a short review of the fuzzy c-means algorithm is given before proceeding further by explaining its application in our text segmentation technique.…”

“…The first peak from the left on the smoothed histogram is choosen as the optimal threshold for the binarization process. After enhancing the image using Shannon up-sampling, Li et al [11] apply a local thresholding method to binarize the enhanced image. A block is marked as background only if its standard deviation is smaller than a fixed threshold.…”