Detecting Main Body Size in Document Images

Directional Hinge Features for Writer Identification: The Importance of the Skeleton and the Effects of Character Size and Pixel Intensity

“…Furthermore, our previous work on Detecting the Main Body Size [45] made apparent that even in a single text written by one writer, Main Body Size fluctuations could be observed. This observation motivated us to utilize this information and explore its contribution towards identifying the writer.…”

“…It is usually referred to words. This characteristic aims to provide a reference for thresholds and sizes of lines, words and characters as it is directly related to the size of the characters [45].…”

“…In the experiments performed on [45] an average error deviation of 2.17 pixels was recorded on handwritten document images, while an average error deviation of 0.67 was recorded for machine-printed documents for MBS estimation using the technique described. Moreover, by identifying the most common MBS in the document, the optimal value for C along with the optimal threshold is also found.…”

Language Identification by Using SIFT Features

“…Generally speaking, the mentioned systems are trying to extract relevant features from the images in order to create a histogram representation of every image (Fig.3) that will be used as input to a classifier. The BoVF system that we implement consists of the following steps: The first step consists of the application of the Main Body detection algorithm, described in [16], to the document image. This will return a number n which indicates how tall is the main body of the majority of the characters in the image.…”