摘要:
An image encoder has image feature extracting means (202) for generating, as shown in the Figure, text data Dc including character codes corresponding to respective text images contained in a document image (201) and auxiliary information on the sizes and positions of the text images based on image data Di on the document image (201). Based on the text data Dc, a predicted document image (205) of the document image (201) is formed, and an arithmetic encoding for changing an event probability model of the document image data Di is performed with reference to the predicted document image data PDi. Thus, the event probabilities of white and black pixels represented by an event probability model are nearly equal to the respective probabilities of white and black pixels in a text image, and the efficiency of the encoding of the text image data by an arithmetic encoder is improved.
摘要:
In an infrared imaging device, shutting means 40 is configured so that it can be opened/closed and so as to shut off an infrared radiation coming into an optical system 20 when it is closed. During an imaging operation, the shutting means 40 is opened, and the optical system 20 causes an infrared radiation from an object 70 to form an image on an infrared detector 10 . The infrared detector 10 outputs a signal according to the amount of infrared light received by each pixel. Correction means 30 corrects sensitivity variations among pixels, the influence of the infrared radiation from the optical system 20 , etc. During a calibration operation, the shutting means 40 is closed, and the correction means 30 determines a correction coefficient for correcting fluctuations in the amount of infrared radiation from the optical system 20 by using the output of the infrared detector 10 imaging the shutting means 40.