Adaptive prediction for binary encoded documents containing a mixture of text, line drawings and halftones
First Claim
Patent Images
1. A system for adaptively predicting a block of image data comprising:
- a plurality of predictors, each using a different predictor pattern for predicting the data bits of each image data block,a comparator for each predictor for generating an error code representative of the accuracy of the predictions made,selector means for selecting one of the predictors, based on its performance on one or several previous data blocks, andmeans responsive to the output of said selector means for outputting the error code for the current block using the selected predictor.
2 Assignments
0 Petitions
Accused Products
Abstract
This is a circuit and method for adaptively selecting one of a plurality of predictor patterns to encode a stream of image data blocks such as would be output from a raster-input-scanner. A variety of halftone and nonhalftone predictor patterns are used and compared, with the best one being selected to predict the next block. Using this technique, the best predictor pattern is usually selected, predicted data buffering is not needed, and no code need be transmitted with the data to the receiver to specify the predictor to be used in the deprediction process since the same basic information is also available to the receiver.
-
Citations
13 Claims
-
1. A system for adaptively predicting a block of image data comprising:
-
a plurality of predictors, each using a different predictor pattern for predicting the data bits of each image data block, a comparator for each predictor for generating an error code representative of the accuracy of the predictions made, selector means for selecting one of the predictors, based on its performance on one or several previous data blocks, and means responsive to the output of said selector means for outputting the error code for the current block using the selected predictor. - View Dependent Claims (2, 3, 4, 5, 6, 7)
-
-
8. The method of adaptively predicting a series of blocks of image data comprising the steps of:
-
first using a plurality of predictors for generating error codes for one or a plurality of previous consecutive data blocks, selecting the best predictor based on its performance, and second using said selected predictor for generating the output error code for the current block. - View Dependent Claims (9, 10, 11, 12)
-
-
13. The method of adaptively predicting a series of blocks of image data comprising the steps of:
-
first using a plurality of halftone predictors to generate error code for a plurality of previous data blocks, second using one or a plurality of nonhalftone predictors to generate error code for the previous data block, selecting;
from a group comprising (a) the halftone predictor with the best performance for all of said plurality of previous data blocks, if there is one and (b) all nonhalftone predictors for the previous block;
the one with the best performance in generating error code for the previous block, andthird using said selected predictor to generate output error code for the current block.
-
Specification