Recognition apparatus and recognition method
AssigneeKabushiki Kaisha Toshiba
InventorKentaro Yokoi
Life of this patent
Estimating the term…
Abstract
A recognition apparatus of this invention for recognizing sorting information as a character string written within a cellophane region of a paper-like material extracts edge components according to a differential image based on an image of light and shade of the paper-like material, determines a cellophane frame based on the edge components and recognizes the character string within the thus determined cellophane frame as an address. Thus, the cellophane region in which the address used as the sorting information is written can be correctly extracted and occurrence of erroneous recognition of the sorting information and erroneous sorting of the paper-like material can be prevented.
Claims17
A recognition apparatus for recognizing sorting information as a character string written within one of a cellophane region and label region of a paper-like material, comprising:
- reading means for reading a density value of each pixel in a main scanning direction and sub-scanning direction of the paper-like material;
- first determining means for determining a portion in which a variation in the density value exceeds a preset value based on the density value of each pixel read by said reading means;
- second determining means for determining said one of the cellophane region and label region of the paper-like material based on the portion which is determined by said first determining means that a variation in the density value exceeds the preset value;
- and
- recognizing means for recognizing the sorting information based on the character string written within said one of the cellophane region and label region determined by said second determining means,
- wherein said first determining means includes
- differentiating means for differentiating the density value of each pixel read by said reading means,
- first extracting means for extracting edge components of one image based on a differential signal of one image obtained by said differentiating means, and
- second extracting means for extracting horizontal and vertical projection images for the edge components of one image extracted by said first extracting means, and
- wherein said second determining means determines a region in which the edge components are concentrated as said one of the cellophane region and label region based on the horizontal and vertical projection images extracted by said second extracting means.
A recognition method for recognizing sorting information as a character string written within one of a cellophane region and label region of a paper-like material, comprising:
- a reading step of reading a density value of each pixel in a main scanning direction and sub-scanning direction of the paper-like material;
- a first determining step of determining a portion in which a variation in the density value exceeds a preset value based on the density value of each pixel read in said reading step;
- a second determining step of determining said one of the cellophane region and label region of the paper-like material based on the portion which is determined in said first determining step that a variation in the density value exceeds the preset value;
- and
- a recognizing step of recognizing the sorting information based on the character string written within said one of the cellophane region and label region determined in said second determining step;
- wherein said first determining step includes
- a differentiating step of differentiating the density value of each pixel read in said reading step,
- a first extracting step of extracting edge components of one image based on a differential signal of one image obtained in said differentiating steps and
- a second extracting step of extracting horizontal and vertical projection images for the edge components of one image extracted in said first extracting step, and
- wherein said second determining step determines a region in which the edge components are concentrated as said one of the cellophane region and label region based on the horizontal and vertical projection images extracted in said second extracting step.
A recognition apparatus for recognizing sorting information as a character string written within one of a cellophane region and label region of a paper-like material, comprising:
- a reading portion which reads a density value of each pixel in a main scanning direction and sub-scanning direction of the paper-like material;
- a density-variation determining portion which determines a portion in which a variation in the density value exceeds a preset value based on the density value of each pixel read by said reading portion;
- a region determining portion which determines said one of the cellophane region and label region of the paper-like material based on the portion which is determined by said density-variation determining portion that a variation in the density value exceeds the preset value;
- and
- a recognizing portion which recognizes the sorting information based on the character string written within said one of the cellophane region and label region determined by said region determining portion;
- wherein said density-variation determining section includes
- a differentiating portion which differentiates the density value of each pixel read by said reading portion,
- a edge extracting portion which extracts edge components of one image based on a differential signal of one image obtained by said differentiating portion, and
- a edge-projection extracting portion which extracts horizontal and vertical projection images for the edge components of one image extracted by said edge extracting portion;
- and
- wherein said region determining portion determines a region in which the edge components are concentrated as said one of the cellophane region and label region based on the horizontal and vertical projection images extracted by said edge-projection extracting portion.
Dependencies are inferred from the claim text.