Sharp Kabushiki Kaisha Corporation

Real World Computing Partnership

Hitachi, Ltd.

Agency of Industrial Science and Technology

The Japan Iron and Steel Federation

Ryuichi Oka

Susumu Seki

Jiro Kiyama

Shigeki Nagaya

Hiroshi Kojima

Yoshiaki Itoh

The present invention is intended to provide an image recognition method which, in recognizing an input image including an object presumable to have continuity like a human figure, enables simultaneously detecting the shape of the object in the input image and cutting-out the object therefrom, and obtaining information on position of the object at the same time, and which also enables recognition of two objects having the same shape but having different sizes. The processing procedure of image recognition according to the present invention consists of: inputting an input frame image of an object by image data inputting means; preparing scaled-down images 1 to n at different resolutions by image size reducing means; calculating neighborhood correlation values for respective scaled-down images 1 to n by neighborhood correlation processing means and checking relationship between them at the same time; calculating the area and the gravity center of each image using pixel values and pixel coordinate value; identifying respective objects at respective resolution on the basis of their relationship; determining shape correlation of each object; and recognizing the shape of each object by comparing the obtained shape auto-correlation with a previously determined reference value.

Image recognition method

A speech recognition system automatically designates a scope of a partial reference pattern. Plural reference patterns, each of which ends in each of composing frames and starts from a preceding frame, are supposed and cumulative distances at every frame are calculated. A partial reference pattern that has a minimal distance value as compared with all other partial reference patterns is taken as a partial input speech recognizing result.

Speech recognizing device and method assuming a current frame is an end

Comparatively macroscopic phonemic units consisting of continuous sounds are used as standard patterns in continuous dynamic programming. In matching intervals between these standard patterns and input patterns, a sampled pattern is labeled according to the phonemes and phonemic particles which form the standard patterns. In the matching intervals, the labeled pattern is called a matching history. The matching history of a standard pattern when the continuous dynamic programming value is below a certain threshold value is used to recognize the phonemes of the input patterns.

Phoneme labelling of frame by accumulating evidence obtained from

Growing community of inventors

Ibaraki, Japan

Ryuichi Oka