A character recognition method involves the inputting of a character attached to an article as an image signal, forming the input image signal into an image pattern, and matching the image pattern with a standard pattern. The character to be recognized is prepared by superimposing on an ordinary character a plurality of marks in a pattern characterizing that character. A letter recognition apparatus discriminates the character by extracting only the characterizing marks from the image signal through image processing and by matching the information concerning the arrangement of the characteristic marks extracted and standard information relating to standard image patterns which is stored in advance.

A method and apparatus for grading generally spherical articles, and specifically fruit such a oranges, which do not orient along a predetermined axis when rotated on a roller conveyor comprises a pair of side scan cameras disposed on opposite sides of the fruit. An overhead camera provides plural data signals representative of the surface quality of the fruit. The side scan cameras provide data indicative of the surface quality of side portions of the fruit. A processor and processing algorithm computes a surface quality indication based upon the data from the overhead camera and modifies the surface quality indication depending upon the data provided by the side scan cameras.

In the grading of shell eggs passing in a stream through a candling bed, images of the eggs are compared with image data characteristic of eggs of known grade so as to assign a grade to each egg in the stream. Positional information representative of the movement of the eggs is combined with the assigned grade to control a mechanical egg routing device. Images of multiple egg streams may be multiplexed and processed together. The images may be Fourier transformed before comparison.

The present invention consists of comparing a feature pattern extracted from an actual observed image and the prediction pattern of an environmental object predicted in relation to the positional information of a mobile object and finding the difference between both the patterns and then guiding and controlling a mobile object so as to reduce the difference, in that a selection sequence rule of the environmental objects serving as guidance targets is selected and applied by an operator and that the prediction patterns for the guidance targets are created in accordance with the selection sequence rule of the objects. Thus, the comparisons between the prediction patterns of the objects according to the selection sequence rule selected and applied beforehand and the feature pattern from the observed image can be made quickly and accurately, so that the autonomous guidance of the mobile object can have the reliability enhanced more.

Lower accuracy picture coincidence detecting means gives positional relationship between picture data of higher resolution picture and lower resolution picture as corresponding pixel position predicted with marker position as reference, i.e., initial position. Block matching processing means sets a search range corresponding to a block of an arbitrary range set within the lower resolution picture in accordance with the initial position to globally search, within the search range, block on the higher resolution picture corresponding to the block of the arbitrary range to find out a candidate of coincidence area to further interpolate between respective pixels of lower resolution picture and higher resolution picture to thereby evaluate, with high accuracy, similarity between the set block and block within the search range to detect a picture in which coincidence is obtained. For this reason, even if corresponding pictures are pictures such that both pixel positions differ from each other in dependency upon pictures of different formats, e.g., different picture angles, it is possible to detect coincidence between pictures or pixels.