Length calculation and determination device, angle calculation and determination device and image determination system
AssigneeSeiko Epson Corporation
InventorsHisao Sato, Katsumi Komagamine
Life of this patent
Estimating the term…
Abstract
A calculation section performs a simplified calculation to obtain lengths and/or angles using calculations of addition and multiplication, without using calculations of division, square root and inverse trigonometric function that impose heavy load to the hardware. A data storage section stores reference length data and reference angle data in a data format that is appropriate for the simplified calculation.
Claims27
A method of identifying a target pattern in an image, said method comprising:
- a) establishing an inventory of characteristic pattern points within said target pattern and establishing a first record of relative distances between selected characteristic pattern points, said relative distances being recorded as the square of the actual distances between said selected characteristic pattern points;
- b) partitioning said image into a plurality of sub-image points;
- c) identifying at least two sub-image points having attributes substantially similar to a corresponding two of said characteristic pattern points;
- d) calculating the square of the distance between said identified two sub-image points and comparing said calculated square to the recorded relative distance of said corresponding two characteristic pattern points;
- e) determining that said target pattern has not been identified in response to no match being found in step (d).
A method of identifying a target pattern in an image, said method comprising:
- a) establishing an inventory of characteristic pattern points within said target pattern;
- b) grouping said characteristic pattern points into three-point sets defining a first angle having a characteristic pattern point at a vertex of said first angle and two characteristic pattern points at side-ends of said first angle, assigning an identification indicator to all three characteristic pattern points in each three-point set specifying which pattern point is at the vertex position and which pattern points are at the side-end positions of said first angle;
- c) establishing a first record of a respective first angle parameter proportional to a corresponding first angle within each three-point set, each first angle parameter being defined as the product of its corresponding first angle's side-lengths multiplied by the cosine of its corresponding first angle;
- d) partitioning said image into a plurality of sub-image points;
- e) identifying at least three sub-image points having attributes substantially similar to a corresponding three of said characteristic pattern points;
- f) calculating a second angle parameter characteristic of a second angle defined by said three identified sub-image points, and comparing said second angle parameter to said first angle parameters;
- g) determining that said target pattern has not been identified in response to no match being found in step (f).
A pattern recognition system for identifying a target pattern within an image, said pattern recognition system comprising:
- a memory store holding first key data descriptive of selected characteristic pattern points within said target pattern, said characteristic pattern points being grouped into three-point sets constituting a first triangle, said first key data including at least one of first length attribute data and first angle attribute data, said first length attribute data being the square of the sides of said first triangle, said first angle attribute data corresponding at least to one of the three angles within said first triangle and being the cosine of a first angle multiplied by the length of the sides adjoining said first angle;
- a processing unit having access to said memory store and having an input for receiving second key data descriptive of selected sub-image points of said image, said selected sub-image points having image attributes substantially similar to corresponding characteristic pattern points and said second key data including position information of its corresponding sub-image point, said processing unit collecting said second key data into data groups corresponding to three sub-image points constituting a second triangle and using said position information to calculate at least one of second length attribute data and second angle attribute data distinctive of said second triangle, said processing unit comparing available second length attribute data to available first attribute data and comparing available second angle attribute data to available first angle attribute data to determine whether said second triangle is congruent to said first triangle.
Dependencies are inferred from the claim text.