Particle recognition device

Details Particle recognition device Particle recognition device

2000-07-25

2004-04-27

Mariam, Daniel (Department: 2621)

Image analysis

Pattern recognition

Feature extraction

Reexamination Certificate

active

06728405

ABSTRACT:

1. TECHNICAL FIELD
The present invention relates to a particle recognition apparatus for separating a particle from a halftone image of a to-be-measured particle group represented by a fertilizer, exploded splinters, or vegetables or fruits and recognizing the particle of an element and, more particularly, to a particle recognition apparatus used to measure a particle size distribution or analyze the shape of a pore in a metal, mineral, or fiber.
2. BACKGROUND ART
As particle recognition apparatuses of this type, conventionally, there are particle recognition apparatuses using labeling or particle recognition apparatuses using distance transform.
[Particle Recognition Apparatus Using Labeling (Prior Art 1)]
A particle recognition apparatus using labeling recognizes a particle with the following procedures.
(1): A particle group to be measured is photographed to obtain a halftone image.
(2): The halftone image obtained in (1) is binarized (“0”, “1”) to obtain a binary image (FIG.
11
(
a
)).
(3): The binary image obtained in (2) is labeled to obtain a label image (FIG.
11
(
b
)).
(4): Each label ({circle around (
1
)} to {circle around (
6
)}) is recognized as one particle.
[Particle Recognition Apparatus Using Distance Transform (Prior Art 2)]
A particle recognition apparatus using distance transform recognizes a particle with the following procedures.
(1): A particle group to be measured is photographed to obtain a halftone image.
(2): The halftone image obtained in (1) is binarized (“0”, “1”) to obtain a binary image.
(3): The binary image obtained in (2) is subjected to distance transform to obtain a distance-transformed image (FIG.
12
(
a
)).
(4): An image Dk obtained by extracting pixels with values equal to or larger than a threshold value k from the distance-transformed image obtained in (3) to obtain a resultant image line Ok (FIG.
13
(
a
)).
(5): The resultant image line Ok obtained in (4) is subjected to exclusive expansion to obtain an image E (FIG.
13
(
b
)).
(6): A cuttable image C (FIG.
13
(
d
)), uncuttable image I (FIG.
13
(
e
)), and new image N (FIG.
13
(
f
)) are classified from an image D
k−1
(FIG.
13
(
c
)).
(7): The image C and image E are ANDed to obtain an image A (FIG.
13
(
g
)).
(8): The image N is smoothed to obtain an image N′ (FIG.
13
(
h
)).
(9): The image A, image I, and image N′ are ORed to obtain an image O
k−1
(FIG.
13
(
i
)).
(10): Steps (5) to (9) are repeated while decrementing the value k to 1, thereby obtaining a particle-separated image.
(11): The particle-separated image obtained in (10) is labeled to obtain a label image (FIG.
12
(
b
)).
(12):: Each label ({circle around (
1
)} to {circle around (
7
)}) is recognized as one particle.
Note that the particle recognition technique using distance transform is disclosed in, e.g., Japanese Patent Laid-Open No. 62-211783 (image processing apparatus).
3. DISCLOSURE OF INVENTION
Problem to be Solved by the Invention
However, according to such a conventional particle recognition apparatus, in the particle recognition apparatus (prior art 1) using labeling, if a plurality of particles in contact with each other are binarized into the binary image, the plurality of particles in contact are recognized as one particle, resulting in low particle recognition accuracy.
More specifically, in particle recognition using an image, generally, the halftone image of particles is often unclear because of the particle photographing environment, Additionally, in the binary image obtained by binarization, a plurality of particles are often binarized in contact with each other. However, the particle recognition apparatus using labeling often recognizes particles in contact with each other as one particle. This often affects the particle recognition result.
If the binarization level is raised to avoid this problem, particles may be recognized as particles with much smaller sizes, or small particles may disappear.
On the other hand, in the particle recognition apparatus (prior art 2) using distance transform, even when a plurality of particles are binarized into a binary image in contact with each other, they can be recognized as the plurality of particles, respectively. This allows accurate particle recognition. However, since this apparatus uses exclusive expansion processing in separating particles, a phenomenon occurs in which the shape of a separated particle becomes different from the original particle shape. Hence, it is difficult to use this apparatus for the application purpose such as pore shape analysis of a metal or mineral. In addition, this apparatus must repeatedly perform not only calculation for distance transform but also calculation for degeneration for image separation, labeling, exclusive expansion, OR, and AND a plurality of number of times, and the calculation amount is enormous. This takes a long processing time, so this apparatus can be introduced only in an environment with a lenient time limitation.
In a chemical plant for a fertilizer or chemical material, the particle size distribution of a particle group being carried on a belt is measured, and the particle size distribution measurement result is used as feedback information to control the charging amount of a material or water. For this purpose, the particle recognition result must be quickly obtained, and the processing time of particle size distribution measurement is also required to be shorter. In a chemical plant for a fertilizer or chemical material, the processing time taken for particle size distribution measurement must be several sec or shorter. Because of the problem of accuracy, prior art 1 is difficult to use. Prior art 2 cannot be used because it takes several min for particle size distribution measurement due to the enormous calculation amount.
Means of Solution to the Problem
The present invention has been made to solve the above problem, and has as its object to provide a particle recognition apparatus capable of recognizing a particle at a high speed and high accuracy.
In order to achieve the above object, the first invention comprises image input means (
1
) for obtaining a halftone image (IG) of a particle group to be measured, binarization means (
2
) for binarizing the halftone image (IG) obtained by the image input means(
1
) to obtain a binary image (IB), distance transform means (
3
) for performing distance transform for the binary image (IB) binarized by the binarization means (
2
) to obtain a distance-transformed image (ID), particle nucleus extraction means (
4
) for performing particle nucleus extraction processing for the distance-transformed image (ID) obtained by the distance transform means (
3
) to obtain a particle nucleus image (IC), and particle expansion means (
5
) for performing particle expansion processing for the particle nucleus image (IC) obtained by the particle nucleus extraction means (
4
) and the distance-transformed image (ID) obtained by the distance transform means (
3
) to obtain a particle-separated image (IS).
According to this invention, the halftone image (IG) of the particle group to be measured is binarized into the binary image (IB), the binary image (IB) is subjected to distance transform to obtain the distance-transformed image (ID), and the distance-transformed image (ID) is subjected to particle nucleus extraction processing to obtain the particle nucleus image (IC), and the particle nucleus image (IC) and distance-transformed image (ID) are subjected to particle expansion processing to obtain the particle-separated image (IS).
According to the second invention, in the first invention, particle nucleus candidates are obtained from the distance-transformed image (ID), and a particle nucleus is extracted on the basis of the distance between the particle nucleus candidates.
According to this invention, the particle nucleus candidates are obtained from the distance-transformed image (ID), the particle nucleus is extracted on the basis of the distance between the particle nucleus candidates to obtain the particle nuc

Affiliated with

Also associated with

Rating

Particle recognition device does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Particle recognition device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Particle recognition device will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3241508