JP2002207994A - Method for processing object image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an object image processing method capable of color identification without being almost affected by a cause of deteriorating color identification accuracy, e.g. fluctuations of illumination and luminance levels. SOLUTION: A place in a state where there is no glass bottle G, etc., is picked up to fetch the background image of the glass bottle G, etc., just before or after the glass bottle G, etc., carried to the prescribed position by a carrying conveyor 1 is picked up by a CCD camera 2. Meanwhile, when the glass bottle G, etc., is carried to the prescribed position, the glass bottle G, etc., is picked up to fetch the image of the glass bottle G, etc. These both images are respectively supplied as RGB signals from the image pickup element of the CCD camera 2 to a color identifying device, the color identifying device performs differential processing between pixels constituting respectively the both images, the shape of the glass bottle G, etc., is extracted from the differential information between the images, and the glass bottle G, etc., is also subjected to color identification from a color information of a proper part.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method for identifying an object, a glass bottle or a fragment thereof (hereinafter, referred to as a glass bottle, etc.), and further identifying and collecting the color of the object.

[0002]

2. Description of the Related Art In general, when a color of a glass bottle is identified, a fluorescent lamp or the like is applied to the glass bottle to take an image of the glass bottle with, for example, a CCD camera or the like. The signals are converted into B signals, and color identification is performed by grayscale image processing for each of these signals. However, there is no example of trying to collect the fragments by color identification.

[0003]

When a fluorescent lamp or the like is used as a light source for irradiation, since the illuminance and the luminance level fluctuate due to, for example, changes in the ambient temperature and aging of the lighting equipment such as the fluorescent lamp, the accuracy of the light source is reduced. Image processing to extract the shape to use the color information of the appropriate part
A disadvantage occurs in that the color identification accuracy is reduced. One way to deal with this inconvenience is to maintain the temperature at a constant temperature, for example, so as not to cause temperature changes. There is a problem from a viewpoint. As another method for coping with the temperature change, for example, Japanese Patent Laid-Open No.
There is a method disclosed in US Pat.
I = (R + G + B) / 3, r = R / I, g =
Assuming that the I, r, g parameters of G / I (I: luminance, r: red component ratio, g: green component ratio) change, the I, r, g reference ranges at the time of initial color table setting When the temperature changes, the I, r, and g reference ranges are reset to cope with the change. However, there are various causes for lowering the color identification accuracy, such as not only the temperature change but also the aging of the lighting equipment as described above, and resetting the reference range in response to all of these causes is an extremely complicated operation. is there. Furthermore, this method is for a glass bottle, and does not describe how to determine a portion from which color information is to be extracted by determining a glass bottle fragment in a line in which glass bottle fragments far smaller than the glass bottle are mixed and flowing.

SUMMARY OF THE INVENTION It is an object of the present invention to accurately recognize the shape of an object without being largely affected by a cause of a decrease in color identification accuracy, for example, a change in illuminance or luminance level, and to provide color information of an appropriate portion for color identification. An object of the present invention is to provide a method of acquiring and performing image processing.

[0005]

According to a first aspect of the present invention, there is provided a method for identifying a color of an object, comprising the steps of: Before or immediately after the image of the object is captured, the background image of the object is captured, and the shape of the object is extracted from the difference information between images obtained by subtracting the brightness of the two images. The color of the object is identified based on the color information. Therefore, at the time of such color identification, since the shape is extracted by the difference from the luminance of the immediately preceding or succeeding image, for example, the shape is hardly affected by the aging of the lighting equipment as well as the fluctuation of the illuminance and the luminance level. , And color information of an appropriate portion is obtained to enable accurate color identification.

[0006]

FIG. 1 shows a first embodiment of the present invention.
This will be described with reference to FIGS. The color identification method according to the present embodiment is used in a recycling system for identifying and collecting glass bottles (articles) and the like (articles) conveyed on a conveyance line. The present color identification method will be described with reference to FIGS. At a predetermined position on the conveyor 1, for example, a color C
A CD camera 2 is installed, and a signal from an image sensor of the CCD camera 2, that is, an RGB signal is supplied to a color identification device (not shown). An illuminator (not shown) such as a fluorescent lamp for irradiating the above-described predetermined position is provided, and the illuminator irradiates an (article) G such as a glass bottle. However, there are various illumination methods, such as diffuse illumination, transmitted illumination, and oblique illumination. According to the present color identification method, accurate color identification can be performed by any of the illumination methods.

A color identification procedure according to the present color identification method will be described with reference to FIG. However, FIG. 1A shows a glass bottle G or the like.
FIG. 1B is a front view and a plan view immediately before being conveyed to a predetermined position, and FIG. 1B is a front view and a plan view when a glass bottle G or the like is conveyed to a predetermined position. Immediately before the CCD camera 2 captures an image of the glass bottle G or the like conveyed to a predetermined position by the conveyor 1, an image of a state where the glass bottle G or the like is not present (a state where the glass bottle G is not conveyed to a predetermined position) is taken. The background image of the glass bottle G or the like is fetched. Then, when the glass bottle G or the like is conveyed to a predetermined position, an image is taken and the image of the glass bottle G is taken in. Needless to say, the image captured at this time is an image in which the background is captured in addition to the glass bottle G or the like. These two images are respectively supplied to the color identification device as RGB signals from the imaging device of the CCD camera 2. Then, in the color identification device, I =
The luminance of each pixel is calculated in accordance with 0.3R + 0.6G + 0.1B, and the luminance I _{iREF of} each pixel in a predetermined area when there is no glass bottle G and the luminance I _{i of} each pixel in a predetermined area when there is a glass bottle G etc. Desired. ΔI _{i for} each pixel
= I _iREF -I _i is obtained, and when the value exceeds a predetermined value and when it is not, a binarization process is performed to extract the shape and compare it with the stored shape data to determine the glass bottle G or the like. At the same time, the color is identified by the color information of an appropriate part (for example, the center) in the extracted shape.

The luminance of each pixel in the case where there is no glass bottle G is not only the newly acquired one, but also the average value of the luminance of each pixel which was previously taken in and without the glass bottle G etc. (I _iREF + I _i-1REF )
May be the luminance when there is no glass bottle G or the like in the color identification of the glass bottle G or the like. This is a more suitable method when a process of capturing a background image immediately after capturing an image of a glass bottle G or the like is performed. Also, when a background image is captured, the position of the belt is determined by binarizing the image, and color identification is performed excluding that portion when performing color identification.

As described above, in the present color identification method, the background image immediately before or immediately after the image of the glass bottle G or the like is captured, an image of the case where the glass bottle G or the like is present is captured, and the difference between the luminances of the two is obtained for each pixel. G for binarization, etc.
Since the shape is extracted and the color is identified from the appropriate part in the extracted shape, the shape can be extracted without being affected by, for example, the aging of the lighting equipment as well as the fluctuation of the illuminance and the brightness level, and accurate color identification can be performed. Will be possible. In fact, FIG.
As shown in FIG. 7, the initial illuminance level decreases with the passage of time. For example, the R, G,
Comparing the R, G, and B components of the glass bottle G or the like irradiated at the illuminance level Q with reference to the B component, accurate image processing cannot be performed particularly with a white glass bottle G or the like. The discrimination and the color discrimination accuracy based thereon are reduced. However, if the image processing is performed based on the illuminance level immediately before or immediately after the image of the glass bottle G or the like is taken (in FIG. If the image processing is performed on the basis of the illuminance level M immediately before the image is taken in), accurate image processing can be performed with almost no influence of the fluctuation of the illuminance level, so that the color identification accuracy does not decrease.

[0010]

According to the object image processing method of the present invention,
The shape of an object can be extracted without being largely affected by a cause of a decrease in color identification accuracy, for example, a change in illuminance or luminance level, and the shape can be accurately identified and the color can be identified.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of image capture in an image processing method according to an embodiment of the present invention.

FIG. 2 is a transition diagram of an illuminance level over time.

[Explanation of symbols]

 1 Conveyor 2 CCD camera G Glass bottle etc.

Continuing on the front page (72) Inventor Akira Ohashi 1-7-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Inside Fuji Heavy Industries Ltd. (72) Inventor Takao Okada 5-6-1, Ikegami, Ota-ku, Tokyo Shinsuke Ikegami Incorporated (72) Inventor Shunji Ueno 5-6-1 Ikegami Ota-ku, Tokyo Ikegami Tsushin Fuki Co., Ltd. F-term (reference) 5B057 AA02 BA02 DA08 DA12 DB02 DB06 DB09 DC09 DC25 DC32

Claims (1)

[Claims] In a color identification method for capturing an image of an illuminated object and identifying the color of the object, a background image of the object is captured immediately before or immediately after capturing the image of the object, and the luminance of both images is determined. An image processing method for an object, characterized in that the shape of the object is extracted from the inter-image difference information obtained by subtracting the shape of the object, and a part for identifying a color is determined.