JP6845839B2 - Container number recognition system - Google Patents

Description

The present invention relates to a system for image recognition of a container number, and more particularly to a container number recognition system suitable for grasping a container number when unloading or unloading a container.

When unloading the container, the inspector visually confirms the container number on the ship side and inputs it to the handy terminal to check whether it matches the container number based on the list sent from the shipping company in advance. It is done. However, due to the visual confirmation by the inspector, mistakes such as misunderstandings and input mistakes occur due to human intervention.

On the gate side, a technique of photographing the container number at the rear part (door part) of the container with a camera and recognizing the number as disclosed in Patent Document 1 has already been put into practical use. In recent years, such automatic container number reading technology is being applied to the unloading side of containers, and attempts are being made to install a camera on a gantry crane to read the container number. Specifically, a camera is attached to a sill beam that constitutes a gantry crane, and the number on the side surface of the container is read using this camera.

The container number is also attached to the container ceiling, but the container ceiling is more susceptible to damage due to wind and rain or contact during cargo handling compared to the side of the container (including the rear), and part of the character string is part of the character string. Since it may be missing and it is easy for recognition failure to occur, the character string on the side of the container is exclusively selected for the automatic recognition of the character string for the container number.

Japanese Unexamined Patent Publication No. 2002-37453

If the automatic character string recognition technology is adopted as in the above technology, it is possible to avoid human error such as misunderstanding or input error. However, the container number attached to the side surface of the container (not including the rear part) is often written vertically. Currently, the popular character recognition system supports horizontal characters, so when recognizing a vertically written character string, it is necessary to cut out the character area one character at a time.

However, the acquired image may not be suitable for recognizing the character area because the container in the distribution process may be dirty or not always in a stationary state, so that the characters can be cut out accurately. This cannot be done and may lead to a decrease in the recognition rate. For this reason, by installing a separate camera on the trolley and spreader to acquire and recognize the image of the container number on the ceiling of the container, it is possible to check the consistency with the container number read from the side of the container and read defects. It was necessary to supplement the part where it occurred.

Therefore, an object of the present invention is to provide a container number recognition system capable of increasing the recognition rate of container numbers without increasing the number of installed cameras in the container handling field.

The container number recognition system according to the present invention for achieving the above object is arranged in a portal tie beam constituting a gantry crane frame, and is a container door portion mounted on a container transport vehicle that handles container cargo in the frame. And the image data captured by the image pickup means and the image data of the ceiling portion are acquired, and the aspect ratio of the acquired image data is changed to create the aspect ratio conversion image data, which is used as the aspect ratio conversion image data. get the character information included, have a, a calculating means for managing the character information as a container number, the calculating means, wherein the vertical distance L4 to the container number written part of the door part from the portal tie beam The value divided by the horizontal distance L2 from directly below the image pickup element in the image pickup means to the rear end of the container transport vehicle is used as the correction ratio, and is multiplied as the vertical magnification when changing the aspect ratio of the image data. The aspect ratio conversion image data is acquired, and both the container number attached to the door portion and the container number attached to the ceiling portion are acquired from the aspect ratio conversion image data . Most of the container numbers on the container door are written horizontally. Therefore, by cutting out the entire container number part as a character string and recognizing it, it is possible to prevent a decrease in the recognition rate due to cutting out the characters that make up the container number one by one, as in the case of container number recognition on the side of the container. Become.

Further, in the container number recognition system having the above-mentioned characteristics, the calculation means compares the container number attached to the door portion with the container number attached to the ceiling portion, and collates and complements the container numbers. It is good to do so. By having such a feature, it is possible to inquire and complement the container numbers of the ceiling part of the container and the door part of the container obtained by one imaging, and it is possible to improve the recognition efficiency of the container number. it can.

According to the container number recognition system having the above-mentioned characteristics, it is possible to increase the recognition rate of the container number in the container handling place without increasing the number of installed cameras.

It is a figure which shows the structure including the arrangement of the container number recognition apparatus which concerns on 1st Embodiment. It is a figure which shows the structure of the gantry crane which arranges the imaging means which concerns on 1st Embodiment. It is a figure which shows an example of the acquired image data and the corrected image data in 1st Embodiment. It is a figure which shows an example of the acquired image data and the corrected image data in 2nd Embodiment.

Hereinafter, embodiments of the container number recognition system of the present invention will be described in detail with reference to the drawings. It should be noted that the embodiments shown below are a part of the embodiments suitable for carrying out the present invention, and even if a part of the configuration is changed within a range that does not affect the effect, the embodiment of the present invention is carried out. Can be regarded as.

[First Embodiment]
First, the container number recognition device according to the first embodiment will be described with reference to FIGS. 1 to 3. In the drawings, FIG. 1 is a diagram showing a configuration including the arrangement of the container number recognition device according to the first embodiment, and FIG. 2 is a configuration of a gantry crane in which the imaging means according to the first embodiment is arranged. It is a figure which shows. Further, FIG. 3 is a diagram showing an example of the acquired image data and the corrected image data in the first embodiment. The container number recognition system 10 according to the present embodiment is a system for automatically recognizing a container number at the time of container handling for a ship (not shown). The configuration is based on the imaging means 12 and the calculation means 14.

[Imaging means]
The imaging means 12 is provided in the gantry crane 20 that handles cargo with the ship. Specifically, the portal tie beam 22 constituting the frame of the gantry crane 20 is arranged at an angle so as to image an obliquely lower side. By adopting the imaging means 12 in such an arrangement form, it is possible to image the rear side surface of the container 50 mounted on the container transport vehicle 30 stopped in the frame for loading and unloading the container 50, that is, the door portion of the container. It will be possible.

Here, the portal tie beam 22 exists at a higher place than the horizontal beam provided for the gate or the like in the container terminal or the sill beam 24 in the gantry crane 20, which is the arrangement position of the container number recognition camera described in the section of the prior art. At the same time, the horizontal distance (distance L2) from the container transport vehicle 30 that is stopped for cargo handling is short. Therefore, the container number appearing in the captured image is in a crushed state as shown in FIG. 3A, and is difficult to recognize as a character string.

[Calculation means]
The calculation means 14 according to the present embodiment corrects the image (acquired image data) acquired via the imaging means 12, acquires character information from the corrected image data (aspect ratio converted image data), and records it as a container number. , Takes the role of managing. The location of the calculation means 14 is not particularly limited.

The correction of the acquired image data by the calculation means 14 may be performed as follows. First, as prior information, the distance between the portal tie beam 22 on which the image pickup means 12 is arranged and the ground (in the case of improving the accuracy, the distance from the image sensor to the ground in the image pickup means 12: distance L1) and the portal tie beam 22 The horizontal distance from directly below to the rear end of the container transport vehicle (in the case of improving accuracy, the horizontal distance from directly below the image sensor in the image sensor 12 to the rear end of the container transport vehicle: distance L2), and the container number from the ground. Enter the approximate distance (distance L3) to the description part A. By subtracting the distance L3 from the distance L1 based on this information, the approximate distance (distance L4) from the portal tie beam 22 to the container number description unit A is obtained.

Next, by dividing the distance L4 by the distance L2, the ratio (correction ratio) of the vertical distance and the horizontal distance from the portal tie beam 22 to the container number description portion A is obtained. After that, as the correction value of the aspect ratio of the image data (acquired image data) acquired via the imaging means 12, the correction image data is acquired by multiplying the magnification in the vertical direction by the correction ratio. As an example, the corrected image data based on the acquired image data shown in FIG. 3 (A) is as shown in FIG. 3 (B).

As described above, the imaging means 12 is arranged at an angle with respect to the portal tie beam 22, and the angle θ is relative to the perpendicular line starting from the portal tie beam 22 (mounting position of the imaging means 12). It can be shown as the angle (angle θ) of the axis C that is the center of imaging. Then, in the present embodiment, the imaging means 12 is angled so that the axis C, which is the center of imaging, faces the container number description portion A at the door portion of the container 50. Therefore, the smaller the angle θ, the larger the correction ratio.

The character information may be acquired by the calculation means 12 by cutting out a character area from the corrected image data and using known character recognition software or the like. Here, the character area may be cut out by cutting out the entire character string including the container number, which is a horizontal character, as the character area. By doing so, the number of times the characters are cut out is reduced as compared with the vertical container number recognition technology that requires cutting out the characters one by one. Therefore, it is possible to prevent a decrease in the recognition rate due to the inability to accurately cut out the character area. Further, after cutting out the character area, filtering processing may be performed on the cut out data to improve the recognition efficiency of the container number. As an example of the filtering process, a process of determining that the recognition unit, which is extremely smaller than the average size of the characters constituting the container number, is dirty or the like and deleting it from the determination information can be mentioned.

The character information acquired based on the corrected image data is recorded in the storage unit as a container number, and is linked to various information and managed.

[effect]
According to the container number recognition system 10 as described above, it is possible to increase the recognition rate of the container number in the container handling place without increasing the number of imaging means (cameras).

[Second Embodiment]
Next, a second embodiment according to the container number recognition system 10 of the present invention will be described. The configuration of the container number recognition system 10 according to the present embodiment is the same as that of the container number recognition system 10 according to the first embodiment described above. Therefore, for the description of each configuration, FIG. 1 will be referred to, and detailed description will be omitted.

[Characteristic part]
The difference between the container number recognition system 10 according to the present embodiment and the container number recognition system 10 according to the first embodiment is the number of acquired information and the recognition efficiency of the container number. Specifically, the imaging range of the imaging means 12 includes not only the door portion but also the ceiling portion, and the image data is acquired. The details of the imaging range on the ceiling side are not limited, but may be generally the range from the center in the longitudinal direction of the container to the door side (closer to the door).

Then, when imaging is performed in the above range as the acquired image data, the calculation means 14 is attached to the container ceiling portion in the imaging range of the imaging means 12 in addition to the container number attached to the door portion. The container number is also acquired as character information. Then, the calculation means 14 compares, collates, and complements the character information (door portion) acquired based on the corrected image data with the character information (ceiling portion) acquired from the acquired image data, thereby recognizing the container number efficiency. To improve.

[effect]
According to the container number recognition system 10 having such a configuration, it is possible to read the container numbers assigned to both the door portion and the ceiling portion without increasing the number of the imaging means 12. Therefore, the recognition efficiency of the container number can be improved. Further, as shown in FIG. 4 (A), the door container number, which is difficult to recognize as character information in the acquired image data, can be converted into characters by creating the corrected image data as shown in FIG. 4 (B). It can be recognized as information. On the other hand, since the container number attached to the ceiling can be recognized as character information even in the corrected image data, the container number attached to the ceiling may be acquired from the corrected image data. By doing so, by creating the corrected image data, it is possible to acquire the container number attached to the ceiling portion and the container number attached to the door portion at one time.

In each of the above embodiments, the acquisition of the container number by the imaging means 12 has been described on the premise that the container transport vehicle 30 equipped with the container 50 is imaged from directly behind and above. Further, for simplification of the explanation, in the actual acquired image, an object (container 50 or container number) whose width is narrower as the portion is farther from the imaging means is described as having no influence of perspective in the width direction. This is because such distortion can be output as acquired image data after being corrected in advance as distortion that occurs on the center side and the edge side of the lens on the camera system.

Therefore, even when the container transport vehicle 30 is stopped at a position deviated from directly below the image pickup means 12 in the width direction, it is possible to support the acquisition of image data. That is, even if the container reflected in the captured image data has distortion in a trapezoidal shape or a rhombus, the container number can be recognized by converting the aspect ratio after correcting the distortion. ..

Further, in the present invention, in order to improve the recognition rate of the container number attached to the container 50, the image pickup means 12 is used to take an image of the door portion having a high probability that the container number is attached horizontally. There is. Therefore, it has a technical feature that the ratio on the vertical side of the aspect ratio of the acquired image data is increased so that the crushed characters can be discriminated. Here, the character discrimination enablement by correcting the crushed character by changing the aspect ratio is not affected by the writing direction of the character string. That is, even if the container number attached to the door portion is written vertically, the aspect ratio conversion (creation of corrected image data) by the container number recognition device 10 according to the present application can be performed. Therefore, as a matter of course, the container number recognition device 10 according to the present invention can also read the vertically written container number.

10 ……… Container number recognition device, 12 ……… Imaging means, 14 ……… Calculation means, 20 ……… Gantry crane, 22 ……… Portal tie beam, 24 ……… Silbeam, 30 ……… Container transportation Vehicle, 50 ……… Container.

Claims (2)

An imaging means that is placed on the portal tie beam that constitutes the frame of the gantry crane and images the door and ceiling of the container mounted on the container transport vehicle that handles the container in the frame.
Image data captured by the imaging means is acquired, the aspect ratio of the acquired image data is changed to create aspect ratio converted image data, character information included in the aspect ratio converted image data is acquired, and the characters possess a calculating means for managing the information as a container number, and
The calculation means divides the vertical distance L4 from the portal tie beam to the container number description portion in the door portion by the horizontal distance L2 from directly below the image pickup element in the image pickup means to the rear end of the container transport vehicle. As a correction ratio, when changing the aspect ratio of the image data, the aspect ratio converted image data is acquired by multiplying it as a magnification in the vertical direction.
A container number recognition system characterized in that both the container number attached to the door portion and the container number attached to the ceiling portion are acquired from the aspect ratio conversion image data. The container number recognition according to claim 1 , wherein the calculation means compares the container number attached to the door portion with the container number attached to the ceiling portion, and collates and complements the container numbers. system.

Images