CN110610141A - A logistics warehousing regular shape cargo identification system - Google Patents

Abstract

The invention discloses a logistics warehousing regular shape cargo identification system, which is composed of a cargo information statistics module, an image acquisition module, an image preprocessing module, a cargo identification module and a cargo adjustment and storage module. The invention uses a machine to identify the type of goods and the way they are placed, and classifies and adjusts the goods for storage, solves the problem of time-consuming and laborious manual handling of goods, and realizes automatic storage and goods management. Shape and color features, using image processing technology, determine the category, placement and angle of the goods, and adjust the storage through the robotic arm.

Description

A logistics warehousing regular shape cargo identification system

technical field

The invention relates to the technical field of logistics sorting and detection, in particular to a logistics warehousing regular shape cargo identification system.

Background technique

Logistics warehousing is an important part of advanced manufacturing field, computer integrated manufacturing system and modern logistics technology, and plays an important role in the development of manufacturing industry. The current warehouse is still in the automation stage. One of the main problems of this type of automated warehouse is that during the picking operation, most docks or ports still need to manually identify and store goods, which is time-consuming and labor-intensive.

At present, there are three main ways to automatically identify objects in and out of the warehouse:

1. Barcode scanning. Barcode identification technology appeared earlier and is the most mature identification technology at present. It identifies relevant information by sticking a set of fixed-format codes containing cargo information on the surface of the cargo, and circulates it in the cargo box through identification equipment and identification technology. identification during the process. In non-automated logistics warehousing systems, this type of cargo identification system is widely used, and it is also used in automated three-dimensional warehouses.

However, the barcode will be damaged during use, and there will be a high bit error rate during automatic identification, and its reliability will be affected. In addition, limited by the conditions of the identification technology itself, the information contained in the barcode is single. With the continuous improvement of the automation level and work efficiency of the three-dimensional warehouse, this type of identification system can no longer meet the application requirements.

Second, RFID technology. Radio frequency identification technology (RFID) using non-contact IC card technology is a technical development direction currently applied in automated three-dimensional warehouses. It uses radio frequency identification (Radio Frequency Identification, RFID) technology, also known as electronic tag (E-Tag) technology, non-contact automatic identification technology realized by radio frequency communication. A radio frequency identification system is generally composed of two parts, namely an electronic tag (transponder, E-Tag) and a reader (reader, Reader). In the practical application of RFID, the electronic tag is attached to the identified item (surface or interior). When the identified object with the electronic tag passes through the readable range of the reader, the reader automatically does not touch The identification information in the electronic tag is read out, so as to realize the function of automatic identification of items or automatic collection of item identification information. The biggest advantage of this method lies in the non-contact working method. The whole identification process does not require manual intervention. It is suitable for automation and is not easy to damage. It can identify high-speed moving objects and identify multiple radio frequency tags at the same time. The operation is fast and convenient. It has good adaptability to harsh environments such as oil stains and dust pollution.

After searching the prior art, it is found that the Chinese patent application number: 201220505565.1, the patent name: an intelligent warehouse management system, the application scheme provides an intelligent warehouse management system provided by the present utility model, including video capture equipment , video transmission equipment, server group, RFID reader, alarm device, LED display, storage equipment, the server group includes data processing server, image processing server, streaming media transmission server, the central control server is the core of the server group, used for Unified control and sending instructions to image processing server, data processing server, streaming media transmission server, so as to realize different functions, and process the information fed back by each server. It does not involve the operation of the image, and has the following defects: RFID tags plus RFID transmitters, readers, encoders and antennas are expensive; manual labeling increases labor; Objects or the environment can have an impact on RFID. Although barcode technology and RFID technology have matured, they are not inherent properties of goods, and are very different from the way humans perceive things. Therefore, how to use the characteristics of the goods to realize the automatic identification of the goods efficiently is an urgent problem to be solved in the field of logistics technology.

3. Object image recognition method. Image recognition refers to the technology that uses computers to process, analyze and understand images to identify targets and objects in various patterns. In general industrial use, an industrial camera is used to take pictures, and then software is used for further identification and processing according to the grayscale difference of the pictures. Chinese Patent Application No. 201810642072.4, the name of the patent application is a method for identifying and classifying goods in supermarket warehouses based on target recognition and KNN algorithm. The application scheme provides a method for identifying and classifying goods in supermarket warehouses based on target recognition and KNN algorithm, including: obtaining goods Image sample data; preprocess the sample data; extract the target contour; select the minimum circumscribed rectangle of the target; calculate the average RGB value and histogram distribution of the entire area; match between the test sample and the training sample through the KNN algorithm . This method only classifies the category of items, and cannot determine the placement of objects, so it is not suitable for logistics warehousing systems. The Chinese application number 201711467294.9, the title of the patent application provides a warehouse cargo intelligent identification system based on cargo image recognition technology, including a data input module, a data storage module, an analysis and comparison module, a phase acquisition module, an output module and an abnormal alarm module. Through image recognition technology, the image of the good sample is compared with the image of the actual goods to judge whether the detected goods are correct good products. Similarly, this method only classifies the category of items, and cannot determine the placement of objects, so it is not suitable for logistics warehousing systems.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a logistics warehousing regular shape cargo identification system, which can identify the type of the cargo, judge the position and arrangement of the cargo, so that the cargo can be sorted and put into the warehouse.

The technical scheme for realizing the object of the present invention is: a logistics warehousing regular shape cargo identification system, comprising a cargo information statistics module, an image acquisition module, an image preprocessing module, a cargo identification module and a cargo adjustment and storage module;

The cargo information statistics module is used to obtain the length, width, height, radius and side length of all sample cargoes, and build a cargo information database according to the average color information;

The image acquisition module is used to collect video information and transmit it to the image preprocessing module;

The image preprocessing module is used to make the difference between each frame of the obtained picture and the background, and process it through image filtering, grayscale, binarization, erosion, expansion, edge detection, contour extraction and image filling;

The cargo identification module is used to judge the shape of the cargo according to the features, extract the color in the ROI area, and finally determine the type and placement of the cargo;

The goods adjustment and warehousing module controls the moving distance and rotation angle of the robotic arm according to the recognition results, and sorts the goods into the warehouse and stacks them neatly.

Compared with the prior art, the present invention has the following significant advantages: (1) compared with the traditional manual sorting and transporting into storage, the method of image processing by a machine greatly liberates manpower and improves the operation efficiency; (2) Compared with barcode identification, the automatic identification of goods can be efficiently realized by using the characteristics of the goods themselves, without worrying about barcode contamination; (3) Compared with RFID tag identification, no manual labeling is required, which simplifies the process and uses Single camera reduces cost.

Description of drawings

Fig. 1 is a block diagram of the logistics warehousing rule-shaped cargo identification system of the present invention.

Fig. 2 is a system block diagram of the hardware part of the system of the present invention.

Fig. 3 is the mechanical structure diagram of the image acquisition module of the present invention.

FIG. 4 is a schematic diagram of the image preprocessing module of the present invention.

FIG. 5 is a schematic diagram of the cargo identification module of the present invention.

Detailed ways

Automated warehousing cargo management can solve the time-consuming and laborious problem of manual sorting and handling in docks, ports and other places, and make warehousing more efficient. The image recognition method makes up for the shortcomings of the existing barcode scanning, RFID, and image recognition methods.

As shown in FIG. 1 , the present invention proposes a logistics warehousing regular shape cargo identification system, including a cargo information statistics module, an image acquisition module, an image preprocessing module, a cargo identification module and a cargo adjustment warehousing module;

The cargo information statistics module is used to obtain the length, width, height, radius, and side length of all sample cargoes, and establish a cargo information database based on the average color information; in order to compare with the characteristics of the processed test cargo, so as to obtain the type and arrangement of the test cargo .

The image acquisition module is used to collect video information and transmit it to the image preprocessing module; image acquisition is the process of acquiring scene images from the work site. Use a CCD camera with the advantages of high sensitivity, strong light resistance, small distortion, small size, long life, and anti-vibration to collect continuous on-site images, and digitize the analog image signals to the image processing module.

The image preprocessing module is used to make the difference between each frame of the obtained picture and the background, and undergo image filtering, grayscale, binarization, erosion, expansion, edge detection, contour extraction and image filling processing;

The cargo identification module is used to judge the shape of the cargo according to the features, then extract the color in the ROI area, and finally determine the type and placement of the cargo;

The goods adjustment and warehousing module controls the moving distance and rotation angle of the robotic arm according to the recognition results, and sorts the goods into the warehouse and stacks them neatly.

The method of obtaining the average color information is: take multiple photos of the goods, convert them to the LAB color space, select the area where the goods are located, perform three-channel color histogram statistics, and obtain the average color of the goods.

The image acquisition module includes a front-end camera, a transmission line and a memory. The front-end camera is connected to the industrial computer through the transmission line for collecting video signals, and the industrial computer is connected to the image preprocessing module for converting the acquired analog video signal into digital. The video sequence is transmitted to the image preprocessing module; the mechanical structure of the image acquisition module is shown in Figure 3. The camera is fixedly installed directly above the conveyor belt, and the fill light is used to reduce the influence of light.

The image preprocessing includes grayscale, image filtering, background subtraction, binarization, erosion, expansion, and image filling; wherein: acquiring the background image and the current image from the memory, and performing grayscale, image filtering, and background subtraction in sequence , binarization, erosion, expansion, image filling processing, the image preprocessing module is connected with the cargo identification module, and the processing process is shown in Figure 4.

The goods identification module includes edge detection, shape detection, outline drawing, color analysis module of interest area and goods information module of comparison database. Among them: the shape detection uses the Laplace edge detection algorithm to calculate the perimeter of the graph and the number of vertices. First determine whether the geometric figure is a circle, that is, determine whether the ratio of the square of the perimeter of the figure to the area is within the specified range, otherwise, determine the number of vertices, equal to 3 is a triangle, and equal to 4 is a rectangle. If it is a rectangle, according to the ratio of length to width, you can continue to subdivide the types of goods, and you can judge whether the goods are placed on the front or side according to this ratio. Outputs the perimeter and area of a geometric figure while identifying the shape of the figure. After the shape is detected, the outline is drawn, the interior is set as the region of interest, and the color is analyzed. Finally, the type and placement of the goods are determined through shape detection and color detection. The specific process is shown in Figure 5.

The adjustment of goods into storage The adjustment of goods into storage is to control the moving distance and rotation angle of the robotic arm according to the recognition result, and classify the goods into storage and stack them neatly. Among them: the hardware part of the whole system of goods adjustment and storage is mainly composed of man-machine interface, PLC control system, camera, drive unit, servo system, shelf position and mechanical parts. Through the man-machine interface, the programmable controller can monitor and write data to monitor the warehouse, identify the goods and extract the goods.

The system of the present invention will be further described below with reference to the accompanying drawings.

Example

A logistics warehousing rule cargo identification system, comprising five parts: a cargo information statistics module, an image acquisition module, an image preprocessing module, a cargo identification module and a cargo adjustment and warehousing module;

The cargo information statistics module is used to obtain the length, width, height, radius, side length, and average color information of all sample cargoes to build a cargo information database. In order to compare with the characteristics of the processed test goods, so as to obtain the type and arrangement of the test goods. To obtain the average color information, you need to take multiple photos of the goods, convert them to the LAB color space, select the area where the goods are located, and perform three-channel color histogram statistics to obtain the average color of the goods.

The image acquisition module receives the analog video signal and digitizes it through A/D, or directly receives the digital video data after the digitization of the camera. The image acquisition part stores the digital image in the memory of the industrial computer. The image acquisition module reads the current frame every 30ms according to the preset program and delay, thereby obtaining continuous video signals. Take 5 background photos with no goods on the conveyor belt and take the average to get the average background image.

Image preprocessing module: For the collected digital images, due to the influence of equipment and environmental factors, they are often subject to different degrees of interference, such as noise, geometric deformation, color misalignment, etc., which will affect the accuracy. For this purpose, the acquired images must be preprocessed. The preprocessing process includes grayscale, image filtering, background subtraction, binarization, erosion, dilation, and image filling.

Grayscale is the process of obtaining a single-channel grayscale image of an RGB color image through a weighted average method. The average background picture and the current frame are grayed and then Gaussian filtering is performed, and the difference map is obtained by subtracting the processed average background picture and the current frame by background subtraction. Set an appropriate threshold, perform corrosion and expansion processing on the difference image after binarization, eliminate noise, make the edge clearer, and obtain a closed boundary, but there are still large and small holes inside, and the hole filling method is more effective. Complete picture of goods.

The cargo identification module is edge detection, judges the shape of the cargo according to the features, and then extracts the color in the ROI area after the outline is drawn, and finally determines the type and arrangement of the cargo.

Shape detection:

(1) Circle: Circle is the most special kind of all regular geometric figures. The ratio of the square of the perimeter to the area fluctuates around 4π, and the ratio of the radius has no effect, so as long as the identified figure satisfies this The condition is a circle.

(2) Rectangle: There are 4 vertices, as long as the boundary points of the detected figure are judged and the number of vertices is calculated, if the number is equal to 4, it is a rectangle; otherwise, it is other geometric figures. Similar to the recognition of circles, the phenomenon that the number of vertices is not 4 may also occur in the actual detection process of the rectangle. It is necessary to perform a "same neighborhood" judgment on the detected vertices to see if some of them are in the same neighborhood. The vertices belong to the same neighborhood range. In this embodiment, there are 8 neighborhoods. If there are, only one is left and the others are removed, so that the result can be well corrected and the occurrence of false detection can be avoided. Using this method, the detected graphics can be well identified as rectangles.

(3) Triangle: The identification method of triangle is similar to that of rectangle, and the number of vertices is also used to determine the geometric shape to which the figure belongs.

The Laplace edge detection algorithm is used to calculate the perimeter of the graph and the number of vertices. After the geometric figure is detected by the Laplace edge detection algorithm, the cumulative number of boundary points is the perimeter of the geometric figure. These boundary points are then further distinguished to find vertices. After the image preprocessing and feature extraction are completed, the shape of the geometric figure can be judged according to the feature information. In this process, first determine whether the geometric figure is a circle, that is, determine whether the ratio of the square of the perimeter of the figure to the area is within the specified range. If it is, it is a circle; otherwise, the number of vertices is judged, equal to 3 is a triangle, and equal to 4 is a rectangle. While recognizing the shape of the geometric figure, it can also output the perimeter and area of the figure. Because the perimeter, area and vertex coordinates of regular geometric figures are known, parameters such as side length and centroid can be calculated.

Color detection:

Output the contour sequence, correspond to the current frame, form the ROI area, calculate the three-channel average (x, y, z) of the RGB images in the area, compare with the sample cargo information database, find the most likely cargo type, and use it as an auxiliary judge.

Placement:

For rectangular goods, according to the ratio of length to width, determine whether the goods are placed on the front or side.

The adjustment of the goods into the warehouse is to control the moving distance and rotation angle of the robotic arm according to the recognition result, and classify the goods into the warehouse and stack them neatly.

As shown in Figure 2, the hardware part of the whole system is mainly composed of man-machine interface, PLC control system, camera, drive unit, servo system, shelf position and mechanical parts. Through the man-machine interface, the programmable controller can monitor and write data to monitor the warehouse, identify the goods and extract the goods. Click the corresponding screen to display the corresponding interface, and it is also more convenient to enter the length, width, height, radius, side length, and average color information of the sample cargo in the manifest.

The image acquisition part is shown in Figure 3. A camera is fixed above the conveyor belt, and there is a circle of fill lights around the camera. When it is sent to the bottom of the camera, the image of the goods will appear in the picture captured by the camera, and the recognition module will identify the type and placement of the goods.

The image preprocessing part is shown in Figure 4, including grayscale, background subtraction, image filtering, binarization, erosion, dilation, and image filling. Obtain the current frame and the average background image, grayscale and Gaussian filtering, respectively, and then subtract the difference image. After binarization, corrosion and expansion processing are performed to eliminate noise, make the edge clearer, and obtain a closed image. Boundary, but there are still large and small holes inside, use the hole filling method to fill the internal holes.

The cargo identification part is shown in Figure 5. First, the edge of the preprocessed image is detected, the contour is extracted, the type of the cargo is judged according to the shape and color, and the placement of the cargo is judged according to the contour information.

The invention designs a logistics warehousing regular shape cargo identification system, which can well judge the type and arrangement of the cargo on the conveyor belt. The system greatly liberates manpower and improves efficiency; compared with barcode recognition, the automatic identification of goods is efficiently realized by using the characteristics of the goods themselves, without worrying about barcode contamination; compared with RFID tag identification, no manual Labeling simplifies the process, and the use of a single camera reduces costs and has good market prospects.

Claims (6)

1. A logistics warehousing rule-shaped cargo identification system, characterized in that it comprises a cargo information statistics module, an image acquisition module, an image preprocessing module, a cargo identification module and a cargo adjustment warehousing module; The cargo information statistics module is used to obtain the length, width, height, radius and side length of all sample cargoes, and build a cargo information database according to the average color information; The image acquisition module is used to collect video information and transmit it to the image preprocessing module; The image preprocessing module is used to make the difference between each frame of the obtained picture and the background, and process it through image filtering, grayscale, binarization, erosion, expansion, edge detection, contour extraction and image filling; The cargo identification module is used to judge the shape of the cargo according to the features, extract the color in the ROI area, and finally determine the type and placement of the cargo; The goods adjustment and warehousing module controls the moving distance and rotation angle of the robotic arm according to the recognition results, and sorts the goods into the warehouse and stacks them neatly. 2. The logistics warehousing rule-shaped cargo identification system according to claim 1, wherein the average color information acquisition method is: taking a plurality of photos of the cargo, converting to LAB color space, selecting the area where the cargo is located, and performing three-channel coloring. Histogram statistics, find the average color of the goods. 3. The logistics warehousing rule-shaped cargo identification system according to claim 1, wherein the image acquisition module comprises a front-end camera, a transmission line and a memory, and the front-end camera is fixedly installed just above the conveyor belt, and communicates with the industrial computer through the transmission line. Connected to the video signal, the industrial computer is connected to the image preprocessing module, and is used to convert the acquired analog video signal into a digital video sequence and transmit it to the image preprocessing module. 4. The logistics warehousing rule-shaped cargo identification system according to claim 1, characterized in that: the image preprocessing module obtains the background image and the current image from the memory, and sequentially performs grayscale, image filtering, background subtraction, two Value, erosion, dilation and image fill processing. 5. The logistics warehousing rule-shaped cargo identification system according to claim 1, wherein the cargo identification module adopts the Laplace edge detection algorithm to calculate the figure perimeter and the number of vertices, and first judges whether the geometric figure is a circle, i.e. Determine whether the ratio of the square of the perimeter of the figure to the area is within the specified range, otherwise, determine the number of vertices, equal to 3 is a triangle, equal to 4 is a rectangle; if it is a rectangle, according to the ratio of length to width, continue Subdivide the types of goods; output the perimeter and area of the graphic while recognizing the shape of the geometric figure; draw the outline after shape detection, set the interior as the area of interest, and analyze the color; finally determine the type of goods and pendulum through shape detection and color detection put way. 6. The logistics warehousing rule-shaped cargo identification system according to claim 5, characterized in that, according to the ratio of the length and width of the rectangle, it is judged whether the cargo is placed on the front or on the side.