فارسی
machine vision systems will be reviewed. These items include:
The best image processing software to implement Object Detection HALCON image processing software . This software is made by the German company MVtec, in which object detection and recognition is called “Matching”. Also, the photos of this article are from ready examples of image processing with HALCON software. To see more examples in detecting and identifying objects, you can download HALCON software download and install and search for the desired example (you can learn how to search for desired examples in HALCON software in the article “ 1000 image processing projects with ready source codes“
Click here to register for “Halcon Software Car Vision Training Course”.
1- Using Object Detection to announce the coordinates of the part to the robot
One of the applications of object recognition in machine vision is to announce the coordinates of the part to the robot. With the help of this algorithm, we can find the part and announce its coordinates to the robot. Let’s break it down a little more. By detecting objects, we can find the coordinates of the piece. After determining the coordinates of the part, it is enough to announce it to the robot. So-called pick and place systems do this. As an example of this application, suppose the robot wants to remove some clamps. In this way, in each step, the Object Detection algorithm must recognize the clip and announce its coordinates. As can be seen in the photos below, this operation continues until the last clip is removed.
2- Application of Object Detection to measure objects
Another application of object detection in machine vision is in measurement. To measure, we often need to find the object first. After finding the object, we can measure it. Object recognition algorithm is used to find the object. For example, in the figure below, we want to examine the razor blades. As shown in the photo with red rectangles, some teeth may be broken. To identify, we need to first determine the location of the ribs according to the figure (blue rectangle). After determining the developmentWe will examine the blades instead of the teeth. To check the blades, it is done by measuring the distance between the edges. The programming steps are described below.
Training stage
Here too, the training stage is similar to the previous one. In order to be able to use the wristband algorithm (identification of objects) in machine vision, we must first make a model of it. Therefore, in the first step, we make a matching model including the top and bottom of the blade according to the photo.
Step of checking the blades
In this step, we want to check the blades using the object detection algorithm. To check the blades, we need to go through the following steps.
First step: we find the top and bottom of the blade.
Second step: Using the coordinates obtained from the top and bottom of the blades, we can determine the yellow rectangle.
Third step: In the yellow rectangle, we check the distance between the teeth. If this distance is too long, we declare it as an error.
Although we are satisfied with mentioning this example here, several articles can be written about the applications of Object Detection in machine vision for measurement.
3-Use of object identification in print inspection
Another application of object recognition in machine vision is in the quality control of printing systems. In print checking, we need to find it before checking the print. In other words, in print checking algorithms, the position of the printed sample should always be fixed. To use the object detection algorithm in the print check, we must find the print using Object Detection. In the next step, you can check the accuracy of the print by using print identification algorithms. For example, suppose we want to check the print in the following photos. To check the accuracy of printing, we must do the following steps.
First step: make a matching model from one of the prints. This step is done only once.
Second step: Look for the matching model in different photos.
The third step: After finding the matching model, use print checking algorithms to find the correctness of the print.
As can be seen in the photos, print checking is another application of object identification in machine vision, which has always attracted the attention of the owners of this industry.
4-Use of Object Detection in identifying different models
Among other applications of object recognition in machine vision, we can mention the identification of different models. We can identify different models of each other with the help of object identification. As an example of this situation, suppose we want to identify three different parts from each other. For this purpose, it is enough:
- Define a model for object recognition algorithm for each of the parts.
- Look for the models in the photos; When the models are found, the type and number of each model will be determined.
This mode is shown in the following photos.
5- Identifying and checking the position using Object Detection
As one of the most important applications of object identification in machine vision and image processing, we can mention the identification and checking of the position of objects; We can detect the position and compare its difference with the reference position and extract the error. For example, in printing the label on the shampoo, the label must be placed on a certain position of the shampoo. For this, we can use the Object Detection algorithm. It is enough to do as follows:
- We define the label model when it is engraved in the correct position on the shampoo as the matching reference model.
- We save the coordinates and angle of the model (coordinates mean the position of the central point of the rectangle in which the model is placed; for example, the position of the central point of the model is equal to … and …)
- In the photos, we find the coordinates of the label with the help of matching algorithm.
- Now we can easily calculate the movement of the label relative to its reference position in the vertical and horizontal directions, for this it is enoughCompare the current coordinates of the model (referring to the model coordinates of the center point of the model in the horizontal and vertical directions) with the reference coordinates and declare their difference as an error. We can also identify the angle difference with respect to the reference.
Identifying objects in machine vision to check and control the position has always been the concern of industrial owners. Similarly, in different industries, we can do the same procedure to check the position of different products.
6- Application of the object identification algorithm in the completeness of the products
The accuracy of the last example of Object Detection applications in machine vision, which we will discuss in this article. Since the purpose of identifying objects is to find an object, then it can be used to count a number of objects, and from counting the number, you can find out the accuracy of the completeness. To clarify this situation, pay attention to the example below, in which each package of cheese contains 4 pieces of cream cheese, two pieces of pepper cheese and two pieces of ham cheese. To check the correctness of the completeness of the cheese and the correctness of the number of different flavors, the object detection algorithm can be used as follows:
- In the training phase, we define a model for each of the cheese flavors, so we need to define three different models according to the photo.
- Now we will check the accuracy of the completeness in each photo, we will find different models and count their number, and if the cheese is not complete, we will identify it.
In this article, we tried to review some of the applications of Object Detection in machine vision by citing some examples. Many other diverse tasks can be done with this algorithm, which we did not mention due to the limitations of this article.
You can get consultation from our experts to carry out your projects.
related content
