Machine vision has become a common jargon in manufacturing industries. It helps to maintain and improve the product quality without intervening with the manufacturing process. A Typical Machine vision system comprises of controller, camera, lens, Illumination Unit and Image Processing Software.
The process of Image Processing involves the conversion of the captured image to electrical signals using CMOS. CMOS converts light signals(Photons) to electric Signals for digital processing. In this processed image, several operations can be done to extract the area of interest and employ tools for shape, size, color, etc. and provide judgment i.e. whether the area of interest aligns with the prerequisite specifications and the allowable deviations.
The complementary metal-oxide-semiconductor (CMOS) image sensor consists of a photodiode and a transistor switch for each pixel. This allows the pixel signals to be amplified individually. By operating the collection of switches, the pixel signals can be retrieved directly and sequentially, and at a much higher speed compared to the conventional CCD.
The ET XPro has been Indigenously developed to provide our customers with a Very Simple programmable Image Processing software. This helps our clients to program the Machine vision system without any expert.
ET XPro is equipped with a wide range of tools to accommodate applications from varied industries. The tools include Find shape, Pixel count, OCR, Blob, 1-D and 2-D code, Measure position, Width measurement, Fuzzy measurement, Fuzzy position, Diameter, and colour tool.
The ET XPro is mainly employed for Verification, Recognition, Inspection, and measurement. The ET XPro can be also be used for guided robots and inspection on-line. The Customized Software design helps the client inspect the product at varied locations.
Our technical guidance aims to help our customers to understand and also learn vision basics for selecting the Camera, Lens, Light selection, Accuracy calculation, Distance calculation, Line speed calculation, O ring selection calculation and the Software tools.
Working Distance (WD)
This signifies the distance from the end of the lens to the target when the focal point is in alignment with the target. This is also called the operating distance. With a CMOS, it forms a proportional expression that says Working Distance is to Field of View as Focal Length is to CMOS Size (Working Distance: Field of View = Focal Length: CMOS Size).
“Focal length” exists as one of the specifications of a lens. As representational models, lenses used for factory automation come in 8 mm 0.32″, 16 mm 0.63″, 25 mm 0.98″, and 50 mm 1.97″ specifications. The WD (working distance) can be equated to the position that aligns with the focal point from the field of view and focal length required for the target that you would like to capture.
The size of the WD and the field of view is determined by the focal length of the lens along with the size of the CMOS, and in ranges that are at the closest possible distance, where a close-up ring is not required, or at distances above this, the relationship can be represented with the following proportional expression.
Field of View
This is the image area within the range of the working distance. In general, the longer the working distance between the target and lens is, the wider the field of view (view angle). Additionally, the width of the field of view is determined according to the focal length of the lens. The angle of the range in which the lens can be used to capture images in regard to the field of view is called the angle of view or the view angle. Because the angle of view becomes larger as the focal length of the lens becomes shorter, the field of view will widen. Conversely, it’s possible to enlarge distant targets when the focal length is long.
Depth of Field
Depth of field means the range that appears in focus through the lens (target-side distance). When the range is wide, it is called “deep depth of field”. Conversely, when the range is narrow, it is called a “shallow depth of field”. Strictly speaking, though only one area can be in focus, to the human eye, images in a certain range appear to form clearly. This range is called depth of field.
Pixel Count Tools
The ET Xpro is equipped with a Pixel Count tool for calculating the Total Pixel of the image. For using the Pixel count tool, the user must choose the colour that needs to be picked. This is done to eliminate the unwanted areas in the image i.e only the region to be inspected is picked. The colour that is picked is considered as a white pixel and the rest is considered black. The number of white pixels in the reference image and the current images are compared and the judgment is made. This judgment is made by adjusting the upper and lower threshold. The number of pixels higher than the upper threshold is NG and also the number of pixels lesser than the Lower threshold is also NG.
As shown in the above image, the orange colour is picked. The orange colour is considered as white pixels and the rest becomes dark. The number of white pixels is counted and is displayed as Total pixel count. Based on the total pixel count in the reference image, the upper and lower tolerance is set.
The Find shape tool in the ET Xpro is used to compare the contours on the products based on size, shape, and position. A Rectangular box is drawn on the region of interest and is chosen as the reference image. After the reference image is stored, the region of interest for all the current images are compared. The output is provided in terms of match %(with respect to the reference image). Based on the threshold set for the match %, the judgment is provided(OK/NG).
As shown in the above image, the word “juice” is selected using the rectangle in the find shape tool. The shape of “juice” from the reference image is considered as 100% match. The current image is compared with the reference image and the match % is found. Based on the tolerance set for the match %, the judgment is made(OK/NG)
Each bar code contains data encoded in it. The data is usually the model name, Date of Manufacture, Price, etc. Manufacturers print the 1D codes on their products for easier traceability and reduced human errors. The ET Xpro has the capability of reading the bar codes at high speeds printed on the products. This not only reduces the time consumed for manual inspection but also prevents the products with faulty bar codes into the delivery line. The faults include Bar code mismatch, Poor print quality, missing bar codes, etc.
The ET XPro can read a wide range of 1D codes that includes :
|2/5 Interleaved||EAN-8 Add-On 2||GS1 DataBar Omnidirectional|
|Codabar||EAN-8 Add-On 5||GS1 DataBar Truncated|
|Code 39||EAN-13||GS1 DataBar Stacked|
|Code 32 (converted from Code 39)||EAN-13 Add-On 2||GS1 DataBar Stacked Omnidirectional|
|Code 93||EAN-13 Add-On 5||GS1 DataBar Limited|
|Code 128||UPC-A||GS1 DataBar Expanded|
|MSI||UPC-A Add-On 2||GS1 DataBar Expanded Stacked|
|PharmaCode||UPC-A Add-On 5|
|UPC-E Add-On 2|
|UPC-E Add-On 5|
The ET Xpro comes with a 2D barcode tool to decode data from various impressions that range from labels to Direct Punch marking. This tool uses an advanced algorithm and technology to ease the traceability and increase efficiency. The tool is designed for reading codes printed on different surfaces, quality grade and size. The tool has been designed to read a wide range of codes which includes
- DataMatrix (ECC200),
- GS1 DataMatrix,
- GS1 Composite (CC-A/CC-B/CC-C)
OCR is trusted for reliable inspection of printing defects, print quality, missing characters, character mismatch, and misprints across the manufacturing industry. Our OCR tool is finely crafted to read the characters that are commonly used/printed on products. It can be integrated with high-speed packaging and printing machines to check the accuracy of data such as MRP, manufactured date, batch code, and chassis number.
OCR Inspection is executed in simple steps
Image capturing: The Preliminary step for the OCR tool is image capturing. This is done by using a camera or using pre-captured images in a specified file location. The image captured is adjusted for brightness and alignment for better visibility. This can be done using image-enhancing filters in the software.
Processing: The next step is image processing. In this, each character is classified into individual blocks, and the outline of the characters is stored. The Software uses a special algorithm to compare the stored character with the available library of fonts. It is also checked for a mismatch, misprint, presence/absence.
Output: The final step is output processing. Afterimage processing is completed, the output is generated as OK/NG. OCR tool has the capability to maintain a record of the OK/NG products. After the completion of the above steps, the tool will provide signals to reject the NG products.
A blob is defined as a region of connected pixels. It a region of an image with the consistency of certain properties such as shape, intensity, or color. These tools can be used for pattern matching techniques. The main advantages of this technique include high flexibility and excellent performance. This tool helps in counting the products or sorting multiple items in the specified region.
The Measure position tool helps to measure the position of the region of interest-based on the user specification. In the above figure it is observed that when the measure position tool is applied, we are able to determine the position of the cap in terms of rows, columns, and angles. If there is an absence or shift in the position of the cap, there is a variation in the values of the rows, columns, or angle. By detecting this change in value based on the position of the product, the products are classified as good or not good products.
The width measurement tool helps in measuring the width of a product. This tool also helps in the identification of gaps between the product/area of interest. To calculate the width, the user has to draw the outline of the product using the rectangular shape provided in the drop-down box. This tool uses either a black or white region to measure the width of the product. The tool measures the distance between the edges of the black region or the white region to calculate the width. The pixels calculated using the width measurement tool can be converted to a millimetre.
The fuzzy measurement tool is an excellent tool to measure the width of the profile. This tool measures the distance between multiple edges measured in the inspection region. This tool is used to measure the maximum and minimum width in target images. It is possible to measure the width in both the X & Y directions. We can even measure the width of the product at a particular segment of the multiple edge points that are detected. This tool will pass or fail the product of the product if the dimensions get out the range.
This tool can be used to measure many edge points in a profile and detects the maximum and minimum points from among the multiple edge points detected in the inspection region and it gives the position of either the maximum or minimum edge point or a specific edge point. It can also find a best fit circle or a best fit line through the information of the edge points detected. The fuzzy position tool can also detect the angle of the edge points of a profile.
This tool helps to determine the diameter of the product. To calculate the diameter, the user has to draw the outline of the product using the circle / oval shape provided in the drop-down box. Using the outline the tool divides the product into several quadrants and determines the diameter of the product. The average diameter, the maximum diameter and the minimum diameter of the product can be found. It is also possible to fine the diameter of the product at a particular segment.
The color tool is used to find the difference in color on the target sample. This tool will differentiate the color based on the changes in the RGB Values and also HSB values. Thus it makes it possible to differentiate the difference between the same color with different shades. (i.e.) this tool can differentiate easily between light blue, mild blue, and dark blue. Which makes color detection more effective.
In many machine vision applications, there is a necessity to convert the image to binary. The Binary filter converts a color image into binary (black and white). This is done by converting the coloured image into a grayscale(0-255 bit). After the image is converted to the grayscale, the upper and lower threshold is set by the user. The range of pixels lying outside the threshold is considered as black pixels and the range between the threshold is considered as white.
The images are converted into pixel blocks of 3×3. When the expand filter is used, the sub-pixel block with the largest bright intensity is considered and is applied as the value for the center block thereby increasing the number of white pixels in the image.
The images are converted into pixel blocks of 3×3. When the shrink filter is used, the pixel block with the largest dark intensity is considered and is applied as the value for the center block. This process reduces the number of white pixels in the image.
In the Blur filter, each pixel is computed as the average of the surrounding pixels. It’s a weighted mean of the surrounding pixels that gives more weight to the pixel near the current pixel. This kind of filter is also called a smoothing filter or low-pass filter. This filter blurs the image and the edges. This will reduce the noise in the image can be due to external factors like poor lighting or environment or internal factors like a few pixels have gone bad.
Changing the luminosity of a picture in a RGB mode can be done by adding a constant to each colour component. It is also possible to change from RGB to HSL to modify the luminosity easily. Thus by using this filter the brightness of the image can be either increased or decreased.