Novel roughing sand granularity detection method based on image processing technology

Abstract

The invention provides a roughing sand granularity detection method based on image processing, and belongs to the field of roughing sand granularity detection in casting. The original sand granularity is an important basis for evaluating the original sand performance, the original sand granularity mainly comprises the original sand granularity size and the concentration and dispersion degree, the original sand granularity is determined by mostly adopting low-efficiency and off-line manual screening detection at present, and the production requirement is difficult to guarantee. The method provided by the invention solves the problems of too long time consumption and complex process of a traditional crude sand granularity detection method. The method comprises a particle vibration dispersion system, an image acquisition system, an image processing system and a particle size analysis system. The roughing sand particle detection system disclosed by the invention can be used for simply and conveniently carrying out real-time detection on the granularity of the roughing sand particles, rapidly and accurately analyzing particle granularity parameters including area, perimeter, particle size, quantity and distribution condition, and has good development potential.

Description

Technical Field

[0001] The patent of this invention belongs to the field of raw sand particle size measurement in the foundry industry, and specifically relates to a new method for measuring raw sand particle size based on image processing method. Background Art

[0002] The particle size of raw sand is one of the important factors that affect the performance of molding sand and further affect the surface quality of castings. The detection of raw sand particle size is of great significance to the quality control of castings.

[0003] The particle size distribution of raw sand particles and the shape and size of the particles have a very important influence on the quality of castings. The traditional manual screening method is characterized by simple screening equipment and easy operation, but there are many screening processes. Not only is the efficiency low, but the test results are also greatly affected by human factors and sieve hole deformation. More advanced detection methods than traditional methods include laser detection and electron microscope detection, which can measure particle data more accurately, but the types of particles detected are limited. They are not widely used. At the same time, advanced detection equipment similar to laser and electron microscope detection have relatively strict requirements on the experimental environment. Therefore, this paper proposes a raw sand particle size detection method based on image processing technology, which is expected to reduce the damage to the raw sand and excessive human participation, streamline the measurement steps, and meet the real-time requirements and reduce time. Summary of the invention

[0004] This paper provides a new method for detecting the particle size of raw sand based on image processing method. On the basis of obtaining the image of raw sand particles, the target raw sand particle image is subjected to noise filtering and image enhancement and other processing through the image preprocessing algorithm, and then the raw sand information characteristics in different mesh size intervals are further obtained through the feature extraction algorithm. Finally, the raw sand particles are classified according to the raw sand information characteristics, so as to realize the rapid detection of the raw sand particle size. The present disclosure provides the following technical solutions to solve the above technical problems. The following steps are involved: S1. Weigh 10g of standard raw sand and analyze its particle size using the traditional screening method. S2. Weigh 1g of the same unscreened standard raw sand as above, place it on a vibration table with parallel light for vibration dispersion until the particle adhesion is relatively low. Use the backlight shooting method to collect particle images through the image acquisition system. S3. Perform image processing on the original sand particle image. S3.1. Crop the image using the outermost particles as the boundary. S3.2, perform 5x5 median smoothing filtering on the cropped image area, and perform grayscale processing after filtering enhancement. S3.3. Perform image binarization on the grayscale processed image to obtain a binary image, and select the global threshold OTSU in the binarization operation. S3.4, segment the binary processed image. This paper adopts a method combining morphological transformation and watershed segmentation to segment the adhesion area into separate particle areas. Figure 2 . S3.5. Marking the position of each particle in the original sand particle image according to the position of the original sand particle area. S3.6. Extract the particle size characteristics of the raw sand particles, obtain the threshold characteristics of the raw sand particles and the image, and use image recognition to obtain the raw sand particle detection parameters by using the threshold characteristics and the raw sand particle position information. S3.6.1. After extracting the image edge, calculate the perimeter L and area S of each raw sand particle, mark and count them, and identify the particle size and shape of the raw sand particles. The Feret minimum diameter was measured as the particle size of the raw sand particles in the image processing method. S3.6.2. Determine the calibration scale factor. After this step, convert the pixel size of the particles in the image to the actual size. Actual particle size = δ × particle pixel size The pixel area is S, the actual area s = S × δ, the pixel perimeter is L, the actual perimeter l = L × δ, and the Feret diameter pixel size is R i , the actual diameter is r i =R i ×δ. S4. Establish a mathematical model between the screening method and the image processing method. If the density of the original sand is ρ, the actual volume of the particle Quality M i =ρV i , the mass ratio of different mesh sizes is S5. Take 1g of raw sand particles of unknown particle size and use S2 to S3 to collect images of the raw sand particles and obtain characteristic parameters. Use the corresponding model described in S4 to convert the pixel size and actual size to obtain the particle size parameters of the raw sand particles, and perform particle size analysis on them.

[0005] According to the new method for determining the particle size of raw sand based on image processing method described in claim 1, it is characterized in that the specific steps of segmenting the adhesion area of ​​the binary image to obtain separate raw sand particle areas also include the following steps: using a segmentation method combining morphological processing and watershed to segment the adhesion area of ​​the binary image.

[0006] The image processing method described in claim 1 is characterized in that: the filtered image is subjected to morphological processing, and then the eroded image is used as a mask to reconstruct the image after the operation, and the maximum value part in the image is extracted. The reconstructed image is subjected to OTSU automatic threshold segmentation, and the holes generated inside the particles after binarization are filled with imfill() and then subjected to watershed segmentation to obtain a better segmentation effect.

[0007] Compared with the prior art, the advantages and positive effects of the present invention are: the present invention is based on an image processing method, by collecting images of raw sand after vibration dispersion, processing the images, and classifying raw sand particles of different sizes according to different program conditions, and then determining the proportion of particles of different sizes in the raw sand, which is expected to reduce the damage to the raw sand and excessive human participation, and simplify the measurement steps. The analysis speed is fast, the accuracy is high, and the degree of intelligence is high.

[0008] The automatic detection system designed according to this method has at least the following beneficial effects: it avoids the time and effort spent in manual screening and reduces the impact of human factors on errors. As long as the staff is familiar with the operation process, they can trigger the corresponding operation by clicking the corresponding button to realize the extraction of the image parameters of the original sand particles. There is no need to input programming control instructions to call the corresponding function, and the original sand particle morphology parameters can be quickly extracted, which reduces the difficulty of extracting particle parameters and saves labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is a flow chart of a new method for detecting the particle size of raw sand based on image processing method.

[0010] Figure 2 1 and 1 , which are the effect diagrams of the method combining morphology and watershed, wherein (a) is a local diagram of adhesion particles, and (b) is the effect diagram after segmentation.

[0011] Figure 3 It is the original sand particle position marking diagram.

[0012] Figure 4 It is the particle size distribution diagram of the raw sand particles.

[0013] Figure 5 It is the particle size distribution diagram of the raw sand particles.

[0014] Figure 6 It is the computer application interface of the original sand particle size automatic detection system.

[0015] Figure 7 This is the composition diagram of the original sand particle size detection system DETAILED DESCRIPTION

[0016] The purpose of the present invention is achieved by the following technical solutions: Figure 1 . S1. Weigh 10g of standard raw sand and analyze its particle size using the traditional screening method. S2. Weigh 1g of the same unscreened standard raw sand as above, place it on a vibration table with parallel light for vibration dispersion until the particle adhesion is relatively low. Use the backlight shooting method to collect particle images through the image acquisition system. S3. Perform image processing on the original sand particle image. S3.1. Crop the image using the outermost particles as the boundary. S3.2, perform 5x5 median smoothing filtering on the cropped image area, and perform grayscale processing after filtering enhancement. S3.3. Perform image binarization on the grayscale processed image to obtain a binary image, and select the global threshold OTSU in the binarization operation. S3.4, segment the binary processed image. This paper adopts a method combining morphological transformation and watershed segmentation to segment the adhesion area into separate particle areas. Figure 2 . S3.5, then mark the position of each particle in the original sand particle image according to the position of the original sand particle area. Figure 3 . S3.6. Extract the particle size characteristics of the raw sand particles, obtain the threshold characteristics of the raw sand particles and the image, and use image recognition to obtain the raw sand particle detection parameters by using the threshold characteristics and the raw sand particle position information. S3.6.1. After extracting the image edge, calculate the perimeter L and area S of each raw sand particle, mark and count them, and identify the particle size and shape of the raw sand particles. The Feret minimum diameter was measured as the particle size of the raw sand particles in the image processing method. S3.6.2. Determine the calibration scale factor. After this step, convert the pixel size of the particles in the image to the actual size. Actual particle size = δ × particle pixel size The pixel area is S, the actual area s = S × δ, the pixel perimeter is L, the actual perimeter l = L × δ, and the Feret diameter pixel size is R i , the actual diameter is r i =R i ×δ. S4. Establish the corresponding model between the screening method and the image processing method. If the density of the original sand is ρ, the actual volume of the particle Quality Mi =ρV i , the mass ratio of different mesh sizes is S5. Take 1g of raw sand particles of unknown particle size and use S2 to S3 to collect images of raw sand particles and obtain characteristic parameters. Use the mathematical model described in S4 to convert pixel size and actual size to obtain particle size parameters of raw sand particles and perform particle size analysis. Figure 4 , Attachment Figure 5 . S6. The software system mainly includes acquisition module, image preprocessing module, particle segmentation module, and calibration measurement module. Figure 6 . S7. The original sand particle size automatic detection system mainly consists of three parts: particle vibration dispersion system, image acquisition system and image processing system. The system still needs online operation by staff in the subsequent image processing stage. Figure 7 .

Claims

1. A new method for measuring the original sand particle size based on image processing technology, It is characterized in that It includes the following steps: 1) Weigh 10 g of standard raw sand and perform particle size analysis using the traditional screening method. 2) Weigh 1g of the same unscreened standard raw sand as above, place it on a vibration table with parallel light for vibration dispersion until the particle adhesion is relatively low. Use the backlight shooting method to collect particle images through the image acquisition system. 3) Performing image processing on the collected sample images, including the following steps: a) Crop the image using the outermost particles as the boundary. b) Perform a 5x5 median smoothing filter on the cropped image area, and perform grayscale processing after filtering enhancement. c) The grayscaled image is subjected to image binarization to obtain a binary image, and a global threshold OTSU is selected in the binarization operation. d) Segmenting the binary image. This paper adopts a method combining morphological transformation and watershed segmentation to segment the adhesion area into separate particle areas. e) marking the position of each particle in the original sand particle image according to the position of the original sand particle area. f) Extract the particle size characteristics of the raw sand particles, obtain the threshold characteristics of the raw sand particles and the image, and use image recognition to obtain the raw sand particle detection parameters - area, perimeter, particle size, quantity and distribution, etc. 4) Establish a corresponding model between the particle size obtained by the screening method and the particle size parameters obtained by the image processing method. 5) Take 1 g of raw sand particles of unknown particle size and use steps 2 and 3 described in power 1 to collect images of the raw sand particles and obtain characteristic parameters, use the corresponding model described in step 4 in power 1 to convert the pixel size and the actual size, obtain the particle size parameters of the raw sand particles, and further perform particle size analysis on the raw sand.

2. The method for detecting raw sand particles according to claim 1, Features The method of combining morphological processing with watershed segmentation is used to segment the adhesion particles, including the following steps: performing opening and closing operations and corrosion and expansion operations on the binary image, and then using watershed segmentation to segment the adhesion area of ​​the binary image.

3. The method for detecting raw sand particles according to claim 1, It is characterized in that The pixel area is S, the actual area s = S × δ, the pixel perimeter is L, the actual perimeter l = L × δ, and the Feret diameter pixel size is R i , the actual diameter is r i =R i ×δ. If the density of the original sand is ρ, the actual volume of the particle Quality M i =ρV i , the mass ratio of different mesh sizes is 4. The image processing method according to claim 1 is characterized in that: the filtered image is subjected to morphological processing, and then the eroded image is used as a mask to reconstruct the image after the operation, and the maximum value part in the image is extracted. The reconstructed image is subjected to OTSU automatic threshold segmentation, and the holes generated inside the particles after binarization are filled with imfill() and then subjected to watershed segmentation to obtain a better segmentation effect.

Images