Method for detecting water-soaked fragments

Abstract

To provide a method for detecting water-soaked and broken particles that can detect them with greater accuracy and effectiveness than conventional methods. [Solution] The method is characterized by comprising: an imaging processing step (S10) of capturing an original image of a rice grain immersed in water; a grain region smoothing processing step (S111) of generating grain region smoothing image data by smoothing the grain region from the captured original image of the rice grain; a contour / broken grain extraction step (S112) of generating contour / broken grain extraction image data by extracting the contour of the grain region and the broken grain portion from the captured original image of the rice grain; and a composite image generation processing step (S113) of generating composite image data by combining the grain region smoothing image data and the contour / broken grain extraction image data.

Description

Technical Field

[0001] The present invention relates to a method for detecting water-soaked cracked grains that can optically detect water-soaked cracked grains in rice grains and the like.

Background Art

[0002] Conventionally, optical detection of water-soaked cracked grains in rice grains has been carried out. For example, in Patent Document 1, as shown in FIG. 6 and the like of this document, an image of the captured rice grains is processed to generate a region image (inverted binary image) that clarifies the region of the rice grains and a contour binary image that emphasizes water-soaked cracks, respectively, and a composite image that emphasizes water-soaked cracks for each rice grain is generated by synthesizing these images. This is an attempt to improve the detection accuracy of the water-soaked cracked part of the rice grains.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, according to the method disclosed in Patent Document 1 above, for example, as shown in FIG. 15 of the drawings of the present application, by binarization processing, the cracked grain part and the background image are distinguished as the same "white". Although this improves the visibility, it is not possible to specify only the cracked grain part and obtain its area.

[0005] In order to specify only the cracked grain part, it is necessary to perform additional separate image processing such as pattern matching or machine learning.

[0006] In view of such problems, an object of the present invention is to provide a method for detecting water-soaked cracked grains that can detect water-soaked cracked grains more accurately and effectively than before. [Means for solving the problem]

[0007] To solve the above problems, the present invention provides a method for detecting water-immersed cracked rice grains, comprising: an imaging processing step of capturing an original image of a water-immersed rice grain; a grain region smoothing processing step of generating grain region smoothed image data by smoothing the grain region from the captured original image of the rice grain; a contour / cracked grain extraction processing step of generating contour / cracked grain extraction image data by extracting the contour of the grain region and the cracked grain portion from the captured original image of the rice grain; and a composite image generation processing step of generating composite image data by combining the grain region smoothed image data and the contour / cracked grain extraction image data.

[0008] Furthermore, the method for detecting water-immersed broken grains is characterized in that the composite image generation processing step includes a summing processing step in which the brightness value of each pixel of the smoothed grain region image data and the brightness value of each pixel of the contour / broken grain portion extraction image data are summed.

[0009] Furthermore, the method for detecting water-immersed cracked particles is characterized by having a cracked particle extraction image generation step, which generates cracked particle extraction image data by extracting pixels with a predetermined brightness value or higher from the composite image data generated in the composite image generation step, and extracting the brightness values ​​of pixels with a brightness value lower than the predetermined brightness value.

[0010] Furthermore, the method for detecting water-soaked cracked grains is characterized by having a merged image generation step in which the cracked grain extraction image data generated in the cracked grain extraction image generation step is combined with the original image of the rice grain to generate a merged image.

[0011] Furthermore, the method for detecting water-immersed broken grains is characterized by having a marked image generation step which calculates the centroid of the broken grain in the broken grain extraction image data generated in the broken grain extraction image generation step, and draws a mark centered on the centroid on the original image to generate a marked image.

[0012] Furthermore, the method for detecting water-immersed broken grains is characterized by having a marked merged image generation process step which calculates the centroid of the broken grain in the broken grain extraction image data generated in the broken grain extraction image generation process step, and draws a mark centered on the centroid in the merged image to generate a marked merged image. [Effects of the Invention]

[0013] According to the present invention, since the broken parts of rice grains can be clearly displayed using synthesized image data, it becomes possible to easily and accurately distinguish between water-soaked and broken grains. Furthermore, since the conventional binarization process is not performed, it is possible to prevent the loss of fine broken grains during the image processing stage. In addition, it becomes possible to quantitatively determine the size of the broken grains from the pixel information. [Brief explanation of the drawing]

[0014] [Figure 1] This is a detection flow for water-soaked broken particles in one embodiment of the present invention. [Figure 2] This is a flowchart illustrating the image processing process in one embodiment of the present invention. [Figure 3] This is an example of the original image and grain region smoothing image data of a grain of rice in one embodiment of the present invention. [Figure 4] This is an example of the original image of a broken grain of rice and the brightness value of the original image in one embodiment of the present invention. [Figure 5] This is an example of grain region smoothing image data of a broken grain portion of a rice grain and the brightness value of the grain region smoothing image data in one embodiment of the present invention. [Figure 6] This is an example of the original image of a grain of rice and the contour / broken grain area extraction image data in one embodiment of the present invention. [Figure 7] This is an example of contour and fragment extraction image data of a broken grain of rice, and the brightness value of the contour and fragment extraction image data, according to one embodiment of the present invention. [Figure 8] This is an example of generating composite image data by combining grain region smoothing image data and contour / cracked grain area extraction image data of a rice grain according to one embodiment of the present invention. [Figure 9] An example of composite image data of the cracked part of rice grains and the luminance value of the composite image data in one embodiment of the present invention. [Figure 10] An example of a method of adding up luminance values in one embodiment of the present invention. [Figure 11] A flow chart explaining an image processing step in another embodiment of the present invention. [Figure 12] An example of composite image data, cracked part extraction image data, and stamped image in another embodiment of the present invention. [Figure 13] An example of a merged image in another embodiment of the present invention. [Figure 14] An example of a stamped merged image in another embodiment of the present invention. [Figure 15] A diagram explaining the problems of the prior art.

Mode for Carrying Out the Invention

[0015] Hereinafter, an embodiment of the method for detecting water-soaked cracked grains of the present invention will be described with reference to the drawings. Note that the present invention is not limited to the embodiments shown below.

[0016] FIG. 1 shows a schematic flow chart of the method for detecting water-soaked cracked grains in the present embodiment. Further, FIG. 2 shows a detailed processing flow chart for the image processing (S11) in FIG. 1.

[0017] That is, the method for detecting water-soaked cracked grains in the present embodiment includes at least an imaging processing step (S10) of imaging an original image of water-soaked rice grains, a grain region smoothing processing step (S111) of generating grain region smoothed image data by smoothing the grain region from the original image of the imaged rice grains, a contour and cracked part extraction processing step (S112) of generating contour and cracked part extraction image data by extracting the contour of the grain region and the cracked part from the original image of the imaged rice grains, and a composite image generation processing step (S113) of generating composite image data by combining the grain region smoothed image data and the contour and cracked part extraction image data.

[0018] The composite image data generated as described above is output to a display device or discrimination device (not shown) in the composite image output step. Then, based on the output composite image data, water-immersed fragments are detected in the fragment discrimination processing step (S12).

[0019] To explain in more detail, in the imaging processing step (S10) described above, a raw image of a rice grain immersed in water is captured, as shown in Figure 3. Then, in the grain region smoothing processing step (S111), the captured raw image of the rice grain is processed to generate grain region smoothed image data, as shown in Figure 3, in which the grain region is smoothed. Note that the grain region smoothed image data is not limited to the visualized image data, but also includes the image data processed in the grain region smoothing processing step (S111).

[0020] In generating the above-mentioned grain region smoothing image data, broken grains and noise present within the rice grains in the original image are removed, and a grain region detection process is performed. This removes noise from the grain regions and reduces misjudgment during grain contact. For example, closing, opening, smoothing, and Gaussian filtering processes can be used when generating the grain region smoothing image data. Figure 4 shows an example of the original image of a broken rice grain and the brightness value of the original image, while Figure 5 shows an example of grain region smoothing image data of a broken rice grain and the brightness value of the grain region smoothing image data.

[0021] In addition, the captured original image of the rice grain is processed in the contour and broken grain extraction step (S112) to generate contour and broken grain extraction image data shown in Figure 6, which extracts the contour of the grain region and the broken grain.

[0022] In generating the above contour and broken grain area extraction image data, the contours of grain regions and broken grain areas are extracted from the original image using processes such as second-order differentiation (Laplacian filter) and first-order differentiation (Sobel, Prewitt filter). Figure 7 shows an example of contour and broken grain area extraction image data and the brightness values ​​of the contour and broken grain area extraction image data for a broken grain of rice. Note that the contour and broken grain area extraction image data is not limited to the visualized image data, but also includes image data processed in the contour and broken grain area extraction processing step (S112).

[0023] Next, as shown in Figure 8, the grain region smoothing image data and the contour / broken grain area extraction image data obtained in each of the above steps are combined in the composite image generation processing step (S113) to generate composite image data. Figure 9 shows an example of the composite image data of the broken grain area of ​​a rice grain and the brightness value of the composite image data.

[0024] In the composite image generation processing step (S113) described above, the brightness value of each pixel in the granular region smoothed image data shown in Figure 5 and the brightness value of each pixel in the contour / split grain area extracted image data shown in Figure 7 are added together. Note that the composite image data is not limited to the visualized image data, but also includes image data processed in the composite image generation processing step (S113).

[0025] Figure 10 illustrates the summing process described above. Specifically, by summing the brightness values ​​of each pixel in the grain region smoothing image data with the brightness values ​​of each pixel in the contour / grain extraction image data, it becomes possible to emphasize even very small grain areas and output them as composite image data.

[0026] In the illustrated 8-bit grayscale image, it can be seen that only the split areas are output with a brightness value of "255" or a value close to it. This makes it easy to identify the split areas. Furthermore, as shown in the figure, it can be seen that there are 15 pixels that show a brightness value of "255", which makes it possible to quantitatively calculate and understand the area of ​​the split areas.

[0027] Furthermore, as shown in Figure 10, even when processing the same original image, there is a positional shift between the boundary line between the background and the grain region obtained by contour and broken grain area extraction image data (indicated boundary line B) and the boundary line between the background and the grain region obtained by grain region smoothing image data (indicated boundary line A). Specifically, boundary line B is shifted outward from the center position of the rice grain compared to boundary line A. In other words, the contours of the grain regions, which are the boundaries, do not overlap. As a result, as shown in the composite image data in Figure 8, the broken grain areas can be brightened and emphasized without the contour areas being emphasized.

[0028] The distinctive processing configuration described above makes it possible to clearly display the broken parts of rice grains in the composite image data, thus enabling easy and accurate identification of water-damaged broken grains. Furthermore, since conventional binarization processing is not performed, it is possible to prevent the loss of fine broken grains during the image processing stage. In addition, it becomes possible to quantitatively determine the size of the broken grains from the pixel information.

[0029] The method for detecting water-immersed, broken rice grains of the present invention can be implemented using various devices. For example, various devices such as a camera on a mobile terminal can be used to capture the original image of water-immersed rice grains. It is also possible to immerse rice grains in water using a container with a transparent bottom, such as a petri dish, and use images captured by a scanner while the grains are immersed. Pre-captured images can also be used, and such images may be obtained via the internet or cloud. Various image processing, display output, measurement data output, and discrimination result output can be performed by program control on a PC, mobile device, cloud server, etc.

[0030] (Other variations) Although one embodiment of the water-immersed broken particle detection method of the present invention has been described above, the present invention is not limited to the above embodiment and also includes the following.

[0031] For example, as shown in the flowchart in Figure 11 and the image in Figure 12, it is also possible to add a process to extract only the fragmented parts from the aforementioned composite image data by a fragmented part extraction image generation processing step (S114).

[0032] This generates image data of extracted cracked grains, as shown in Figure 12, making it possible to measure the length and size of the cracks based on the number of pixels. Based on this measurement data, it is possible to grade and evaluate the rice grains, which can be used by users such as rice milling factory managers for price evaluation analysis. Note that the image data of extracted cracked grains is not limited to visualized image data, but also includes image data processed in the image data extraction processing step (S114).

[0033] Furthermore, as shown in the flowchart in Figure 11 and the image in Figure 12, a marked image can be generated by the marked image generation processing step (S115). Specifically, in the extracted grain image data, the centroid points (coordinates) of consecutive white pixels in each grain are calculated, and in the original image, marks (such as the white square outlines shown) are drawn around these centroid points to generate the marks. The color of the marks can be set as appropriate, such as white or red, and their shape is not limited to squares but can also be circular.

[0034] Furthermore, as shown in the flowchart in Figure 11 and the image in Figure 13, a merged image can be generated by combining the original image and the extracted fragment image data described above in the merged image generation processing step (S116). The color of the fragment can be set to white, red, or other appropriate colors.

[0035] Furthermore, as shown in Figure 14, the marked merged image generation processing step (S117) also allows for the generation of a merged image with markings, similar to the marked image in Figure 12. By configuring the system to generate marked images, merged images, and marked merged images in this way, it becomes possible to display and output the images in a visually very easy-to-understand manner.

[0036] While embodiments and modifications of the present invention have been described above, the embodiments described above are for the purpose of facilitating understanding of the present invention and do not limit it. The present invention can be modified and improved without departing from its spirit, and the present invention includes equivalents thereof. Furthermore, the combinations or omissions of the components described in the claims and specification are possible to the extent that at least some of the above-mentioned problems can be solved or at least some of the effects can be achieved.

Claims

1. The imaging process step involves capturing an original image of rice grains submerged in water, A grain region smoothing step generates grain region smoothed image data obtained by smoothing the grain region from the original image of the captured rice grain, A contour and broken grain extraction processing step generates contour and broken grain extraction image data by extracting the contour of the grain region and the broken grain portion from the original image of the captured rice grain, The process includes a composite image generation step of generating composite image data by combining the smoothed grain region image data and the contour / grain extraction image data. A method for detecting water-soaked broken particles, characterized by the features described above.

2. The composite image generation process step includes a summing process step that sums the brightness values ​​of each pixel in the granular region smoothed image data and the brightness values ​​of each pixel in the contour / split grain area extracted image data. The method for detecting water-soaked broken particles according to claim 1, characterized by the features described above.

3. The composite image generation process step includes a process step to generate fragmented fragment image data by extracting pixels with a predetermined brightness value or higher from the composite image data generated in the composite image generation process step, and extracting the brightness values ​​of pixels with a brightness value lower than the predetermined brightness value. A method for detecting water-soaked broken grains according to claim 1 or 2, characterized by the above.

4. The process includes a merged image generation step, which combines the image data of the extracted grain portion generated in the grain portion extraction step with the original image of the rice grain to generate a merged image. The method for detecting water-soaked broken particles according to claim 3.

5. The process includes a step to generate a marked image, which involves calculating the centroid of the fragmented portion in the fragmented portion extracted image data generated in the fragmented portion extraction step, and drawing a mark centered on the centroid on the original image to generate a marked image. The method for detecting water-soaked broken particles according to claim 3.

6. The process includes a step to generate a marked merged image, which involves calculating the centroid of the fragmented portion in the fragmented portion extracted image data generated in the fragmented portion extraction process step, and drawing a mark centered on the centroid in the merged image to generate a marked merged image. The method for detecting water-soaked broken particles according to feature 4.

Images