TWI569766B - A method for predicting saliva image recognition in female ovulation - Google Patents

Description

Saliva image identification method for predicting female ovulation

The invention relates to a saliva image recognition method, in particular to a saliva image recognition method for predicting a female ovulation period.

In general, women's menstrual cycle is about 27 to 30 days, of which the menstrual period lasts about 3 to 7 days, while the ovulation period lasts about 6 days. Therefore, in order to detect the ovulation period to obtain birth control or birth control, Several kinds of detection methods such as basal body temperature method, vaginal secretion analysis and urine test have been developed. Among many detection methods, saliva detection is a relatively new technology, which mainly distinguishes the crystal pattern of dry saliva. The ovulation state of the saliva master, according to the existing research data, the crystal pattern can be divided into punctate, semi-female and whole dentate, and corresponding to the female non-ovulation period, possible ovulation period and ovulation period.

However, the existing saliva detection methods mostly apply saliva to a glass slide, and then observe and recognize through a microscope, which is not only inconvenient to operate, but the recognition success rate depends on the experience of the identifier, so that the industry has developed such as the United States. Patent Publication No. US20080255472 discloses a method of periodically capturing a saliva image, performing image binarization processing, converting the saliva image into a binarized image, and then analyzing the black pixel density of the binarized image. And record the graph into a density statistical graph, so that the saliva master can understand his own through the density statistical graph. Ovulation status.

However, the above patent adopts the method of defining the grayscale value of the pixel to filter out noise such as noise in the image, which is easy to be misjudged, resulting in poor recognition success rate, and the saliva master can only observe the ovulation through the density statistical graph afterwards. In the state, it is impossible to immediately know the ovulation state at that time. Therefore, after observing the above-mentioned shortcomings, the inventor of the present invention considered that the existing saliva image recognition method is necessary for further improvement, and the present invention is produced.

The main object of the present invention is to provide a saliva image recognition method for predicting a female ovulation period, which is capable of automatically detecting and analyzing a female saliva image in a simple, rapid, safe and accurate manner, and automatically determining the female from the saliva image. Ovulation status.

In order to achieve the above object, the saliva image recognition method for predicting a female ovulation period includes the following steps: an image capturing step is performed by using an image capturing device to obtain a dry saliva image, and the image is obtained by the image capturing device. a grayscale process for converting the dried saliva image into a grayscale saliva image; and an optimized image step of removing noise from the grayscale saliva image by using a filter and enhancing image detail of the grayscale saliva image; The image binarization step sets a threshold value, and compares the grayscale value of each pixel of the grayscale saliva image with the threshold value, and if the grayscale value of the pixel is lower than or equal to the threshold value, Defining the pixel as a black pixel, and if the grayscale value of the pixel is higher than the threshold, defining the pixel as a white pixel, thereby obtaining a binarized image; and a pattern recognition step, passing through a determining machine The binarized image is analyzed and automatically determined to be one of a non-ovulatory image, a possible ovulation image, and an ovulation image.

The method for identifying a saliva image for predicting a female ovulation period provided by the present invention can be simple, rapid, and safe by the step of acquiring an image, the step of optimizing the image, the step of binarizing the image, and the step of identifying the pattern. Accurately and automatically detect and analyze the saliva image of a woman, and automatically determine the female line in one of the non-ovulation period, the possible ovulation period, and the ovulation period by the saliva image, thereby helping the user to predict the ovulation period and thereby obtaining autonomy Control of birth or birth control.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a schematic flow chart of a first preferred embodiment of the present invention.

Figure 2 is a schematic flow chart of a second preferred embodiment of the present invention.

Figure 3 is a flow chart showing a third preferred embodiment of the present invention.

Figure 4 is a flow chart showing a fourth preferred embodiment of the present invention.

Please refer to FIG. 1 , which is a schematic flow chart of a first preferred embodiment of the present invention. The method for identifying a saliva image for predicting a female ovulation period is disclosed. The saliva image recognition method includes the following steps. : In the image capturing step, an image of the dry saliva is taken by an image capturing device, and the image of the dried saliva is converted into a grayscale saliva image by image graying.

An optimized image step is to remove the grayscale saliva shadow by using a filter Like the noise, and enhance the image details of the grayscale saliva image.

An image binarization step is to set a threshold value, and compare the grayscale value of each pixel of the grayscale saliva image with the threshold value, if the grayscale value of the pixel is lower than or equal to the threshold value, Then, the pixel is defined as a black pixel, and if the grayscale value of the pixel is higher than the threshold, the pixel is defined as a white pixel, thereby obtaining a binarized image.

The one type identification step analyzes the binarized image through a determination mechanism, and automatically determines that the binarized image is one of a non-ovulatory image, a possible ovulation image, and an ovulation image.

For a further understanding of the structural features of the present invention, the application of the technical means, and the intended effect, the manner of use of the present invention will be described. It is believed that the present invention may be further understood and understood as follows: As shown in FIG. 1 , the saliva image recognition method of the present invention is mainly performed by using MATLAB software, and the flow of the steps is as follows. First, the user needs to use the image capturing device to capture and obtain the dry saliva image, wherein the image capturing device can It is a general mobile phone, which is convenient for the user to shoot anytime and anywhere, or a special instrument developed for the saliva image shooting, thereby ensuring the correctness of the image capturing, and does not limit the user to obtain the dry saliva image. the way.

Then, using the image graying process, converting the dried saliva image into the grayscale saliva image, and removing the noise of the grayscale saliva image by using a filter, and enhancing the image detail of the grayscale saliva image, in this embodiment In the middle, a median filter is used to remove noise such as speckle noise in the grayscale saliva image, and further A high-enhancement filter is used to suppress low-frequency information, thereby enhancing the edge features of the grayscale saliva image and sharpening the image details.

Then, the grayscale saliva image is binarized by comparing the grayscale value of each pixel of the grayscale saliva image with the threshold value, if the grayscale value of the pixel is lower than or equal to The threshold is defined as a black pixel, and if the grayscale value of the pixel is higher than the threshold, the pixel is defined as a white pixel, thereby obtaining the binarized image, wherein the threshold is User-defined or automatically selected based on edge information and noise in the image.

Finally, the binary image is analyzed by the determination mechanism, and the binary image is automatically determined to be one of a non-ovulation image, a possible ovulation image, and an ovulation image. In this embodiment, the determination mechanism is By using the number of white pixels, the number of white pixels in the image is divided by the sum of the number of black pixels and white pixels to calculate the binary value as the pattern of the saliva of the saliva image becomes more prominent. The white pixel density of the image is then compared with a preset first threshold value and a second threshold value. If the white pixel density is lower than the first threshold value, the binarization is determined. The image is a non-ovulation image, and if the white pixel density is between the first threshold and the second threshold, determining that the binarized image is a possible ovulation image, and if the white pixel density is higher than the first The threshold value is determined as the ovulation period image, wherein the first threshold value and the second threshold value vary according to the user's physical condition, so the first threshold value and the second threshold value After the system is defined by the users themselves, or tracking analysis for some users of the ovulation state time, according to statistics obtained automatically defined.

Thereby, the saliva image recognition method of the present invention can automatically detect and analyze the saliva image of a woman in a simple, rapid, safe and accurate manner, and automatically determine that the female line is in a non-ovulation period, a possible ovulation period, and ovulation. One of the periods, which can help users predict the ovulation period, and then gain control of their own birth or birth control.

Please refer to FIG. 2 at the same time, which is a schematic flowchart of a second preferred embodiment of the present invention. The saliva image identification method is different from the foregoing first preferred embodiment in that the optimized image step and the The image binarization step further includes an edge detection step of performing edge detection processing on the grayscale saliva image to first search for a pixel indicating that the grayscale value of the image changes significantly, so that the subsequent execution is performed. In the image binarization step, the edge features of the binarized image are made cleaner and more obvious, and the accuracy of the pattern recognition can be improved. The edge detection processing can be performed by Sobel, Canny, Prewitt, and the like. An operation method is performed. In this embodiment, the Sobel operation mode is adopted.

Please refer to FIG. 3 at the same time, which is a schematic flowchart of a third preferred embodiment of the present invention. The saliva image identification method is different from the foregoing second preferred embodiment in that the image capturing step and the The step of optimizing the image further includes an image cropping step of locking and cropping the gray block saliva image to reduce the amount of the gray scale saliva image to improve the speed of the overall image operation, and Achieve the effect of reducing the false positive rate.

Please refer to FIG. 4 at the same time, which is a schematic flowchart of a fourth preferred embodiment of the present invention, the saliva image identification method and the foregoing third preferred embodiment. The difference is that between the image binarization step and the pattern recognition step, there is further included an image thinning step, which is to simplify the white pixel of the binarized image into a thinned image with a width of 1 pixel. And the image thinning step and the pattern identifying step further comprise a feature capturing step of performing a Hough transform process on the thinned image to extract an edge feature included in the image, and the The judgment mechanism adopts the method of data exploration to judge. The method of data exploration can be carried out by one of the methods such as Bayesian classification, neural network classification, and decision tree classification. In this embodiment, decision tree classification is adopted. For data exploration, the specific implementation examples are as follows.

Since the fern pattern is mainly composed of a plurality of long line segments and a plurality of short line segments which are staggered or parallel, the six line segment features, that is, the bus segment and the long line segment, can be extracted from the saliva image with the fern pattern. The short line segment, the parallel line segment, the percentage of the long line segment and the percentage of the parallel line segment, and the above line segment features are established into a decision tree using the J48 classification of the weka software, so that the edge feature of the above-mentioned Hough transform is compared to the decision The node of the tree can quickly identify the line segment features of the thinned image, and further judges one of the non-ovulation image, the possible ovulation image, and the ovulation image, thereby being able to be explored by means of data exploration. Further improve the recognition success rate.

Further, the features of the present invention and the achievable expected efficacy thereof are as follows: the saliva image recognition method for predicting a female ovulation period according to the present invention, wherein the image acquisition step, the optimized image step, and the image binarization are performed. step And the type identification step can automatically detect and analyze the saliva image of the woman in a simple, rapid, safe and accurate manner, and automatically determine the female line in one of the non-ovulation period, the possible ovulation period and the ovulation period by the saliva image. In this way, it is necessary to help the user to predict the ovulation period, and then to obtain control of the independent birth or birth control.

In summary, the present invention has excellent advancement and practicability in similar products, and at the same time, the technical materials of such structures are frequently investigated at home and abroad, and the same structure is not found in the literature. The invention already has the invention patent requirements, and the application is filed according to law.

However, the above-mentioned ones are merely preferred embodiments of the present invention, and the equivalent structural changes of the present invention and the scope of the claims are intended to be included in the scope of the present invention.

Claims (10)

A saliva image recognition method for predicting a female ovulation period comprises the following steps: an image capturing step is performed by taking an image capturing device to obtain a dry saliva image, and performing image graying treatment to dry the saliva The image is converted into a grayscale saliva image; an optimized image step is to remove the noise of the grayscale saliva image by using a filter, and enhance the image detail of the grayscale saliva image; an image binarization step is to set a threshold And comparing the grayscale value of each pixel of the grayscale saliva image with the threshold. If the grayscale value of the pixel is lower than or equal to the threshold, the pixel is defined as a black pixel, and if the pixel If the grayscale value is higher than the threshold, the pixel is defined as a white pixel, thereby obtaining a binarized image; the one type identification step analyzes the binarized image through a determining mechanism, and automatically determines the second The imaged image is one of a non-ovulatory image, a possible ovulation image, and an ovulation image. According to the method for identifying a saliva image for predicting a female ovulation period according to the first aspect of the patent application, the judging mechanism analyzes the white pixel density of the binarized image. According to the method for identifying a saliva image for predicting a female ovulation period according to the second aspect of the patent application, wherein a first threshold value and a second threshold value are set, when the white pixel density is lower than the first threshold value And determining that the binarized image is a non-ovulated image, and when the white pixel density is between the first threshold and the second threshold The value determines that the binarized image is a possible ovulation image, and the white pixel density is higher than the second threshold value, and the binarized image is determined to be an ovulation image. The saliva image identification method for predicting a female ovulation period according to the first aspect of the patent application scope, wherein the determination mechanism is determined by means of data exploration. According to the fourth aspect of the patent application, the saliva image identification method for predicting female ovulation period, wherein the decision tree classification is used for data exploration. According to the fifth aspect of the patent application, the method for identifying saliva images for predicting female ovulation period, wherein a bus segment, a long line segment, a short line segment, a parallel line segment, and a long line are extracted from a saliva image with a fern pattern. A six-segment feature, such as the percentage of the segment and the percentage of parallel segments, creates a decision tree. The saliva image recognition method for predicting a female ovulation period according to the first aspect of the patent application, wherein the step of acquiring the image and the step of optimizing the image further comprises an image clipping step for the gray scale The saliva image is locked and cropped by a particular block to reduce the amount of data in the grayscale saliva image. The saliva image recognition method for predicting a female ovulation period according to the first aspect of the patent application, wherein between the optimized image step and the image binarization step, an edge detection step is further included. The grayscale saliva image performs edge detection processing to find pixels that indicate significant changes in grayscale values in the image. The saliva image recognition method for predicting a female ovulation period according to the fourth aspect of the patent application, wherein the image binarization step and the pattern recognition step further comprise an image thinning step. White pixels of the binarized image Simplified into a thinned image having a width of 1 pixel, and between the image thinning step and the pattern identifying step, further comprising a feature capturing step of performing a Hough transform process on the thinned image to extract The edge features contained in the image. The saliva image recognition method for predicting a female ovulation period according to the first aspect of the patent application, wherein the optimization image step uses a median filter to remove noise such as speckle noise of the gray scale saliva image, and A high-enhancement filter suppresses low-frequency information, thereby enhancing the edge features of the grayscale saliva image and sharpening the image details.