JP5869822B2 - Dental image processing system - Google Patents

Description

  The present invention relates to a dental image processing system.

  Due to the optical resolution and small size of the digital camera, it is possible to clearly capture a limited space such as in the oral cavity. The intraoral image can be used for informed consent, etc. Use in the dentistry industry, such as recording, has become increasingly popular.

  In Japanese Patent Laid-Open No. 2008-228818, intraoral image processing by a computer is proposed, and a feature amount based on color information of each pixel is acquired as tooth image data.

JP 2008-228818 A

The oral cavity is affected by saliva and is a narrow and deep space, so diffuse reflection due to illumination light, etc. at the time of shooting, and intense light spots are scattered. In this case, it is difficult to determine the color due to misrecognition, thin gums, and dirty teeth.

In view of the above, the present invention provides an image input means for inputting a photographed intraoral image, from an image input by the image input means,
(1) Light spot detection means for detecting a light spot, a light spot is a group of four or eight connections, a label to identify each mass and a part that is not a light spot, Based on the label, the area of the connected cluster of each light spot is calculated, the area is compared with a predetermined threshold, and if it is smaller than the threshold, the light spot of noise is determined by determining that it is a noise light spot. Means for detecting the light spot of the noise, by replacing the noise light spot by replacing it with the label of the surrounding mass, and using the information from which the noise light spot has been removed to detect the boundary between the gingiva and the tooth , Eliminate false detection due to the effect of light spot, and enable accurate recognition of gingiva and teeth.

(2) Light spot detection means to detect, the light spot is expanded to a certain extent by the expansion filter, the average value of a certain range of colors around the mass is calculated, and the expanded light spot cluster of by replacing the average color calculated color area, means for removing noise light spot, using the information to remove noise spots, in gum and teeth of the boundary detection using color information, of the light spot It eliminates false detections due to effects and enables accurate recognition of gingiva and teeth.

Further, the present invention relates to an image input means for inputting a photographed intraoral image, and from the image input by the image input means, the boundary value between the teeth and the gingiva corresponding to the gingiva, depending on whether the color information is above or below a certain threshold. First boundary detection means to detect, and from the image input by the image input means, a color information determination method different from the first boundary detection means is used, and the boundary is detected based on whether the color information is above or below a certain threshold. The degree of coincidence between the boundary value obtained by the second boundary detection means and the first boundary detection means and the boundary value obtained by the second boundary detection means is calculated by calculation, and the proportion of coincidence increases. As described above, the threshold values of the first boundary detection means and the second boundary detection means are determined, and the first boundary detection means and the second boundary detection means have the above (1) or (2) noise. tooth stains and gum by subjecting the light spots removed as pretreatment Thin recognition was possible to detect critical difficult gums and teeth.
The color information in the present invention is, for example, when using any of the R, G, B information of each dot of the original image, or from the R, G, B information of each dot of the original image to the HLS or HSV color space. After conversion, any one of H, L, and S information, information combining them, information about any of H, S, and V, or information combining them can be used.
Not limited to this, it can be used in other color spaces, such as Lab, XYZ, and CMYK.

The present invention is thin by detecting a light spot that is erroneously recognized in image processing among light spots generated by photographing and removing the influence of the light spot as a pre-processing, and then detecting a boundary between teeth and gums. It is possible to improve the accuracy of intraoral recognition in the intraoral image by accurately detecting the boundary between the gingiva and the tooth, which is difficult to detect due to the gingiva and dirt.

FIG. 1 is a diagram showing an embodiment of the present invention. FIG. 2 is a diagram showing an embodiment of the present invention. It is a figure explaining the operation | movement of FIG. It is a figure explaining the operation | movement of FIG.

The present invention detects a light spot in order to remove a noisy light spot from a locally high brightness value and a color range closer to white caused by irregular reflection of illumination light during photographing caused by saliva or a narrow intraoral space. The light spot detection means, the light spots are grouped into groups of 4 or 8 connections, and labels are provided to identify each cluster and a portion of the light spot that is not a cluster of light spots. Means for determining the light spot of the noise by calculating the area of the connected cluster of the points, comparing the area with a predetermined threshold value, and determining that the smaller than the threshold value is the noise light spot, The determined lump is provided with means for removing a noise light spot by replacing it with a label of a lump around the lump.
Light spot detection means for detecting the light spot as means for removing other noise light spots, the light spot is expanded to some extent by the expansion filter, and the average value of a certain range of colors around the mass And means for removing the noise light spot by replacing the color of the expanded light spot cluster area with the calculated average color.

  Furthermore, the present invention relates to an image input means for inputting a photographed intraoral image, and from the image input by the image input means, the boundary value between the teeth and the gingiva corresponding to the gingiva, depending on whether the color information is above or below a certain threshold. First boundary detection means to detect, and from the image input by the image input means, a color information determination method different from the first boundary detection means is used, and the boundary is detected based on whether the color information is above or below a certain threshold. The degree of coincidence between the boundary value obtained by the second boundary detection means and the first boundary detection means and the boundary value obtained by the second boundary detection means is calculated by calculation, and the proportion of coincidence increases. As described above, there is provided means for determining the threshold value of the first boundary portion detecting means and the threshold value of the second boundary portion detecting means.

The present invention makes it possible to form an image having a more accurate tooth-gingival boundary portion by combining the noise point removing means and the tooth-gingival boundary detecting means.
In the above two inventions, each configuration may be formed by an electronic circuit such as a gate array, but may be executed by computer software, for example, informed consent by clarification of intraoral data. It is used for use as a medical image, and as an automatic intraoral diagnosis device by automatic recognition of teeth and gingiva.

Hereinafter, an embodiment of the present invention will be described in detail with reference to FIG.
In FIG. 1, 101 is an image input means that can digitally process images obtained from various optically recognizable devices such as digital images taken with a digital camera and digital images captured with an image scanner. It is a means to input in a state.
Reference numeral 102 denotes a light spot detection unit that detects a luminance value, color information, and the like from a digital data value, and detects this portion as a range when the value is a predetermined luminance value and color information.
Reference numeral 103 denotes a light spot range setting unit that detects a point having a high luminance value for each pixel as a point to shine and sets the range. Further, the vicinity of 4 and 8 may be set by connecting with the pixel value of the same light spot.
Reference numeral 104 denotes label adding means for adding labels having different values to the light spot range set by the light spot range setting means 103.
Further, in this background, it is preferable to add a different label to each of teeth and gums. The label is in a state in which the color information of the light spot and the identification code corresponding thereto are linked together.
Reference numeral 105 denotes comparison means, which measures the area of each light spot range set by the light spot range setting means 103, and determines and outputs a noise light spot below a predetermined value.
Reference numeral 106 denotes replacement means, which is a means for replacing the part determined as the noise light spot by the comparison means 105 with a color corresponding to a label attached to the background around the part.

Next, the operation shown in FIG. 1 will be described with reference to FIG.
The digital image of the tooth 32 and the gingiva 31 input by the image input means 101 is the light spot detection means after the luminance value and color of each pixel, pixels at regular intervals, or pixels grouped together are detected. The light spot 33 is detected based on the luminance value and color.
The light spot range setting means 103 connects the neighborhoods of 4 and 8 with pixel values of the same light spot to form a mass (34) (35). Next, an identification code is added for each cluster of light spots by the label adding means 104 (r1 to r8 in FIG. 3C), and labels ra and rb are also added to the background for each different color information. To do. Next, the area of each light spot cluster is calculated, and when it is smaller than a predetermined threshold (34), it is detected as a noise light spot, and this is converted into color information corresponding to the surrounding background label by the replacement means 106. replace. This replacement eliminates the noisy spot.

Another embodiment will be described with reference to FIG.
In FIG. 1B, the description of the same configuration as in FIG.
Reference numeral 107 denotes a light spot range extending means, which is formed by an expansion filter or the like, and forms an extended light spot by extending the light spot range obtained by the light spot detecting means in the same area or in one direction. Means. Further, when the extended light spots are overlapped with each other, it is means for forming an extended light spot with the outline of the overlapping light spots as a whole. Here, the term “overlap” may include a state in which the light spot contours are in contact with each other or are not in contact with each other and the distance is closer. The point which shines for every pixel may be detected, and it may extend by connecting 4 neighborhood and 8 neighborhood with the pixel value of the same light spot with respect to the circumference | surroundings.
Reference numeral 108 denotes ambient color information detection means for detecting color information around the extended light spot range.

Reference numeral 109 denotes a replacement unit. If the extended light spot range obtained by the light spot range extension unit is to be erased, such as a noise light point, the surrounding color information is detected. It is means for replacing the color information obtained by means 108 and removing it. FIG. 3 (d) shows the state. Reference numeral 35 denotes an extended light spot, and 36 denotes a surrounding color information detection site. FIG. 3 is a curve, and the light spot range is described, but actually shows a state where square pixels are gathered.
In the embodiment shown in FIG. 1B, the means for replacing with the background color information shown in FIG. 1A is replaced with surrounding color information. Further, after the light spot is detected by the light spot detecting means 102, the light spot range extending means 107 expands each range by a fixed amount, and sets a state where there is no light spot at an adjacent position. By setting the expanded state, the light spots are not adjacent to each other, the surrounding color information 36 is detected, the color information is replaced with the light spot range 35, and the light spot range is deleted.
The embodiment shown in FIG. 1 is shown in a block diagram and can be realized by an apparatus configuration, but may preferably be realized by software by a computer.

Next, another embodiment will be described in detail with reference to FIG.
In FIG. 2, 201 is an image input means, which has the same configuration as 101, and is a means for inputting a digital image so that it can be digitally processed.
Reference numeral 202 denotes first boundary detection means, which is a boundary between the gingiva and the tooth, and is a color included in the gingiva, and sets a critical color threshold that is difficult to detect whether it is a tooth or gingiva. Is a means for comparing the individual pixel data, connecting the coincident and substantially coincident portions and outputting them as critical contour data.
Reference numeral 203 denotes a second boundary detection unit, for example, a predetermined adjustment threshold value such as a luminance value corresponding to gingival color data is set, and a part that matches or substantially matches this threshold value is detected and connected. Thus, it is means for forming boundary data for adjustment.

Reference numeral 204 denotes difference value detection means for taking the difference between the critical contour data and the calculation adjustment boundary data and outputting the difference boundary data.
Reference numeral 205 denotes a comparison detection unit. When the difference boundary data is equal to or smaller than a predetermined value, this is set as a boundary value. If the difference boundary data is larger than the predetermined value or range, the threshold setting units 206 and 207 are set as threshold values. Output a signal to change the value of.
Reference numeral 207 denotes threshold setting means for changing and adjusting the critical color threshold, and reference numeral 208 denotes threshold setting means for changing and adjusting the adjustment threshold.

Next, the operation of FIG. 2 will be described in detail with reference to FIG.
Based on the critical color threshold value of the gingiva, the digital image of the gingiva and teeth input by the image input unit 201 is based on the critical color threshold value of the gingiva as shown in FIG. Data 41 is formed. Reference numeral 42 denotes actual tooth-gum boundary data.
Further, the second boundary detection unit 203 detects the adjustment boundary data 43 as shown in FIG. 4B based on a predetermined adjustment threshold for the data input from the image input unit 201.

The adjustment boundary data formed by the second boundary detection unit 203 and the critical boundary data formed by the first boundary detection unit 202 are input to the difference value detection unit 204 to form difference value data, and the comparison detection unit 205. Is input.
If the difference value data is smaller than or coincides with the predetermined value, the comparison detection unit 205 outputs the boundary data 44. If the difference value data is larger than the predetermined value, the comparison detection unit 205 outputs to the threshold setting unit 206 and the threshold setting unit 207. A threshold change signal is output to change the boundary color threshold used by the first boundary detection unit 202, and the first boundary detection unit 207 changes the adjustment threshold value of the second boundary detection unit 203. To do. In addition, as described above, one of the boundary color threshold and the adjustment threshold may be adjusted without adjusting both.
After changing the threshold value, the difference value detection unit 204 similarly compares the new critical contour data and the calculation adjustment boundary data, and the comparison detection unit 205 detects and compares the value with a predetermined value. The operation is repeated until the difference data is equal to or lower than the predetermined value.
As described above, in this embodiment, it is possible to accurately detect gingival-tooth boundary data that is difficult to identify.

  The present invention enables accurate image processing of the tooth outline and the gingival-teeth boundary that are difficult to detect in the image processing of the intraoral data, so that a clear intraoral image can be obtained and used in the dental field. Digital images can be used.

101 Image input means 102 Light spot detection means 103 Light spot range setting means 104 Label addition means 105 Comparison means 106 Replacement means

Claims (4)

  Light spot detection means for detecting a light spot, light spot range setting means for setting a light spot as a group of light by four or eight connections, and for identifying each of the light spots and a part of the light spot that is not a light spot Label adding means for labeling, calculating the area of a cluster of connected light spots based on the label, comparing the area with a predetermined threshold value, and determining that the light spot is smaller than the threshold value as a noise light spot Comparison means for comparing and determining noise light spots, and dental image processing provided with replacement means for removing noise light spots by replacing the light spots determined as noise light spots with the surrounding mass labels system. Light spot detection means for detecting a light spot, the light spot is expanded to a certain extent by an expansion filter, the average value of a certain range of colors around the mass is calculated, and the expanded light spot cluster A dental image processing system comprising means for removing a noise light spot by replacing the color of the area of the area with the calculated average color. Image input means for inputting a photographed intraoral image, first boundary detection means for detecting a critical boundary value between teeth and gingiva based on a critical color threshold corresponding to the gingiva from the image input by the image input means; A second boundary detection unit for detecting an adjustment boundary value based on an adjustment threshold value based on a value of a predetermined reference color, which is a color of gingiva, from the image input by the image input unit; An arithmetic means for calculating a predetermined value on the contour value obtained by the detecting means to form an arithmetic boundary value, comparing the arithmetic boundary value with the critical boundary value, comparing the degree of coincidence of the values, and the degree of coincidence is not sufficient 3. The dental image information processing system according to claim 1, further comprising comparison detection means for adjusting one or both of the critical color threshold and the adjustment threshold. 4. The dental image information system according to claim 3, wherein the predetermined reference color is one selected from primary colors such as RGB and CYMK.

Images