CN103335727A - Thermal imaging image processing method based on setting of multiple emissivities for visible light divided area - Google Patents

Abstract

A thermal imaging image processing method based on visible light region segmentation and setting multiple emissivity, comprising the following steps: S1, taking a visible light map and an infrared heat map; S2, using a visible light map, using a segmentation algorithm to divide a target object into n regions, n ≥2, set corresponding emissivity in n regions to obtain n contours; S3, use n contours in step S2 to define or divide the infrared heat map, obtain n regions of the infrared heat map, and divide the infrared heat map The n regions are spliced or synthesized, there will be gaps after splicing or synthesis, and the gaps are smoothed with the help of heat conduction characteristics. After the above steps, for isothermal irregular objects (different in shape, roughness, texture), the heat map taken, after pseudo-color transformation, the color tends to be consistent, and the temperature variance of each area decreases; for non-isothermal objects objects, the measured temperature distribution is more accurate.

Description

A Thermal Imaging Image Processing Method Based on Visible Light Region Segmentation Setting Multiple Emissivity

technical field

The invention relates to a thermal imaging image processing method for setting multiple emissivity based on visible light region division.

Background technique

The infrared imaging technology in foreign countries started earlier, so the infrared temperature measurement technology based on this has also developed rapidly. Foreign countries have been engaged in the development of infrared temperature measurement products since the 1920s. The earliest developed pyrometer implementation method is to eliminate the influence of the real temperature due to the difference in emissivity by comparing the brightness ratio of the target at a selected wavelength. error. This method is not suitable for most materials in general, and will cause relatively large errors, but it is still relatively effective for the measurement of gray bodies. The development of infrared temperature measurement technology in my country is also mainly based on this general working principle. The infrared radiation emitted by the object is obtained through the optical system, and the light in the infrared wavelength range is transmitted through the filtering function of the lens and enters the infrared focal plane. The infrared focal plane converts the detected infrared radiation into a voltage signal, and then processes the digital signal to obtain the thermal image and temperature information of the object.

Thermal imaging temperature measurement is an emerging non-contact temperature measurement technology, which not only obtains the two-dimensional temperature field distribution of the target in real time, but also has functions such as target identification and temperature early warning. Emissivity is one of the factors that affect thermal imaging temperature measurement. In the laboratory environment, a single emissivity is used for relatively regular geometry. In practical application, the target objects are often irregular and have multiple emissivity. Following this "idealized" laboratory method will lead to inevitable errors.

Contents of the invention

The technical problem to be solved by the present invention is to provide a thermal imaging image processing method based on visible light region segmentation and setting multiple emissivity, so as to solve the above-mentioned problems in the background technology.

The technical problem solved by the present invention adopts following technical scheme to realize:

A thermal imaging image processing method based on visible light region segmentation setting multiple emissivity, comprising the following steps:

S1. Using a visible light array detector and an infrared thermal imaging camera to shoot the target object under the same field of view and the same resolution, and obtain a visible light map and an infrared thermal image;

S2. Based on the many complementary characteristics of the visible light map and the infrared heat map, use the visible light map to segment the target object using a segmentation algorithm, and divide it into n areas, n≥2, and set the corresponding emissivity of the n areas:

${\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; /</span>\sqrt[\mathrm{<span\; class="notranslate">\; no</span>}]{\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

T _r is the test temperature, T ₀ is the real temperature, ε is the emissivity.

Obtain n contours, wherein the segmentation algorithm includes but not limited to Candy, watershed method, threshold method, Otsu and combinations thereof;

S3, using the n contours in step S2 to define or divide the infrared heat map to obtain n regions of the infrared heat map, and splicing or synthesizing the n regions of the infrared heat map, there will be gaps after splicing or synthesizing, The gap is smoothed with the help of heat conduction characteristics. The processing methods include but are not limited to based on the heat conduction equation:

$\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; sin</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{<span\; class="notranslate">\; \&infin;</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; n\&pi;x</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span>{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; k\; n</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

in $A^{\mathrm{no}}=2{\&Integral;}_0^1\cos \left(2x\right)\cos \left(\mathrm{n\&pi;x}\right)\mathrm{dx}={(-1)}^{\mathrm{no}}\frac{-4\sin \left(2\right)}{{\mathrm{no}}^2{\mathrm{\&pi;}}^2-4}$

and its variants or derivatives, radial diffusion method

The idea of the radial diffusion algorithm is: expand the contour line to n widths (the original width is 1 pixel), n is an odd number, and the specific number of n is determined when it can just cover all the gaps in the stitching. In this paper, n=5; Circulate point by point along the original contour line, and find the line segment Li of length n perpendicular to the contour line at point i (the node coordinates of Li are from outside to inside: Ci1...Cin, and the contour line is at the midpoint Cin/2+1), From the problem setting, we can get TCi1=T2, TCin=T1, divide the line segment Li into n-1 equal parts, and then we can deduce the temperature of other points.

After the above steps, for isothermal irregular objects (different in shape, roughness, texture), the heat map taken, after pseudo-color transformation, the color tends to be consistent, and the temperature variance of each area decreases; for non-isothermal objects objects, the measured temperature distribution is more accurate.

In the present invention, the visible light array detector and the infrared thermal imaging camera are required to be consistent in terms of focal length, physical shape of the lens, and optical characteristics.

In the present invention, the infrared thermal imager is preferably an online monitoring infrared thermal imager produced by Zhuhai Yiduo Monitoring Technology.

In the present invention, the infrared thermal imager adapts to a wavelength of 2-14 μm.

In the present invention, the visible light area array detector adopts a color CCD or a black and white CCD.

In the present invention, in step S2, it is preferable to use a canny detector to segment the visible light image.

In the present invention, in step S3, when smoothing the gap, it is preferable to adopt the radial diffusion method.

Due to the adoption of the above technical solutions, the present invention has the following beneficial effects:

The main content of the present invention is to provide a set of methods for improving the measured temperature error due to multiple emissivity of irregular target objects, which is used for thermal imaging temperature measurement of complex target objects.

When the instrument is set to a single emissivity, no matter from the theoretical calculation or the actual observation effect, when the actual emissivity is higher than the emissivity set by the instrument, the measured temperature will be higher, and vice versa, referred to as Gauguin Higher, lower lower. When the target is known to be isothermal, its infrared heat map (after pseudo-colorization) should ideally be the same color. If the emissivity of one area is higher than the set emissivity, the emissivity of the other area is lower than When the set emission is performed, the infrared heat map will have a large color difference, as shown in Figure (2). The color temperature at this time cannot reflect the real temperature of the object. The solution adopted by the invention is to use a method of setting multiple emissivity by area based on visible light segmentation for the same object to be measured, so as to make an isochromatic infrared heat map of an object with known isothermal temperature. The purpose of this method is to set the corresponding emissivity according to different regions through the improved region growing algorithm. In view of the characteristics of the algorithm itself, it will make the adjacent parts of the regions unable to fit together. The present invention adopts the radial diffusion method according to the heat conduction characteristics to make the junction Smooth transition. In this way, the heat map is more uniform and smooth to almost a whole, and the objective evaluation is consistent with the subjective evaluation. The final thermal image more accurately reflects the true temperature of the target object.

The method is suitable for the diagnosis of crop diseases and insect pests, and can also be applied to temperature measurement and medical diagnosis of industrial equipment, which can avoid or reduce misdiagnosis and misjudgment; in infrared telemetry of ground objects, it can make it easier to distinguish and identify ground objects.

Description of drawings

Figure 1 is the VI diagram of a 20*10*5mm aluminum block taken in a general laboratory environment;

Figure 2 is the IR image of a 20*10*5mm aluminum block taken in a general laboratory environment;

Figure 3 shows the edge map extracted by the canny detector in the MATLAB environment;

Fig. 4 is an effect diagram after superimposing Fig. 3 in Fig. 2 for region definition;

Figure 5 and Figure 6 are two renderings of the corrected gray-scale heat map converted to a pseudo-color map under the same detector due to the different fitting equations between the gray scale and temperature of the detector;

Fig. 7 is the overlay diagram of Fig. 5 and Fig. 6;

Figure 8 shows the effect diagram after the smooth transition of "color temperature" in the MATLAB environment, that is, the final effect diagram of this method.

Detailed ways

In order to make the technical means, creative features, goals and effects achieved by the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

A thermal imaging image processing method based on visible light region segmentation setting multiple emissivity, comprising the following steps:

S1. Use visible light array detectors and infrared thermal imaging cameras to shoot target objects in the same field of view and the same resolution. Among them, the visible light area array detectors use color CCDs, and the infrared thermal imaging cameras are produced by Zhuhai Yiduo Monitoring Technology. The on-line monitoring infrared thermal imager has the same requirements on the focal length, physical shape of the lens, and optical characteristics. The infrared thermal imager adapts to a wavelength of 2-14 μm, and the visible light map and infrared heat map are obtained as follows:

Use the online monitoring thermal imaging camera (Flir core, Daheng processor) produced by Zhuhai Yiduo Monitoring Technology, the default emissivity is 0.95, the color output mode is Fusion, the indoor is a general laboratory environment, and the indoor temperature is 20°C , the relative humidity is 65%, and the doors and windows are closed, it is considered that there is no air convection in the room. The size of the aluminum block is 20*10*5mm, the surface is frosted, and the middle is painted into an irregular shape with paint. After being placed indoors for 2 hours (considered as heat exchange balance), the same field of view is taken. Figure 1 and Figure 2 are the visible light map and infrared heat map taken in the same field of view, with a resolution of 320*240;

S2. Based on the mutual characteristics of the visible light map and the infrared heat map, use the visible light map to segment the target object using a segmentation algorithm, and divide it into n regions, n≥2,

Set the corresponding emissivity for n regions respectively:

${\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; T</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; /</span>\sqrt[\mathrm{<span\; class="notranslate">\; no</span>}]{\mathrm{<span\; class="notranslate">\; \&epsiv;</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

T _r is the test temperature, T ₀ is the real temperature, ε is the emissivity, where n=4, this is the theoretical basis, derived by the thermal radiation company.

Obtain n contours, wherein the segmentation algorithm includes but not limited to Candy, watershed method, threshold method, Otsu and their combinations, all of which are known algorithms;

as follows:

From the experimental conditions, it can be seen that all parts of the aluminum block are at the same temperature. The temperature of the black paint in the middle is obviously higher than that of the surrounding blank area through visual inspection of the control temperature bar. The temperature difference of the instrument shows 4.4 °C. Caused by different emissivity. Using the following method, you can make an isochromatic infrared heat map of an object with known isothermal temperature;

The segmentation of the black area in Figure 1 is realized by using the canny detector. Spray black paint (emissivity Es=1) evenly on the surface of the target object, and then adjust the emissivity of the infrared thermal imaging camera until the temperature of the surface without painting is the same as or close to the temperature of the painted surface to obtain the emission of the non-painted part of the target object Rate. Using the above experimental method, the emissivity of the non-painted part can be obtained as Er=0.92. According to the two different emissivity Es, Er and the original grayscale image, the corrected grayscale heat map can be reversely deduced, and then converted into a pseudo-color image. The painted part will be shifted to blue, and the non-painted part will be shifted to orange. The effect is shown in Figure 5 and 6. After superposition, it is Figure 7. Visually, the color difference between the two areas is significantly reduced, and the color of the pseudo-color image tends to be consistent. However, there are obvious defects at the boundary, which is caused by the pixels at the boundary neither belonging to Figure 5 nor Figure 6.

S3, using the n contours in step S2 to define or divide the infrared heat map to obtain n regions of the infrared heat map, and splicing or synthesizing the n regions of the infrared heat map, there will be gaps after splicing or synthesizing, The gap is smoothed with the help of heat conduction characteristics. The processing methods include but are not limited to based on the heat conduction equation:

$\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; sin</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{<span\; class="notranslate">\; \&infin;</span>}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; no</span>}}\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; n\&pi;x</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; exp</span>}<span\; class="notranslate">\; (</span>{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; k\; n</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

in $A^{\mathrm{no}}=2{\&Integral;}_0^1\cos \left(2x\right)\cos \left(\mathrm{n\&pi;x}\right)\mathrm{dx}={(-1)}^{\mathrm{no}}\frac{-4\sin \left(2\right)}{{\mathrm{no}}^2{\mathrm{\&pi;}}^2-4}$

and its variants or derivatives, the radial diffusion method,

The idea of the radial diffusion algorithm is: expand the contour line to n widths (the original width is 1 pixel), n is an odd number, and the specific number of n is determined by just covering all the gaps in the stitching; point by point along the original contour line Loop, find the line segment Li of length n perpendicular to the contour line at point i (the node coordinates of Li are from outside to inside: Ci1...Cin, and the contour line is at the midpoint of Cin/2+1), TCi1 can be obtained from the problem setting =T2, TCin=T1, divide the line segment Li into n-1 equal parts, and the temperature of other points can be deduced.

as follows:

The situation in Figure 7 is simulated by using the heat conduction process: the heat conduction has reached a certain steady state, that is, the temperature in the transition zone from the high temperature zone to the low temperature zone does not change any more. It is assumed that T2 corresponds to the high temperature of the painted zone, and T1 corresponds to the low temperature of the non-painted zone. Temperature, the temperature in the transition zone should transition smoothly from T2 to T1 along the curve of the heat conduction equation from the inside to the outside. As for the curve at which time point to choose, which section of the curve to choose is uncertain, because the difference between T1 and T2 is relatively small, you can Radial diffusion is used along the contour to achieve smooth transitions. The effect after processing is shown in Figure 8.

Referring to the characteristic attributes of texture statistics, information entropy, standard deviation, smoothness, contrast third moment, and consistency are used to evaluate the grayscale components in Figure 2 and Figure 8. Table (1) was measured using the crop digital image analysis system (V2.0) developed by the Agricultural Information Institute of Hunan Agricultural University

Table 1 Evaluation parameter test comparison table

The results show that in Figure 8 compared with Figure 2, the entropy value becomes smaller, the standard deviation becomes smaller, the contrast becomes smaller, the smoothness becomes smaller, the third-order distance becomes smaller, and the consistency increases, indicating that the heat map is more uniform, smoother and more consistent, that is The heat map of the aluminum block after setting the emissivity by area is almost a whole, and the objective evaluation is consistent with the main evaluation.

After the above steps, for isothermal irregular objects (different in shape, roughness, texture), the heat map taken, after pseudo-color transformation, the color tends to be consistent, and the temperature variance of each area decreases; for non-isothermal objects objects, the measured temperature distribution is more accurate.

The basic principles and main features of the present invention and the advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments, and what described in the above-mentioned embodiments and the description only illustrates the principles of the present invention, and the present invention will also have other functions without departing from the spirit and scope of the present invention. Variations and improvements are possible, which fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents.

Claims (7)

1. cut apart the graphic images disposal route that many emissivity are set based on the visible region for one kind, it is characterized in that: comprise the steps:

S1, employing visible light planar array detector and thermal infrared imager are taken target object under same visual field, same resolution, obtain visible light figure and infrared chart;

S2, many mutually by characteristic based on visible light figure and infrared chart, utilize visible light figure, adopt partitioning algorithm to cut apart to target object, be divided into n zone, n 〉=2, n zone arranges corresponding emissivity respectively and obtains n profile, and wherein, partitioning algorithm is including but not limited to Candy, watershed method, threshold method, Otsu and combination thereof;

S3, utilize n profile among the step S2 that infrared chart is defined or divides, obtain n zone of infrared chart, and with n of infrared chart zone splicing or synthetic, can there be break joint after splicing or synthesizing, to break joint by thermal conduction characteristic, carry out smoothing processing, the method for processing including but not limited to based on heat-conduction equation and distortion thereof or derive, the radial diffusion method.

2. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: described visible light planar array detector and thermal infrared imager all require consistent on the physical form of focal length, camera lens, optical characteristics.

3. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: the on-line monitoring thermal infrared imager that thermal infrared imager selects for use Zhuhai more than one monitoring science and technology to produce.

4. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: the wavelength that thermal infrared imager adapts to is 2～14 μ m.

5. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: described visible light planar array detector adopts colored CCD or black-white CCD.

6. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: among the step S2, utilize the canny detecting device that visible light figure is cut apart.

7. according to claim 1ly a kind ofly cut apart the graphic images disposal route that many emissivity are set based on the visible region, it is characterized in that: among the step S3, when break joint is carried out smoothing processing, adopt the radial diffusion method to handle.

Images