TWI744785B - Method for recognizing and locating object basde on thermal image - Google Patents

Abstract

A method for recognizing and locating an object based on a thermal image of the object includes the following steps: loading information of the thermal image of the object, wherein the information includes plural temperature values corresponding to plural pixels arranged in a matrix of n rows and m columns; performing a horizontal edge judgement on the temperature values corresponding to the pixels located at i column, thereby obtaining a left edge and a right edge of the object, in which i=1~M; performing a vertical edge judgement on the temperature values corresponding to the pixels located at j row and located between the left edge and the right edge, thereby obtaining a top edge and a bottom edge of the object, in which j=1~N; combining the left edge, the right edge, the top edge, and the bottom edge so as to obtain location of the object.

Description

Object identification and positioning method based on thermal image

The embodiment of the disclosure relates to an object recognition and positioning method, and particularly relates to a thermal image-based object recognition and positioning method.

Generally speaking, identifying and locating the position of the object in the image is by analyzing the visible light image, which uses the pixel feature values in the image to analyze the position of the object through calculation. The specific method is to first set the identification conditions corresponding to the object, so as to find the pixels that meet the conditions in the obtained visible light image. For example, in the color image, use the value of RGB, HSV, YUV, etc. as the identification Condition, use the single value threshold (0~255) setting in the grayscale image, through the full image search, mark the pixels in the visible image that meet the identification conditions, and then design a specific method for subsequent object identification calculations , Find the most suitable area and mark the position of the object.

However, if the light source in the environment where the object is located is insufficient or the variation is large, the visible light image cannot be used for identification. In addition, if there are objects with recognizable characteristics around the recognized object, or extremely close to or overlap with the object to be recognized, it will cause recognition errors. In addition, if the background is complex and there are many edges that do not belong to the ontology, it will also cause poor recognition results.

Another way to identify and locate the position of an object in the image is to analyze the thermal image of the object. However, if the object is partially obscured, has a complex background, or the image information is defective, it will cause recognition errors and lead to the position of the object Misjudgment.

The purpose of this disclosure is to provide a thermal image-based object identification and positioning method, including: loading two-dimensional thermal image information of the object, wherein the two-dimensional thermal image information includes a plurality of corresponding rows arranged in M rows and N rows. Multiple temperature values of the pixel, where M and N are positive integers; perform horizontal boundary judgment on the N temperature values of the N pixels located in the i-th row to obtain the left and right boundaries of the object, where i=1~ M; Perform vertical boundary judgment on the multiple temperature values of the multiple pixels located in the j-th column and between the left boundary and the right boundary to obtain the upper boundary and the lower boundary of the object, where j=1~N ; And combine the left border, right border, upper border and lower border to get the position of the object.

In some embodiments, the horizontal boundary judgment is based on the temperature range, the number threshold, the dispersion threshold, and the width threshold; wherein the judgment is based on the temperature range, the number threshold, the dispersion threshold, and the height threshold. Vertical boundary judgment.

In some embodiments, the determination of the horizontal boundary of the N temperature values of the N pixels located in the i-th row includes: [Step S11]: determining the N temperature values of the N pixels located in the i-th row Among them, whether the number of temperature values that meet the temperature range is greater than the number threshold; in [Step S11], if the number is greater than the number threshold, go to [Step S12]: Determine N of the N pixels in the i-th row Whether the dispersion degree of the temperature value is greater than the dispersion degree threshold; in [Step S12], if the dispersion degree is greater than the dispersion threshold, go to [Step S13]: determine whether the left boundary has been found; in [Step S13], if If the left boundary is not found, the i-th row is the left boundary, and enter [Step S14]: add 1 to i, and return to [Step S11].

In some embodiments, the determination of the horizontal boundary of the N temperature values of the N pixels located in the i-th row includes: in [Step S11], if the number is not greater than the number threshold, proceed to [Step S15 ]: Determine whether i is equal to M; in [step S15], if i is not equal to M, go back to [step S14]; in [step S15], if i is equal to M, then the Mth row is the right boundary, And end the horizontal boundary judgment.

In some embodiments, the determination of the horizontal boundary of the N temperature values of the N pixels in the i-th row includes: in [Step S12], if the degree of dispersion is not greater than the threshold of the degree of dispersion, enter [ Step S16: Determine whether the left boundary has been found; in [Step S16], if the left boundary is not found, go back to [Step S14].

In some embodiments, the determination of the horizontal boundary of the N temperature values of the N pixels in the i-th row includes: in [Step S16], if the left boundary has been found, then proceed to [Step S17]: Determine whether the distance between the i-th row and the left border is greater than the width threshold; in [Step S17], if the distance is not greater than the width threshold, go back to [Step S15]; in [Step S17], if the distance is greater than the width threshold Value, the i-th row is the right boundary, and the horizontal boundary judgment ends.

In some embodiments, the vertical boundary determination performed on the plurality of temperature values of the plurality of pixel points located in the j-th column and located between the left boundary and the right boundary includes: [Step S21]: determining that it is located Among the plurality of temperature values of the plurality of pixel points in the jth column and located between the left border and the right border, whether the number of the plurality of temperature values that meet the temperature range is greater than the number threshold; in [Step S21 ], if the number is greater than the number threshold, proceed to [Step S22]: determine whether the dispersion degree of the temperature values of the plurality of pixel points located in the j-th column and between the left boundary and the right boundary is greater than Discrete degree threshold; In [Step S22], if the degree of dispersion is greater than the threshold of Discrete degree, proceed to [Step S23]: Determine whether the upper boundary has been found; In [Step S23], if the upper boundary is not found, then proceed to [Step S23]: Column j is the upper boundary, and enter [Step S24]: add 1 to j, and return to [Step S21].

In some embodiments, the vertical boundary determination performed on the plurality of temperature values of the plurality of pixel points located in the j-th column and located between the left boundary and the right boundary includes: in [step S21], If the quantity is not greater than the quantity threshold, proceed to [Step S25]: Determine whether j is equal to N; In [Step S25], if j is not equal to N, go back to [Step S24]; In [Step S25], if If j is equal to N, the Nth column is the lower boundary, and the vertical boundary judgment ends.

In some embodiments, the determination of the vertical boundary of the plurality of temperature values of the plurality of pixel points located in the j-th column and between the left boundary and the right boundary includes: in [step S22], If the degree of dispersion is not greater than the threshold of the degree of dispersion, proceed to [Step S26]: determine whether the upper boundary has been found; in [Step S26], if the upper boundary is not found, return to [Step S24].

In some embodiments, the determination of the vertical boundary of the plurality of temperature values of the plurality of pixel points located in the j-th column and between the left boundary and the right boundary includes: in [step S26], If the upper boundary has been found, proceed to [Step S27]: Determine whether the distance between the j-th column and the upper boundary is greater than the height threshold; in [Step S27], if the distance is not greater than the height threshold, go back to [Step S25] ; In [Step S27], if the distance is greater than the height threshold, the j-th column is the lower boundary, and the vertical boundary judgment is ended.

In order to make the above-mentioned features and advantages of the present disclosure more obvious and understandable, the following specific embodiments are described in detail in conjunction with the accompanying drawings.

The embodiments of the present invention are discussed in detail below. However, it can be understood that the embodiments provide many applicable concepts, which can be implemented in various specific contents. The discussed and disclosed embodiments are for illustrative purposes only, and are not intended to limit the scope of the present invention.

The technology disclosed in this disclosure uses a thermal image sensor to determine the position of an object in a thermal image that meets a specified temperature range. The present disclosure can be applied to the identification method used in the temperature detection system of the ladle (LD) wall in a steelmaking plant. The thermal image-based object identification and positioning method proposed in this disclosure will be able to correctly find the steel The actual position of the barrel completes the identification and positioning of the steel barrel.

FIG. 1 is a flowchart of a method 1000 for identifying and locating objects based on thermal images according to an embodiment of the present disclosure. The method 1000 for identifying and positioning objects based on thermal images includes steps 1100, 1200, 1300, and 1400. In step 1100, the two-dimensional thermal image information of the object is loaded, where the two-dimensional thermal image information includes a plurality of temperature values corresponding to a plurality of pixels arranged in M rows and N columns, where M and N are positive integers.

In step 1200, based on the temperature range T _range , the number threshold T _th , the dispersion threshold D _th and the width threshold W _th, perform horizontal boundary judgment on the N temperature values of the N pixels in the i-th row to obtain The left and right boundaries of the object, where i=1~M.

In step 1300, based on the temperature range T _range , the number threshold T _th , the discrete degree threshold D _th and the height threshold H _th , a plurality of pixels located in the j-th column and located between the left boundary and the right boundary of the object are aligned Perform vertical boundary judgment on multiple temperature values of, to obtain the upper boundary and lower boundary of the object, where j=1~N.

In step 1400, the left boundary, the right boundary, the upper boundary and the lower boundary of the object are combined to obtain the position of the object.

2 is a flowchart of the horizontal boundary determination in step 1200 of the thermal image-based object recognition and positioning method 1000 according to an embodiment of the present disclosure. The horizontal boundary determination in step 1200 includes steps S10 to S19. In step S10, i is set to 0. Next, in step S14, i is accumulated by 1, that is, i=i+1.

_{Next, in step S11, it is determined whether the number T i} of the temperature values in the _{temperature range T range} among the N temperature values of the N pixels in the i-th row is greater than the number threshold T _th . Specifically, the temperature range T _{range in} step S11 is the temperature range range set by the user to which the object to be tested fits. _{In other words, step S11 is to find the number T i} of pixels in the i-th row that meet the specified temperature interval, and determine whether the number T _i is greater than the number threshold T _th .

In step S11, if the number T _{i is} greater than the number threshold T _th , proceed to step S12 to determine whether the dispersion degree Di of the N temperature values of the N pixels in the i-th row is greater than the dispersion degree threshold D _th . Specifically, if it is determined in step S12 that the dispersion degree _{Di is} greater than the dispersion degree threshold D _th , it means that the N temperature values in the i-th row may include the temperature value of the object and other background temperature values.

In step S12, if the dispersion degree D _{i is} greater than the dispersion degree threshold value D _th , proceed to step S13: determine whether the left boundary has been found. In step S13, if the left boundary is not found, proceed to step S18: the i-th row is the left boundary, and then return to step S14.

In step S11, if the number T _{i is} not greater than the number threshold T _th , proceed to step S15 (node A through the relay): determine whether i is equal to M. In step S15, if i is not equal to M, go back to step S14. In step S15, if i is equal to M, proceed to step S19: the i-th row is the right boundary.

In step S12, if the dispersion degree _{Di is} not greater than the dispersion degree threshold value D _th , proceed to step S16: determine whether the left boundary has been found. In step S16, if the left boundary is not found, return to step S14. In step S16, if the left boundary has been found, proceed to step S17: determine whether the distance between the i-th row and the left boundary is greater than the width threshold W _th . For example, if the object to be tested is covered by other low-temperature objects and the continuity that meets the temperature condition is interrupted, the position of the object to be tested will not be correctly identified. Therefore, the present disclosure uses the width threshold W _th to improve the performance. The object to be tested is covered by other low-temperature objects and the position of the object to be tested cannot be correctly identified.

In step S17, if the threshold distance is not greater than the width W _th, the process returns to step S15 (via the relay node A); in step S17, if the distance is greater than the width of the threshold W _th, the process proceeds to step S19: The i-th row is the right boundary.

3 is a flowchart of the vertical boundary determination in step 1300 of the thermal image-based object recognition and positioning method 1000 according to an embodiment of the present disclosure. The vertical boundary determination in step 1300 includes steps S20 to S29. In step S20, j is set to 0. Then, in step S24, j is accumulated by 1, that is, j=j+1.

Next, in step S21, it is determined whether the number of temperature values T _{j that conform to the temperature range T range} among the multiple temperature values of the multiple pixels located in the j-th column and between the left and right boundaries of the object is greater than the quantity threshold The value T _th . Specifically, the temperature range T _{range in} step S21 is the temperature range range set by the user to which the object to be tested fits. _{In other words, step S21 is to find out the number T j} of pixels in the j-th column and between the left and right edges of the object that meet the specified temperature interval, and determine whether the number T _j is greater than the number threshold T _th .

In step S21, if the number T _{j is} greater than the number threshold T _th , proceed to step S22 to determine the degree of dispersion of the temperature values of the pixels located in the j-th column and between the left and right edges of the object Whether D _j is greater than the dispersion degree threshold D _th . Specifically, if it is determined in step S22 that the degree of dispersion D _{j is} greater than the threshold value D _{th of the} degree of dispersion, it represents multiple temperatures of multiple pixels located in the j-th column and between the left and right edges of the object. The value may include the temperature value of the object and other background temperature values.

In step S22, if the dispersion degree D _{j is} greater than the dispersion degree threshold D _th , proceed to step S23: determine whether the upper boundary has been found. In step S23, if the upper boundary is not found, proceed to step S28: the jth column is the upper boundary, and then return to step S24.

In step S21, if the number T _{j is} not greater than the number threshold T _th , proceed to step S25 (node B through the relay): determine whether j is equal to N. In step S25, if j is not equal to N, return to step S24. In step S25, if j is equal to N, proceed to step S29: the jth column is the lower boundary.

In step S22, if the dispersion degree D _{j is} not greater than the dispersion degree threshold D _th , proceed to step S26: determine whether the upper boundary has been found. In step S26, if the upper boundary is not found, return to step S24. In step S26, if the upper boundary has been found, proceed to step S27: determine whether the distance between the j-th column and the upper boundary is greater than the height threshold H _th . For example, if the object to be tested is covered by other low-temperature objects and the continuity that meets the temperature condition is interrupted, the position of the object to be tested will not be correctly identified. Therefore, the present disclosure uses the height threshold H _th to improve the performance. The object to be tested is covered by other low-temperature objects and the position of the object to be tested cannot be correctly identified.

In step S27, if the distance is not greater than the height threshold H _th , go back to step S25 (passing the relay node B); in step S27, if the distance is greater than the height threshold H _th , then go to step S29: The jth column is the lower boundary.

Specifically, the present disclosure is based on multiple judgment conditions (temperature range, quantity threshold, dispersion threshold, width threshold, and height threshold) as the condition basis for object identification. This disclosure does not use a single judgment condition as the basis for identifying the object. The condition basis for object identification, therefore, the present disclosure can solve the problem that partial occlusion, complicated background, or defective image information, etc. thermal images will cause identification errors and misjudge the position of the object. For example, if the object to be tested is covered by other low-temperature objects and the continuity that meets the temperature condition is interrupted, using a single judgment condition as the condition for object identification will not be able to correctly identify the location of the object to be tested. This disclosure uses A combination of multiple judgment conditions is used as the condition for object identification, so as to improve the problem that when the object to be tested is covered by other low-temperature objects, the position of the object to be tested cannot be correctly identified. For example, in the case of a complex background, only the temperature range is used as the condition for object recognition. If there are other objects that meet the temperature range and overlap with the object to be measured, other objects will also be recognized As the object to be tested, this disclosure uses a combination of multiple judgment conditions as the condition for object identification, thereby improving when there are other objects in the surrounding temperature range that overlap with the object to be tested and the object to be tested cannot be correctly identified The position of the object.

In summary, this disclosure proposes a thermal image-based object recognition and positioning method, which can identify and locate the object to be tested by using two-dimensional thermal image information, even if the object to be tested is partially obscured, has a complex background or thermal image When the information is defective, the object to be tested can still be correctly identified and located. The thermal image-based object identification and positioning method proposed in this disclosure can not only improve the recognition accuracy, but also reduce the computational complexity and speed up the identification speed. For the M*N two-dimensional thermal image, this disclosure only needs to be less than 2*( M*N) calculation can find the object to be tested. Therefore, the calculation amount of the thermal image-based object identification and positioning method proposed in this disclosure is more efficient and faster than other algorithms.

The features of several embodiments are summarized above, so those who are familiar with the art can better understand the aspect of the present disclosure. Those who are familiar with this technique should understand that they can easily use the present disclosure as a basis to design or modify other processes and structures, thereby achieving the same goals and/or the same advantages as the embodiments described herein. . Those who are familiar with this art should also understand that these equivalent constructions do not depart from the spirit and scope of this disclosure, and they can make various changes, substitutions and alterations without departing from the spirit and scope of this disclosure.

1000: Object recognition and positioning method based on thermal image 1100, 1200, 1300, 1400: steps S10~S19, S20~S29: steps A, B: Node

From the following detailed description in conjunction with the accompanying drawings, a better understanding of the aspect of the present disclosure can be obtained. It should be noted that, according to industry standard practice, each feature is not drawn to scale. In fact, in order to make the discussion clearer, the size of each feature can be increased or decreased arbitrarily. [Fig. 1] is a flowchart of an object recognition and positioning method based on thermal images according to an embodiment of the present disclosure. [Fig. 2] is a flowchart of the horizontal boundary determination of the thermal image-based object recognition and positioning method according to the embodiment of the present disclosure. [FIG. 3] It is a flowchart of the vertical boundary determination of the thermal image-based object identification and positioning method according to the embodiment of the present disclosure.

Domestic deposit information (please note in the order of deposit institution, date and number) without Foreign hosting information (please note in the order of hosting country, institution, date, and number) without

1000: Object recognition and positioning method based on thermal image

1100, 1200, 1300, 1400: steps

Claims (8)

A thermal image-based object identification and positioning method includes: loading one-dimensional thermal image information of an object, wherein the two-dimensional thermal image information includes a plurality of pixels corresponding to a plurality of pixels arranged in M rows and N columns Temperature values, where M and N are positive integers; perform a horizontal boundary judgment on the N temperature values of the N pixels in the i-th row to obtain a left boundary and a right boundary of the object, where i =1~M; perform a vertical boundary judgment on the temperature values of the pixels located in the jth column and between the left and right boundaries to obtain an upper boundary and a lower boundary of the object, where j =1~N; and combine the left boundary, the right boundary, the upper boundary and the lower boundary to obtain the position of the object; among them, it is based on a temperature range, a quantity threshold, a dispersion threshold, and a width The horizontal boundary is determined based on the threshold; wherein, the vertical boundary is determined based on the temperature range, the number threshold, the discrete degree threshold, and a height threshold; wherein the vertical boundary judgment is performed on the N of the i-th row The determination of the horizontal boundary performed by the N temperature values of the pixels includes: [Step S11]: Determine the N temperature values of the N pixels located in the i-th row, which meets the temperature range Whether a first number of the temperature values is greater than the number threshold; in [Step S11], if the first number is greater than the number threshold, then Enter [Step S12]: Determine whether a first degree of dispersion of the N temperature values of the N pixels in the i-th row is greater than the threshold of the degree of dispersion; in [Step S12], if the first degree of dispersion If the degree of dispersion is greater than the threshold of the degree of dispersion, proceed to [Step S13]: determine whether the left boundary has been found; in [Step S13], if the left boundary is not found, the i-th row is the left boundary, and enter [Step S14]: Add 1 to i, and return to [Step S11]. The thermal image-based object identification and positioning method according to claim 1, wherein the determination of the horizontal boundary of the N temperature values of the N pixels in the i-th row further includes: in [Step In S11], if the first quantity is not greater than the quantity threshold, proceed to [Step S15]: Determine whether i is equal to M; In [Step S15], if i is not equal to M, go back to [Step S14]; In [Step S15], if i is equal to M, the M-th row is the right boundary, and the horizontal boundary determination ends. The thermal image-based object identification and positioning method according to claim 2, wherein the determination of the horizontal boundary of the N temperature values of the N pixels in the i-th row further includes: in [step In S12], if the first degree of dispersion is not greater than the threshold of the degree of dispersion, proceed to [Step S16]: determine whether the left boundary has been found; in [Step S16], if the left boundary is not found, return to [step S14]. The method for identifying and positioning objects based on thermal images as described in claim 3, wherein the determination of the horizontal boundary of the N temperature values of the N pixels in the i-th row further includes: in [Step In S16], if the left boundary has been found, proceed to [Step S17]: determine whether a distance between the i-th row and the left boundary is greater than the width threshold; in [Step S17], if the distance is not greater than the width Return to [Step S15] in [Step S17]; in [Step S17], if the distance is greater than the width threshold, the i-th row is the right boundary, and the horizontal boundary determination ends. The thermal image-based object recognition and positioning method according to claim 1, wherein the vertical boundary is performed on the temperature values of the pixels located in the j-th column and between the left boundary and the right boundary The determination includes: [Step S21]: Determine a second number of the temperature values of the temperature values within the temperature range among the temperature values of the pixels located in the j-th column and between the left border and the right border Is it greater than the number threshold; in [step S21], if the second number is greater than the number threshold, proceed to [step S22]: determine the j-th column and between the left boundary and the right boundary Whether a second degree of dispersion of the temperature values of the pixels is greater than the threshold of the degree of dispersion; in [Step S22], if the second degree of dispersion is greater than the threshold of the degree of dispersion Value, enter [Step S23]: Determine whether the upper boundary has been found; In [Step S23], if the upper boundary is not found, the jth column is the upper boundary, and enter [Step S24]: Set j Add 1 and return to [Step S21]. The thermal image-based object identification and positioning method according to claim 5, wherein the vertical boundary is performed on the temperature values of the pixels located in the j-th column and between the left boundary and the right boundary The judgment further includes: in [step S21], if the second quantity is not greater than the quantity threshold, proceed to [step S25]: judge whether j is equal to N; in [step S25], if j is not equal to N, then Go back to [Step S24]; In [Step S25], if j is equal to N, the Nth column is the lower boundary, and the vertical boundary judgment is ended. The thermal image-based object recognition and positioning method according to claim 6, wherein the vertical boundary is performed on the temperature values of the pixels located in the j-th column and between the left boundary and the right boundary The judgment further includes: in [step S22], if the second degree of dispersion is not greater than the threshold of the degree of dispersion, proceed to [step S26]: determine whether the upper boundary has been found; in [step S26], if it is not found This upper boundary returns to [Step S24]. Thermal image-based object recognition and The positioning method, wherein the determination of the vertical boundary of the temperature values of the pixels located in the j-th column and between the left boundary and the right boundary further includes: in [step S26], if it has been found For the upper boundary, proceed to [Step S27]: determine whether a distance between the j-th column and the upper boundary is greater than the height threshold; in [Step S27], if the distance is not greater than the height threshold, return to [ Step S25]; In [Step S27], if the distance is greater than the height threshold, the jth column is the lower boundary, and the vertical boundary judgment is ended.

Images