Intelligent positioning method for abnormal temperature of solid waste based on infrared image

Abstract

The present application relates to the field of image processing, more particularly, the present application relates to a solid waste abnormal temperature intelligent positioning method based on infrared image, the method comprises: first, screening candidate hot spots through temperature threshold, and obtaining the synchronous visible light image and the historical temperature sequence. Then, based on the static optical-thermodynamic coupling characteristics and the dynamic time domain thermal inertia, the artifact confidence is calculated for each pixel in the hot spot. Subsequently, the overall real risk score of the candidate hot spot is obtained by comprehensively considering the confidence of all pixels in the region. Finally, the score is compared with the preset threshold to accurately determine and locate the real abnormal heat source. The present application fuses multi-modal information, can effectively distinguish real heat source and artifact, and significantly reduces the false positive rate.

Description

Technical Field

[0001] This invention relates to the field of image processing. More specifically, this invention relates to a method for intelligent location of abnormal temperatures in solid waste based on infrared images. Background Technology

[0002] Solid waste disposal sites, such as landfills and industrial waste dumps, are crucial infrastructure for modern society, but they also harbor significant safety hazards. Due to the complex composition of waste, containing large amounts of combustible materials, and the potential for anaerobic fermentation and chemical reactions that generate heat, the internal temperature can rise, easily leading to spontaneous combustion or fires. Such fires not only cause severe economic losses and environmental pollution but can also threaten the lives of people in the surrounding area. Therefore, 24 / 7, high-precision temperature anomaly monitoring of solid waste disposal sites and the timely detection of potential fire hotspots are core aspects of safety management.

[0003] Infrared thermal imaging technology has become the mainstream technology for monitoring temperature distribution in such scenarios due to its advantages of non-contact, long-distance, and wide-area temperature measurement. Traditional monitoring systems typically identify high-temperature points by setting a fixed or simple dynamic temperature threshold on the infrared thermal image. When the temperature of a certain area in the image exceeds the threshold, the system triggers an alarm. However, this identification method, which relies solely on a single temperature dimension, has exposed many drawbacks in practical applications, the core problem being an extremely high false alarm rate.

[0004] The primary cause of false alarms is the prevalence of reflection artifacts. In outdoor environments, sunlight is the biggest source of interference. When sunlight shines on the surfaces of certain objects in the storage yard, such as metal fragments, glass bottles, plastic films, or damp surfaces, specular reflection occurs. This reflected solar radiation, when received by infrared thermal imagers, forms extremely hot bright spots on the image, with readings far exceeding the surrounding ambient temperature, leading the system to incorrectly identify them as abnormal hotspots. The temperature characteristics of these artifacts are remarkably similar to those of real early-stage fires, making them difficult to distinguish based on temperature data alone, resulting in frequent false alarms triggered by security personnel.

[0005] In summary, how to eliminate the interference of reflection artifacts and achieve accurate heat source detection has become the research focus of this scheme. Summary of the Invention

[0006] To address the challenge of accurately detecting heat sources by eliminating reflection artifacts, this invention proposes an intelligent method for locating abnormal temperatures in solid waste based on infrared images. This method includes the following steps:

[0007] a. Initially screen candidate anomalous hotspots on the original infrared image using temperature thresholds;

[0008] b. For each of the candidate abnormal hotspots, acquire its data in the synchronously acquired visible light and infrared images, and extract its historical temperature sequence;

[0009] c. For each pixel within the candidate anomalous hotspot, calculate a pixel-level artifact confidence based on its static optical-thermodynamic coupling characteristics and dynamic temporal thermal inertia characteristics.

[0010] d. Calculate the overall true risk score of the candidate abnormal hotspot based on the artifact confidence of all pixels within the candidate abnormal hotspot;

[0011] e. Compare the actual risk score with a preset alarm threshold to determine whether the candidate abnormal hotspot is a real abnormal heat source, and locate the hotspot that is determined to be a real abnormal heat source.

[0012] This invention uses temperature as a preliminary screening criterion to identify candidate anomalous hotspots, and subsequent analysis is based solely on these candidate hotspots, effectively reducing computational load. Furthermore, it introduces artifact confidence to accurately distinguish between real heat sources and false reflective hotspots, significantly reducing the false alarm rate. Further, it describes the differences between reflection artifacts and real hotspots in static optics and thermodynamics to initially reflect the situation where a pixel is a reflection artifact. Moreover, considering that some highly reflective areas may also be real hotspots, static optics and thermodynamics alone cannot identify the real hotspots corresponding to highly reflective areas; therefore, it introduces dynamic time-domain thermal inertia features to identify the real hotspots corresponding to reflective areas.

[0013] Preferably, the preliminary screening of candidate abnormal hotspots by temperature threshold includes:

[0014] The original infrared image is binarized and connected component analysis is performed to identify each independent connected region as a candidate anomalous hotspot.

[0015] This invention can filter out candidate abnormal hotspots through simple binarization processing. This method is relatively simple and has higher efficiency.

[0016] Preferably, calculating the artifact confidence at the pixel level includes:

[0017] Calculate the confidence score of a static artifact that represents a static feature;

[0018] Calculate a dynamic time-domain thermal inertia index that characterizes dynamic features;

[0019] The pixel-level artifact confidence is calculated by combining the static artifact confidence with the dynamic time-domain thermal inertia index.

[0020] This invention analyzes artifact confidence using two components: static artifact confidence and dynamic time-domain thermal inertia index. The static artifact confidence assesses the degree to which the spatial and physical properties of a hotspot at the current moment match an artifact; the dynamic time-domain thermal inertia index assesses how its temperature changes over a past period match an artifact. Combining these two components forms a complete chain of evidence in both space and time, significantly enhancing the reliability of the final judgment.

[0021] Preferably, the static artifact confidence is obtained by coupling a specular reflection potential index calculated based on a visible light image and a local thermodynamic inconsistency calculated based on an infrared image.

[0022] This invention combines two typical characteristics of artifacts—the reflection properties at the viewing angle and the abrupt changes in temperature—to jointly determine possible artifact situations, thereby achieving a more comprehensive artifact determination.

[0023] Preferably, the specular reflection potential index is calculated based on the luminance and saturation components of the visible light image after conversion to the HSV color space.

[0024] This invention takes into account the characteristics of high brightness and low color saturation in reflective areas, and uses this characteristic to accurately determine the situation as a reflective feature.

[0025] Preferably, the local thermodynamic inconsistency is obtained by calculating the difference between the temperature of a pixel and the average temperature of a neighborhood that does not include itself.

[0026] Preferably, the dynamic time-domain thermal inertia index is calculated based on the statistical characteristics of historical temperature sequences.

[0027] Preferably, the dynamic time-domain thermal inertia index is obtained by calculating the ratio of the standard deviation to the mean absolute value of the historical temperature sequence and mapping it through an exponential function.

[0028] Preferably, the overall true risk score of the candidate abnormal hotspot is the weighted average temperature of all pixels within the candidate abnormal hotspot, with the degree of artifact suppression as the weight, wherein the degree of artifact suppression is derived from the artifact confidence score.

[0029] The present invention has the following beneficial effects:

[0030] This invention uses temperature to initially screen out candidate anomalous hotspots, and subsequent analysis is based only on these candidate anomalous hotspots, effectively reducing the amount of computation.

[0031] Furthermore, by introducing artifact confidence, we can accurately distinguish between real heat sources and false reflective hotspots, significantly reducing the false alarm rate;

[0032] Furthermore, by describing the differences between reflection artifacts and real hotspots in static optics and thermodynamics, we can initially reflect the situation where pixels are reflection artifacts.

[0033] Furthermore, considering that some highly reflective areas may also be real hot spots, the real hot spots corresponding to highly reflective areas cannot be identified by static optics and thermodynamics alone. Therefore, dynamic time-domain thermal inertia characteristics are introduced to identify the real hot spots corresponding to reflective areas. Attached Figure Description

[0034] Figure 1 This is a flowchart of the steps of the intelligent location method for abnormal temperature of solid waste based on infrared images provided in the embodiments of the present invention;

[0035] Figure 2 This is a visible light image provided in an embodiment of the present invention;

[0036] Figure 3 This is the original infrared image provided in the embodiments of the present invention;

[0037] Figure 4 These are candidate abnormal hotspot region images provided in embodiments of the present invention;

[0038] Figure 5 These are real abnormal heat source images provided in the embodiments of the present invention. Detailed Implementation

[0039] Please see Figure 1 The diagram illustrates a flowchart of the steps in an intelligent location method for abnormal temperatures in solid waste based on infrared images, as provided in an embodiment. The method includes the following steps:

[0040] S1: Simultaneously acquire the original infrared image and the spatially registered visible light image.

[0041] Specifically, a coaxial or paraxial imaging system consisting of an infrared thermal imager and a visible light RGB camera is deployed to simultaneously acquire the raw infrared image and the spatially registered visible light image at the current moment. Figure 2 and Figure 3 The images are shown in both visible light and raw infrared light. The visible light image reveals some highly reflective areas.

[0042] S2: Initially screen out candidate abnormal hotspots on the original infrared image using temperature thresholds.

[0043] Preferably, as an example, candidate anomalous hotspots are initially screened in the original infrared image using a temperature threshold, including:

[0044] First, the original infrared image at the current moment is binarized based on a preset temperature threshold, where pixel values ​​above the preset temperature threshold are set to 1, and pixel values ​​below the temperature threshold are set to 0.

[0045] Subsequently, connected component analysis was performed on the pixels with a value of 1 in the binarized image, and each independent connected region was identified as a candidate anomalous hotspot.

[0046] Figure 4 The image shows candidate hotspot regions, where white pixels represent these regions. The image reveals that the segmented candidate hotspot regions contain some reflection artifacts, which need to be eliminated.

[0047] S3: For each of the candidate abnormal hotspots, acquire its data in the synchronously acquired visible light and infrared images, and extract its historical temperature sequence.

[0048] It should be noted that, in order to obtain the corresponding pixels of candidate anomalous hotspots in the visible light image, pixel-level spatial alignment of the original infrared image and the visible light image is required first. Then, the pixels corresponding to the candidate anomalous hotspots are obtained in the visible light image.

[0049] Preferably, as an example, for each of the candidate anomalous hotspots, its data in the synchronously acquired visible light and infrared images are obtained, and its historical temperature sequence is extracted, including:

[0050] Step 1: Alignment.

[0051] Based on Zhang Zhengyou's calibration method, the intrinsic and extrinsic parameters and distortion coefficients of the infrared thermal imager and the visible light camera, both based on a pinhole camera model, are obtained. The rotation and translation matrices between the two are then calculated to establish a precise mapping relationship between pixels. Based on this mapping relationship, the original infrared image and the visible light image at the current moment are aligned.

[0052] Step 2: For each candidate abnormal hotspot, acquire its data in the synchronously acquired visible light and infrared images.

[0053] Based on the alignment results, regions corresponding to candidate anomalous hotspots are obtained in visible light and infrared images, respectively.

[0054] Step 3: Obtain historical temperature data.

[0055] The system acquires infrared images from a predetermined number of previous time points. Using a bounding box intersection-union (IOU) matching algorithm or a more advanced target tracking algorithm, it performs cross-frame tracking of candidate anomalous hotspots within these infrared images. Based on the tracking results, it extracts the historical temperature value corresponding to each pixel within the candidate anomalous hotspots from the infrared images from the predetermined number of previous time points, forming a historical temperature sequence. For example, the predetermined number can be 20.

[0056] S4: For each pixel within the candidate abnormal hotspot, calculate a pixel-level artifact confidence level based on its static optical-thermodynamic coupling characteristics and dynamic time-domain thermal inertia characteristics.

[0057] It should be noted that since there may be some reflection artifacts in the extracted candidate abnormal hotspots, the following distinctions need to be made based on the differences between the reflection artifacts and the real heat sources in the instantaneous image and their temporal variations.

[0058] S40: Obtain static optical-thermodynamic coupling characteristics.

[0059] It should be noted that real heat sources and reflection artifacts are somewhat distinguishable in instantaneous images. Based on the distinguishability of instantaneous images, some distinguishing features are extracted below.

[0060] S400: Calculates the pixel-level specular reflection potential index.

[0061] It should be noted that specular reflection areas typically exhibit high brightness and low saturation, and therefore, this characteristic can be used to extract discriminative features.

[0062] Preferably, as an example, calculating the pixel-level specular reflection potential index includes:

[0063] Convert the visible light image from the RGB color space to the HSV color space to obtain the brightness and saturation values ​​of each pixel.

[0064]

[0065] in, This represents the pixel-level specular reflection potential index of the j-th pixel in the i-th candidate anomalous hotspot. This represents the saturation value of the j-th pixel within the i-th candidate anomalous hotspot. This represents the brightness value of the j-th pixel within the i-th candidate anomalous hotspot. This represents the maximum brightness value of all pixels and is used for normalization.

[0066] It is understandable that specular reflection areas are characterized by high brightness and low saturation. Therefore, the greater the brightness and the lower the saturation of a pixel, the greater the possibility of specular reflection at that pixel location, and thus the greater the pixel-level specular reflection potential index at that pixel location.

[0067] S401: Calculate local thermodynamic inconsistencies.

[0068] It should be noted that a real, internally generated heat source diffuses its heat into the surrounding medium, forming a thermal halo with a temperature gradient on an infrared image; while a solar spot artifact is a surface reflection of external energy, so its temperature boundary with the surrounding area will exhibit a great discontinuity.

[0069] Preferably, as an example, calculating local thermodynamic inconsistencies includes:

[0070]

[0071] in, This represents the temperature value of the j-th pixel within the i-th candidate anomalous hotspot. This represents the average temperature value of all pixels within a preset neighborhood centered on the j-th pixel of the i-th candidate anomalous hotspot. This indicates the ambient temperature at the current moment. This indicates the local thermodynamic inconsistency of the j-th pixel within the i-th candidate anomalous hotspot. This indicates a preset zero-prevention parameter used to prevent the denominator from being 0. For example... Take 0.001.

[0072] It is understandable that when a high-temperature point is a solar spot artifact, its temperature will be much higher than that of its immediate vicinity, resulting in a large inconsistency in the calculated local thermodynamics.

[0073] S402: Calculate the confidence level of static artifacts based on pixel-level specular reflection potential index and local thermodynamic inconsistency.

[0074] Preferably, as an example, the confidence level of static artifacts is calculated based on pixel-level specular reflection potential index and local thermodynamic inconsistency, including:

[0075]

[0076] in, This represents the confidence level of the static artifact of the j-th pixel in the i-th candidate anomalous hotspot. This indicates the local thermodynamic inconsistency of the j-th pixel within the i-th candidate anomalous hotspot. This indicates linear normalization.

[0077] Understandably, this involves using two combinations to jointly determine the probability that a pixel in a candidate anomalous hotspot is a reflection artifact. A larger value indicates a higher probability that the candidate anomalous hotspot is a reflection artifact.

[0078] S41: Obtain dynamic time-domain thermal inertia characteristics.

[0079] It should be noted that the static artifact confidence level obtained in the above steps can effectively identify most instantaneous solar spot artifacts formed by specular reflection. However, if a highly reflective object is heated under prolonged sunlight, it will also appear highly reflective in visible light images, and in infrared images, it will exhibit a significant temperature difference with its surrounding environment due to its high temperature. Therefore, such a heated highly reflective object cannot be identified using the above features.

[0080] It should be further explained that, due to the specific heat capacity of objects, the temperature of an object heated by the sun changes gradually over time, exhibiting a low-frequency variation. This characteristic can be used to identify highly reflective objects that have been heated by the sun.

[0081] Preferably, as an example, obtaining the dynamic time-domain thermal inertia index includes:

[0082]

[0083] in, This represents the variance of all data in the historical temperature sequence of the j-th pixel within the i-th candidate anomalous hotspot. This represents the mean of all data in the historical temperature sequence of the j-th pixel within the i-th candidate anomalous hotspot. This represents an exponential function with the natural constant as its base. It represents the dynamic time-domain thermal inertia index of the j-th pixel in the i-th candidate anomalous hotspot.

[0084] Understandable It reflects the relative fluctuation of heat at a pixel location. The larger the value, the greater the instability of heat at that pixel location. Since the actual heat does not fluctuate significantly over time, the probability of the actual heat at that pixel location is smaller, and therefore the smaller the dynamic time-domain thermal inertia index.

[0085] S42: Calculate the confidence level of artifacts.

[0086] Preferably, as an example, calculating the artifact confidence includes:

[0087]

[0088] in, This represents the artifact confidence of the j-th pixel in the i-th candidate anomalous hotspot.

[0089] Understandable This value reflects the likelihood that a pixel is an artifact in a momentary image; the larger the value, the greater the likelihood that the pixel is an artifact. This reflects thermal inertia; the larger the value, the less likely the pixel is an artifact.

[0090] S5: Calculate the overall true risk score of the hotspot based on the artifact confidence of all pixels within the candidate abnormal hotspot.

[0091] It should be noted that the above process analyzes each pixel in the candidate abnormal hotspots to determine if each pixel is an artifact. The following process requires a comprehensive assessment of the situation of all pixels in the candidate abnormal hotspots to determine if each candidate abnormal hotspot is a real hotspot.

[0092] Preferably, as an example, based on the artifact confidence of all pixels within the candidate anomalous hotspot, the overall true risk score of the hotspot is calculated, including:

[0093]

[0094] in, This represents the number of pixels in the i-th candidate anomalous hotspot. This represents the true risk score of the i-th candidate abnormal hotspot.

[0095] Understandable This reflects the probability that a pixel represents real heat; the higher the value, the greater the probability that the pixel represents real heat. (Utilizing...) As a weight, the temperature of each pixel We perform a weighted summation to obtain the true temperature after eliminating artifact interference, which is the overall true risk score mentioned above.

[0096] S6: Compare the actual risk score with the preset alarm threshold to determine whether the candidate abnormal hotspot is a real abnormal heat source, and locate the hotspot that is determined to be a real abnormal heat source.

[0097] Preferably, as an example, the actual risk score is compared with a preset alarm threshold to determine whether the candidate abnormal hotspot is a real abnormal heat source, and the hotspot determined to be a real abnormal heat source is located, including:

[0098] The actual risk score is compared with a preset alarm threshold. Candidate abnormal hotspots with actual risk scores greater than the preset alarm threshold are identified as actual abnormal heat sources. The location of the actual abnormal heat source is then marked.

[0099] Figure 5The image shows a real anomalous heat source; the area marked by the black rectangle in the image is the real anomalous heat source. The image shows that reflection artifacts have been eliminated, revealing the true anomalous heat source.

[0100] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent location of abnormal temperatures in solid waste based on infrared images, characterized in that, include: a. Initially screen candidate anomalous hotspots on the original infrared image using temperature thresholds; b. For each of the candidate abnormal hotspots, acquire its data in the synchronously acquired visible light and infrared images, and extract its historical temperature sequence; c. For each pixel within the candidate anomalous hotspot, based on its static optical-thermodynamic coupling characteristics and dynamic temporal thermal inertia characteristics, calculate a pixel-level artifact confidence score, including: calculating a static artifact confidence score characterizing the static characteristics; calculating a dynamic temporal thermal inertia index characterizing the dynamic characteristics; and calculating the pixel-level artifact confidence score by combining the static artifact confidence score and the dynamic temporal thermal inertia index. The confidence level of static artifacts is obtained by coupling a specular reflection potential index calculated based on a visible light image and a local thermodynamic inconsistency calculated based on an infrared image. The dynamic time-domain thermal inertia index is calculated based on the statistical characteristics of historical temperature sequences; The dynamic time-domain thermal inertia index is obtained by calculating the ratio of the standard deviation to the mean absolute value of the historical temperature series and mapping it through an exponential function. d. Calculate the overall true risk score of the candidate abnormal hotspot based on the artifact confidence of all pixels within the candidate abnormal hotspot; e. Compare the actual risk score with a preset alarm threshold to determine whether the candidate abnormal hotspot is a real abnormal heat source, and locate the hotspot that is determined to be a real abnormal heat source.

2. The intelligent location method for abnormal temperature of solid waste based on infrared images according to claim 1, characterized in that, The preliminary screening of candidate abnormal hotspots using temperature thresholds includes: The original infrared image is binarized and connected component analysis is performed to identify each independent connected region as a candidate anomalous hotspot.

3. The intelligent location method for abnormal temperature of solid waste based on infrared images according to claim 1, characterized in that, The specular reflection potential index is calculated based on the luminance and saturation components of a visible light image after conversion to the HSV color space.

4. The intelligent location method for abnormal temperature of solid waste based on infrared images according to claim 1, characterized in that, The local thermodynamic inconsistency is obtained by calculating the difference between the temperature of a pixel and the average temperature of a neighborhood that does not include itself.

5. The intelligent location method for abnormal temperature of solid waste based on infrared images according to claim 1, characterized in that, The overall true risk score of the candidate anomalous hotspot is the weighted average temperature of all pixels within the candidate anomalous hotspot, with the degree of artifact suppression as the weight, wherein the degree of artifact suppression is derived from the artifact confidence score.

6. The intelligent location method for abnormal temperature of solid waste based on infrared images according to claim 1, characterized in that, Mark the location of the actual abnormal heat source.

Images