Infrared image local target enhancement method for boiler radiation field and related device

Abstract

This application discloses a method and related apparatus for enhancing local targets in infrared images of boiler radiation fields, relating to the field of infrared image processing technology. The method includes: acquiring an infrared image sequence of a target radiation scene and performing temporal preprocessing on the infrared image sequence to obtain an input image; separating the background and local targets in the input image by spatial scale to model and obtain a background image representing the background radiation distribution; performing background compensation processing on the input image and the background image to obtain a background compensation image, and performing image enhancement processing on the background compensation image to improve the contrast of the local targets. This application reconstructs the grayscale relationship between the background and local targets in the image, suppresses continuously varying large-scale background components, and avoids the simultaneous amplification of background radiation and noise during direct enhancement. This effectively solves the problem that weak local targets in complex radiation scenes are obscured by strong backgrounds, making them difficult to clearly identify and accurately locate.

Description

Technical Field

[0001] This application relates to the field of infrared image processing technology, and in particular to a method and related apparatus for enhancing local targets in infrared images of boiler radiation fields. Background Technology

[0002] In complex radiation environments such as industrial boilers and high-temperature kilns, the internal combustion process is accompanied by intense flame radiation, smoke and dust flow, and fly ash particle interference. To monitor the operating status of the equipment, infrared imaging technology is typically used to acquire temperature distribution images within the environment, in order to identify local targets such as coking on heated surfaces or areas of abnormal local temperatures.

[0003] Currently, the visualization processing of infrared images typically involves preprocessing and enhancing the acquired single or multiple frames of raw images. Common implementation methods include using algorithms such as wavelet transform and median filtering to remove noise, followed by histogram equalization, Laplacian edge enhancement, or adaptive contrast stretching to directly improve the contrast and clarity of the entire image, making the details in the image more apparent, thereby assisting operators in observing the conditions inside the furnace.

[0004] However, when faced with a boiler radiation field characterized by large-scale, continuous, strong background radiation and weak, blurred local targets, directly enhancing the original image often fails to effectively distinguish between background interference and the true local targets. Because background radiation dominates the image's grayscale distribution and changes drastically, conventional enhancement methods tend to simultaneously amplify background noise and non-target interference structures, resulting in no substantial improvement in the visibility of local targets, and even introducing artifacts, making spatial localization and clear display of local targets difficult. Summary of the Invention

[0005] In view of the above problems, this application provides a method and related apparatus for enhancing local targets in infrared images of boiler radiation fields, to solve the problem that local temperature anomaly regions in infrared images are difficult to clearly identify and locate due to strong background interference in complex radiation environments. The specific solution is as follows:

[0006] The first aspect of this application provides a method for local target enhancement in infrared images oriented towards boiler radiation fields, the method comprising:

[0007] Acquire an infrared image sequence under the target radiation scene, and perform time-series preprocessing on the infrared image sequence to obtain an input image;

[0008] By performing spatial scale separation on the background and local targets in the input image, a background image is modeled to characterize the background radiation distribution.

[0009] The input image and the background image are subjected to background compensation processing to obtain a background compensation image, and the background compensation image is subjected to image enhancement processing to improve the contrast of the local target.

[0010] In one possible implementation, the step of performing temporal preprocessing on the infrared image sequence to obtain the input image includes:

[0011] Determine multiple infrared images of the infrared image sequence within a preset time window;

[0012] The input image is obtained by statistically calculating multiple grayscale values ​​corresponding to the same pixel position in the multiple infrared images.

[0013] In one possible implementation, the step of modeling a background image to characterize the background radiation distribution by performing spatial scale separation of the background and local targets in the input image includes:

[0014] Morphological erosion is performed on the input image using a structuring element, wherein the size of the structuring element is larger than the typical projection size of the local target in the infrared image sequence.

[0015] The background image is obtained by using the morphological erosion operation result as the labeled image and the input image as the constraint image for morphological reconstruction.

[0016] In one possible implementation, the step of performing background compensation processing on the input image and the background image to obtain a background compensated image includes:

[0017] An initial compensation image is obtained by performing a pixel-by-pixel difference operation between the input image and the background image;

[0018] The background compensation image is obtained by constraining the grayscale values ​​in the initial compensation image.

[0019] In one possible implementation, the image enhancement processing of the background-compensated image includes:

[0020] The background compensation image is divided into multiple local regions;

[0021] Histogram equalization is performed on each local region separately. During the histogram equalization process, histogram peaks exceeding the preset threshold are cropped and the cropped portions are redistributed to each gray level.

[0022] The difference algorithm is used to smoothly fuse the histogram equalization results of each local region.

[0023] In one possible implementation, the infrared image local target enhancement method for boiler radiation fields further includes:

[0024] The input image is preprocessed to remove noise.

[0025] A second aspect of this application provides an infrared image local target enhancement device for a boiler radiation field, the infrared image local target enhancement device for a boiler radiation field comprising:

[0026] An image preprocessing module is used to acquire an infrared image sequence under the target radiation scene and perform time-series preprocessing on the infrared image sequence to obtain an input image;

[0027] The background separation module is used to model a background image that characterizes the background radiation distribution by performing spatial scale separation between the background and local targets in the input image.

[0028] The background compensation and image enhancement module is used to perform background compensation processing on the input image and the background image to obtain a background compensation image, and to perform image enhancement processing on the background compensation image to improve the contrast of the local target.

[0029] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the method for enhancing local targets in an infrared image oriented towards a boiler radiation field, as described in the first aspect or any implementation thereof.

[0030] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0031] The memory is used to store computer programs;

[0032] The processor is used to execute the computer program so that the electronic device can implement the method for local target enhancement of infrared images oriented towards boiler radiation fields, as described in the first aspect or any implementation thereof.

[0033] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs that, when executed by an electronic device, enable the electronic device to implement the method for enhancing local targets in an infrared image oriented towards a boiler radiation field, as described in the first aspect or any implementation thereof.

[0034] Using the above technical solution, this application provides a method and related apparatus for enhancing local targets in infrared images of boiler radiation fields. The method includes: acquiring an infrared image sequence under a target radiation scene and performing temporal preprocessing on the infrared image sequence to obtain an input image; separating the background and local targets in the input image by spatial scale to model and obtain a background image representing the background radiation distribution; performing background compensation processing on the input image and the background image to obtain a background compensation image; and performing image enhancement processing on the background compensation image to improve the contrast of the local targets. This application is based on the imaging mechanism of continuously and slowly varying background radiation and discrete distribution of hot surface coking or local temperature anomaly regions in the complex combustion environment of a boiler. It utilizes the spatial scale difference between the background and local targets to construct a background image, and reconstructs the grayscale relationship between the local targets and the background through background compensation, so that subsequent image enhancement processing mainly acts on the grayscale range corresponding to the local targets. In other words, this application reconstructs the grayscale relationship between the background and local targets in the image, suppresses the continuously varying large-scale background components, and avoids the situation where background radiation and noise are amplified simultaneously during direct enhancement. This effectively solves the problem that local targets are difficult to clearly identify and accurately locate when they are covered by strong backgrounds in complex radiation scenes, ensuring that image enhancement operations are mainly used for local targets that deviate from the background. This significantly improves the stability and engineering applicability of image processing and provides a highly reliable visualization basis for target detection under boiler radiation fields. Attached Figure Description

[0035] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0036] Figure 1 A flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, provided as an embodiment of this application;

[0037] Figure 2 A partial flowchart illustrating a method for enhancing local targets in infrared images of boiler radiation fields, provided in an embodiment of this application;

[0038] Figure 3 This is another part of the flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, as provided in an embodiment of this application.

[0039] Figure 4 This is another part of the flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, as provided in an embodiment of this application.

[0040] Figure 5This is another part of the flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, as provided in an embodiment of this application.

[0041] Figure 6 This application provides an example of an infrared image enhancement technique for local targets in a boiler furnace combustion environment, as shown in the following illustration. Figure 6 (a) is the original infrared image. Figure 6 (b) is the background image obtained through spatial scale separation modeling. Figure 6 (c) is the background-compensated image obtained after background compensation processing. Figure 6 (d) is the output result after image enhancement processing;

[0042] Figure 7 A schematic diagram of the structure of an infrared image local target enhancement device for a boiler radiation field provided in an embodiment of this application;

[0043] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0044] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0045] As will be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0046] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0047] In complex radiation scenarios such as industrial boilers and high-temperature kilns, the furnace interior contains high-temperature flames, black smoke, and fly ash, resulting in high-intensity infrared radiation with a continuous and slowly changing spatial distribution. Meanwhile, areas of coking or temperature anomalies are typically small in scale and have low radiation contrast, making them easily obscured by these complex media. Under these conditions, background radiation often dominates in infrared images, and target structures appear as weak, blurry, or even indistinguishable local features in the original image. Directly enhancing the original infrared image often amplifies both background radiation and noise interference, failing to effectively highlight weak target structures and potentially introducing pseudo-structures, thus reducing the image's structural realism and engineering reliability. Therefore, effectively improving the visualization of local targets in infrared images and achieving stable target display and positioning has become a pressing issue that needs to be addressed.

[0048] To address the aforementioned problems, this application provides a method for enhancing local targets in infrared images oriented towards boiler radiation fields, effectively highlighting local targets even in environments with strong interference. The method for enhancing local targets in infrared images oriented towards boiler radiation fields, according to this application, will be described in detail below with reference to the accompanying drawings.

[0049] See Figure 1 , Figure 1 This is a flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, as provided in an embodiment of this application. Figure 1 As shown in the figure, the infrared image local target enhancement method for boiler radiation field provided in this application embodiment may include steps S101 to S103, which are described in detail below.

[0050] S101, acquire the infrared image sequence under the target radiation scene, and perform time-series preprocessing on the infrared image sequence to obtain the input image.

[0051] In this embodiment, the target radiation scene is an imaging environment with strong background interference and weak local targets, such as a boiler furnace combustion environment. This environment includes a continuously distributed high-temperature flame background and discretely distributed coking or temperature anomaly areas. Multiple frames of infrared images are acquired within a continuous time window using an infrared imaging device to form an infrared image sequence, for example, 60 to 200 frames of infrared images are acquired continuously. The obtained infrared image sequence undergoes temporal preprocessing, i.e., statistical calculations or filtering are performed in the time dimension to suppress transient infrared fluctuations caused by flame flickering, smoke flow, and short-term obstruction by fly ash, thereby generating an input image that can characterize the stable radiation distribution under the current operating conditions.

[0052] In one possible implementation, temporal preprocessing can be achieved through median synthesis, time averaging, or robust statistics. For example, for multiple grayscale values ​​corresponding to the same pixel location in an infrared image sequence, the median can be taken as the grayscale value of that pixel in the input image, thereby eliminating extreme value interference caused by the instantaneous movement of fly ash or the violent flickering of flames. This time-robust processing prioritizes the preservation of background and target structures that appear stably within the time window, reducing the impact of random noise and transient interference, and providing a high signal-to-noise ratio data source for subsequent spatial scale separation. See also Figure 2 , Figure 2 This is a partial flowchart illustrating a method for enhancing local targets in an infrared image of a boiler radiation field, as provided in an embodiment of this application. Figure 2 As shown in the embodiment of this application, an infrared image local target enhancement method for boiler radiation field is provided. In step S101, "performing time-series preprocessing of the infrared image sequence to obtain the input image" may include steps S201 to S202, which are described in detail below.

[0053] S201, determine multiple infrared images in the infrared image sequence within a preset time window.

[0054] In this embodiment, the preset time window is a time range or frame range used to select consecutive image frames for time-series statistical calculations. The setting of this time window is based on the duration characteristics of transient interferences such as flame flickering, smoke flow, and fly ash obstruction under boiler combustion conditions. Specifically, the preset time window can be set to include 50 to 200 consecutive infrared images, for example, selecting 60 consecutive infrared images as a processing unit.

[0055] Taking the boiler furnace combustion environment as an example, background and target structures (such as coking areas or abnormal temperature areas) exhibit temporal stability during boiler combustion, while transient disturbances (such as fly ash flashes and localized smoke obstruction) are characterized by short duration, random occurrence, and rapid location changes. Therefore, in infrared image sequences, background and target structures appear continuously in most image frames, while transient disturbances typically appear only in a few. To address this, statistical processing of multiple infrared images within a time window can enhance the weight of persistent stable structures and reduce the weight of short-lived random disturbances, thereby weakening the impact of transient disturbances and random noise.

[0056] S202, the input image is obtained by statistically calculating the multiple gray values ​​corresponding to the same pixel position of multiple infrared images.

[0057] In this embodiment of the application, for multiple infrared images within the same time window, multiple gray values ​​at the same pixel location can be aggregated, such as any one or more of median value synthesis, time averaging, and robust statistical processing, to generate a single frame input image.

[0058] Specifically, for each pixel location, multiple grayscale values ​​corresponding to it in multiple infrared images are extracted to form a grayscale value sequence. Then, the statistical feature value of this grayscale value sequence is calculated as the grayscale value of the input image at that pixel location. For example, when using median value synthesis, the median of the grayscale value sequence is taken as the grayscale value of the input image at that pixel location. Alternatively, when using time averaging, the arithmetic mean is taken as the grayscale value of the input image at that pixel location. This application utilizes statistical principles to suppress transient infrared fluctuations caused by flame flicker, smoke flow, and short-term occlusion by fly ash, while preserving background and target structures that stably appear within the time window.

[0059] This temporal preprocessing effectively filters out random noise and transient high-brightness interference, making the generated input image more representative of the true radiation distribution under the current boiler operating conditions. Its signal-to-noise ratio is significantly better than the original single-frame image, and it eliminates the pseudo-changes caused by the dynamic combustion process. This provides a high-quality and highly stable data source for subsequent background modeling based on spatial scale separation, and avoids distortion in background image construction caused by transient artifacts in the input image.

[0060] It should be noted that infrared images taken in a boiler furnace combustion environment differ significantly from conventional static thermal target imaging. Flame radiation within the furnace exhibits temporal fluctuations; smoke flow and fly ash particles can cause short-term occlusion or transient high-brightness disturbances in localized areas. Therefore, a single frame image often fails to stably reflect the true radiation distribution under current operating conditions. Based on this, the temporal preprocessing in this application is not merely for general random noise reduction, but prioritizes preserving structures that stably appear within the time window and suppresses transient interference caused by flame flickering, smoke flow, and short-term fly ash occlusion with a temporal duration shorter than a preset number of frames. This ensures that the obtained input image better characterizes the stable background radiation distribution under the current boiler operating conditions.

[0061] In one possible implementation, before performing spatial scale separation and background modeling on the input image, denoising preprocessing can be used to reduce high-frequency random noise and residual transient interference components in the image, improve the signal-to-noise ratio of the input image, and provide a more stable and cleaner image source for subsequent construction of large-scale background images. This prevents noise from being misjudged as target structures or causing background image distortion during morphological operations or background compensation. In this regard, the local target enhancement method for infrared images of boiler radiation fields provided in this application embodiment further includes the following steps:

[0062] The input image undergoes denoising preprocessing.

[0063] In this embodiment, the input image can be preprocessed for denoising using discrete wavelet transform. Specifically, discrete wavelet transform can decompose the image into sub-bands of different spatial scales and directions, thereby distinguishing low-frequency components representing the main structural information and high-frequency components representing noise details. During processing, different threshold factors are set for the horizontal, vertical, and diagonal detail sub-bands in each wavelet decomposition layer; by comparing the wavelet coefficients of each sub-band with the set thresholds, coefficients below the threshold are zeroed or shrunk to suppress high-frequency noise, while coefficients above the threshold are retained to maintain the integrity of the target edge and detail structure.

[0064] Although the input image obtained from the aforementioned temporal preprocessing has suppressed some temporal fluctuations, random noise may still exist in the spatial domain of a single frame. To address this, denoising preprocessing can be performed to output a denoised image with a significantly improved signal-to-noise ratio. This denoised image, as the direct input to the subsequent spatial scale separation step, effectively avoids noise points interfering with the matching of structural elements in morphological erosion operations or creating false background undulations during background reconstruction. This significantly improves the robustness and engineering reliability of the entire infrared image local target enhancement method under harsh imaging conditions.

[0065] Therefore, temporal preprocessing suppresses transient fluctuations caused by flame flicker and smoke flow in the temporal dimension, preserving the stable radiative structure; while denoising preprocessing further filters out high-frequency random noise and subtle interference in the spatial frequency dimension. The combination of these two processes ensures that the image input to the background modeling stage possesses both temporal stability and spatial purity. Based on this, subsequent morphological background modeling using large-scale structural elements can more accurately extract the slowly varying background radiation distribution within the furnace, avoiding background image distortion caused by noise or transient interference. Furthermore, during background compensation and image enhancement, the contrast improvement between the target and background regions is more realistic and reliable, effectively reducing the risk of pseudo-anomalies and ultimately achieving clear visualization and precise localization of subtle temperature anomalies in the complex combustion environment of a boiler.

[0066] S102, by performing spatial scale separation on the background and local targets in the input image, a background image is obtained to characterize the background radiation distribution.

[0067] In this embodiment, taking advantage of the imaging characteristics of a large and continuously slow change in the spatial scale of background radiation and a small spatial scale of local targets, a mathematical morphology method is used to distinguish the background and local targets in the input image in the frequency domain or morphological domain, thereby achieving spatial scale separation. Then, a background image is obtained through morphological reconstruction modeling. This background image is image data that only contains large-scale continuous background radiation structure and does not contain medium- and small-scale local target structures, and is used to characterize the distribution trend of background radiation in the furnace.

[0068] In one possible implementation, a structuring element larger than the typical projected size of the local target in the infrared image sequence is first selected to perform morphological erosion on the input image to eliminate small- to medium-scale bright target structures. Then, the erosion result is used as the labeled image, and the input image is used as the constraint image for morphological reconstruction, restoring the large-scale background structure while avoiding background morphological distortion. For example, if the typical projected size of the coking region is 5×5 pixels, a 15×15 pixel structuring element can be used for erosion to ensure that the coking region is completely suppressed, while large-scale background areas such as the furnace wall are preserved. After the above spatial scale separation modeling, the resulting background image will retain only a large-scale, gradually varying grayscale distribution, while filtering out local coking bright spots. See also... Figure 3 , Figure 3 This is another schematic flowchart illustrating a method for enhancing local targets in infrared images of boiler radiation fields, provided as an embodiment of this application. Figure 3 As shown in the embodiment of this application, an infrared image local target enhancement method for boiler radiation field is provided. In this method, step S102, "by performing spatial scale separation of the background and local target in the input image, a background image for characterizing the background radiation distribution is obtained", may include steps S301 to S302. These steps are described in detail below.

[0069] S301 uses a structuring element to perform morphological erosion on the input image. The size of the structuring element is larger than the typical projection size of the local target in the infrared image sequence.

[0070] In this embodiment, morphological erosion operation utilizes mathematical morphology principles to scan each pixel of the input image using a sliding window, assigning the minimum grayscale value within the area covered by the structuring element to the central pixel. The structuring element is a matrix or shape template defined during morphological operations to detect the geometric shape of the image. Its specific shape can be rectangular, circular, elliptical, or cross-shaped, and its size directly determines the spatial scale of image features that can be preserved or suppressed. A typical projection size is the pixel range occupied by a local target (such as a coking area or temperature anomaly point within a boiler furnace) on the infrared imaging plane, which can be determined based on the boiler heating surface structure dimensions, infrared camera installation position, lens field of view, observation distance, and statistical results of anomaly areas under historical operating conditions.

[0071] Structural elements can act as scale filters. When the size of the structuring element is significantly larger than the typical projected size of the local target, during the erosion process, the local target region cannot fully accommodate the structuring element, and its bright grayscale values ​​are replaced by the surrounding lower background grayscale values, thus effectively suppressing or eliminating it in the calculation result. Meanwhile, the large-scale background radiation distribution, due to its slow and continuous spatial changes, can support the sliding of large-sized structuring elements, thus preserving their basic outline. For example, if the diameter of a typical coking spot in the furnace is approximately 15 pixels according to on-site calibration, a square structuring element with a size of 21×21 pixels or a circular structuring element with a radius of 11 pixels can be selected for the erosion operation. This size constraint ensures that all transient bright interferences smaller than this scale (such as flame flash points and fly ash particles) and local abnormal targets are removed during the erosion stage, leaving only the skeleton information of the large-scale background. Under the action of large-sized structuring elements, white local bright target regions (corresponding to coking or abnormally high-temperature areas) in the input image are significantly suppressed or even disappear, while large areas of slowly varying background regions, although their grayscale values ​​are reduced, still maintain their overall continuous shape.

[0072] By utilizing spatial scale differences to initially separate the background from the target, and through the erosion operation of large-sized structural elements, all bright target structures of small and medium scales in the image are forcibly removed, providing a pre-screened labeled image basis for the subsequent construction of a clean background image, thereby avoiding the mixing of target structures in the subsequent reconstruction process.

[0073] S302, using the morphological erosion operation result as the labeled image and the input image as the constraint image, performs morphological reconstruction to obtain the background image.

[0074] In this embodiment, morphological reconstruction is an iterative morphological filtering process that starts from the labeled image and recovers the geometric structure of the image under the constraints of the constrained image. The image after erosion is used as the labeled image, which contains background skeleton information after suppressing local targets, but details are lost and gray values ​​are generally low; the input image is used as the constrained image, which provides the upper limit boundary of pixel gray values ​​during the reconstruction process.

[0075] Morphological reconstruction typically involves repeatedly dilating the labeled image and then comparing the result of each dilation with the constraint image pixel by pixel until the image state no longer changes (i.e., convergence is achieved). During morphological reconstruction, the background skeleton in the labeled image gradually expands outward, attempting to restore its original shape and grayscale level. However, due to the constraints of the constraint image, the grayscale value in the reconstruction process will never exceed the grayscale value of the corresponding position in the input image. For eroded local target regions, since the value at the corresponding position in the labeled image has dropped to the background level, and the surrounding area is also background value during the dilation process, no matter how much dilation is performed, the result is limited by the actual low background value at that position in the original image (or truncated due to the presence of the target), thus failing to recover the original bright target structure. However, for large-scale background regions, the dilation operation can gradually raise their grayscale value to a level close to that of the original input image, thereby accurately restoring the continuous distribution characteristics of the background.

[0076] For example, during iterative reconstruction, if a pixel has a grayscale value of 50 in the labeled image and 120 in the constrained image, the dilation operation will attempt to increase the grayscale of the neighborhood of that pixel. However, the final output value is limited to a range not exceeding 120. Furthermore, since the target region has been filled in the labeled image, the reconstruction algorithm cannot overcome this trough to recover the isolated peak. After the above reconstruction process, the background image shows that the previously existing localized high-brightness anomalies have completely disappeared, replaced by a smooth and continuous background radiation distribution. The edge structure and gradient trend of the background are highly consistent with the original scene, without the boundary blurring or artifacts commonly seen in traditional opening operations. By introducing constraints, while restoring the large-scale background morphology, small- and medium-scale local targets are permanently removed, resulting in an ideal background image that possesses both high-fidelity background structure and is free from any target interference. This solves the problem of traditional filtering methods struggling to balance background integrity and target suppression.

[0077] Therefore, by using structuring elements larger than the typical projection size of local targets for morphological erosion, and then using the erosion results as markers and the original image as constraints for morphological reconstruction, accurate spatial scale separation between the background and local targets is achieved. By pre-setting the structuring element size and matching it with the projection size of the local targets, the erosion stage completely suppresses small-scale targets. Furthermore, by leveraging the constraint recovery mechanism of morphological reconstruction, the reconstruction path of local targets is blocked while restoring the large-scale background radiation distribution. This collaborative processing ensures that the final background image accurately reflects the continuously varying background features under complex radiation scenes, while completely removing discrete local high-temperature anomalies. This provides high-purity benchmark data for subsequent background compensation processing, effectively avoiding compensation residue or overcompensation problems caused by inaccurate background estimation, and significantly improving the accuracy and stability of local target enhancement.

[0078] S103, perform background compensation processing on the input image and the background image to obtain a background compensation image, and perform image enhancement processing on the background compensation image to improve the contrast of local targets.

[0079] In this embodiment, during background compensation processing, the background image is subtracted from the input image through pixel-by-pixel difference operations, thereby compressing the large-scale background radiation component. By compressing the background radiation component, the gray-level distribution relationship between the background and the local target can be reconstructed, thus preserving and enhancing the local target. For the background-compensated image obtained after background compensation processing, image enhancement processing is performed through contrast adjustment to further highlight the details of the local target. By performing cascaded processing of background compensation followed by enhancement, the problem of simultaneous amplification of background noise caused by direct enhancement is avoided. This allows the enhancement operation to act more on the gray-level range where the target structure is located, thereby significantly improving the visualization effect and positioning accuracy of local high-temperature targets against complex and strong backgrounds, effectively solving the problem of difficulty in distinguishing the target from the background.

[0080] In one possible implementation, a pixel-by-pixel difference operation can be performed between the input image and the background image, followed by grayscale range constraint processing to obtain the background compensation image. This allows the grayscale dynamic range to be mainly concentrated in the local target region. See also Figure 4 , Figure 4 This is another schematic flowchart illustrating a method for enhancing local targets in infrared images of boiler radiation fields, provided as an embodiment of this application. Figure 4 As shown in the embodiment of this application, an infrared image local target enhancement method for boiler radiation field is provided. In step S103, "performing background compensation processing on the input image and the background image to obtain a background compensation image" may include steps S401 to S402. These steps are described in detail below.

[0081] S401, Perform pixel-by-pixel difference operation on the input image and the background image to obtain the initial compensation image.

[0082] In this embodiment, during the pixel-by-pixel difference operation, the gray value of each pixel in the input image is subtracted from the gray value of the corresponding pixel in the background image, thereby generating an initial compensation image based on the gray value difference between each pixel. Taking the combustion environment of a boiler furnace as an example, if the gray value of a certain pixel in the input image is 180 (including background radiation 150 and coking target radiation increment 30), while the gray value of the same pixel in the background image is 150 (only background radiation), then after the difference operation, the gray value of that pixel is 30, thereby extracting the weak target signal that was originally submerged in the strong background.

[0083] By performing a subtraction operation, the dominant large-scale background radiation component in the input image is significantly compressed or even eliminated. Since the local target is not preserved in the background image (its gray value is close to the surrounding background), it exhibits a significant positive gray-level deviation after the difference operation. This can effectively remove low-frequency background interference by using spatial high-pass filtering, so that the initial compensation image mainly reflects the local abnormal signals that deviate from the background image.

[0084] S402, constrain the gray values ​​in the initial compensation image to obtain the background compensation image.

[0085] Since the background image is an estimated value obtained through morphological erosion and reconstruction, in some areas with complex textures or high noise, the estimated value of the background image may be slightly higher than the actual grayscale value of the input image, resulting in negative grayscale values ​​after the difference operation. Furthermore, extreme transient interference may also cause the difference result to exceed the normal dynamic range. Therefore, in this embodiment, upper and lower limit constraints are set on the grayscale values ​​in the initial compensation image. Specifically, all grayscale values ​​less than zero can be forcibly set to zero or a preset minimum non-negative threshold to eliminate the risk of nonlinear distortion caused by negative values; simultaneously, grayscale values ​​exceeding the maximum allowable value can be truncated to prevent data overflow.

[0086] After constraint processing, the negative value regions (usually manifested as black noise or artifacts) originally generated by the difference are effectively suppressed, the overall gray-level distribution of the image is limited to a reasonable dynamic range, the background area tends to be a uniform dark tone, while the target area retains its bright features. This not only ensures the integrity of the image gray-level histogram, but also avoids erroneous gradient calculations or pseudo-structure amplification in subsequent enhancement algorithms when processing negative or extreme values.

[0087] Therefore, it is evident that pixel-by-pixel differencing and grayscale constraint processing can effectively suppress large-scale backgrounds and stably extract local targets in complex radiation scenarios. Specifically, the differencing operation precisely removes low-frequency background radiation from the input image, significantly improving the relative salience of local targets; while the constraint processing eliminates negative noise and numerical instability that may be introduced during the differencing process, ensuring the purity of the background compensation image. After the above processing, the previously blurred outline of the coking area inside the furnace becomes clearly discernible, background interference is significantly reduced, and no obvious artifacts are introduced. This provides a high-quality input foundation for subsequent enhancement processing such as histogram equalization, thereby achieving accurate visualization and localization of temperature anomaly areas in complex combustion environments.

[0088] In one possible implementation, during the image enhancement process of the background compensation image, a constrained adaptive histogram equalization algorithm can be used. This algorithm divides the background compensation image into multiple local regions, performs histogram equalization on each region separately, and crops and redistributes histogram peaks exceeding a preset threshold. Finally, a difference algorithm is used to smoothly fuse the results from each region. See also Figure 5 , Figure 5 This is another schematic flowchart illustrating a method for enhancing local targets in infrared images of boiler radiation fields, provided as an embodiment of this application. Figure 5 As shown in the embodiment of this application, an infrared image local target enhancement method for boiler radiation field is provided. In step S103, "image enhancement processing of background compensation image" can include steps S501 to S503, which are described in detail below.

[0089] S501 divides the background compensation image into multiple local regions.

[0090] In this embodiment, the background compensation image is spatially segmented into several non-overlapping or partially overlapping sub-images to obtain multiple local regions. The size of the local regions can be set according to the typical projection size of the target structure in the background compensation image and the spatial frequency of noise particles. For example, when the resolution of the background compensation image is 640×480, it can be divided into 8×8 rectangular sub-regions, each containing 80×60 pixels. If the local regions are divided too large, they tend to be equalized globally, making it difficult to highlight small high-temperature coking areas; if the divisions are too small, high-frequency noise is easily amplified and computational complexity is increased.

[0091] By dividing the image into local regions, we can adopt differentiated enhancement strategies for different regions based on the non-uniformity of local gray-level distribution. This avoids local overexposure or underexposure caused by global enhancement, and thus more precisely captures the radiation distribution characteristics of different locations within the furnace. This provides a basis for subsequent independent contrast adjustment of each region and enhances the local statistical significance of weak high-temperature anomaly areas.

[0092] S502 performs histogram equalization on each local region separately, and during the histogram equalization process, the histogram peaks that exceed the preset limit threshold are cropped and the cropped parts are redistributed to each gray level.

[0093] In this embodiment, during histogram equalization, the grayscale distribution of pixels in each local region can be mapped to a uniform distribution using a transformation function, thereby expanding the grayscale dynamic range within that local region. Furthermore, during histogram equalization, when the number of pixels at a certain grayscale level (typically a residual background or noise accumulation area) within a local region is excessive, causing the histogram peak to exceed a preset threshold, the excess pixels are collected, and the total number of cropped pixels is evenly distributed to other grayscale levels within that local region. This maintains the overall brightness balance and information conservation of the image, significantly improving the texture clarity of the coking area while effectively suppressing interference from non-target pseudostructures such as fly ash and soot scattering common in complex boiler combustion environments, ensuring higher reliability of the enhanced image.

[0094] It should be noted that the preset threshold is used to control the magnitude of local contrast enhancement, preventing sharp peaks in the histogram caused by a small number of extremely high grayscale pixels in a local area (such as fly ash particle reflections or flame edge disturbances), which could lead to an excessively steep slope in the transform function, resulting in over-enhancement of non-target structures and noise amplification. This preset threshold can be determined based on the square root of the total number of pixels in the local area or an empirical coefficient.

[0095] S503 uses the difference algorithm to smoothly fuse the histogram equalization results of each local region.

[0096] Since histogram equalization and peak clipping are performed independently in each local region, the transformation functions between adjacent regions may differ. Direct stitching can lead to noticeable grid-like artifacts in the image. To address this, this embodiment employs bilinear or bicubic interpolation algorithms to process the histogram equalization results of each local region, eliminating the block artifacts caused by segmented processing and ensuring the spatial continuity and naturalness of the final output image.

[0097] In practical applications, the difference algorithm can be used to calculate the weighted average of the transformation results of four adjacent local regions surrounding the pixel to be processed, thus achieving a smooth transition. Specifically, for pixels located at the intersection of the four local regions, different weights are assigned according to their distance from the center point of each region, thereby generating smooth grayscale values. By using the difference algorithm to smoothly fuse the histogram equalization results of each local region, not only are the high-contrast details of each enhanced local region preserved, but the visual discontinuities caused by block boundaries are also eliminated. This results in a softer and more realistic outline of temperature anomaly areas in the final infrared image, making it easier for operators to accurately identify the focusing location.

[0098] Therefore, dividing the background compensation image into multiple local regions allows the enhancement operation to adapt to the spatial non-uniformity of radiation distribution within the furnace. Furthermore, by cropping histogram peaks exceeding a preset threshold and redistributing the cropped portions, the enhancement amplitude of local contrast is limited, preventing the synchronous amplification of transient high-brightness interference such as flame flickering and fly ash particles, thus avoiding the formation of pseudo-anomalies. Further, the histogram equalization results are smoothly fused using an interpolation algorithm, eliminating the block effect caused by segmentation and ensuring the visual continuity of the image. This allows for progressive optimization with the preceding background compensation process. Background compensation addresses the problem of targets being invisible due to large-scale background radiation dominance, while image enhancement solves the problem of targets being visible but with blurred details and insufficient contrast. Through background compensation and image enhancement, the grayscale statistical relationship between the background and the target is reconstructed, and the visibility of the target is maximized within a local range, thereby achieving stable display and accurate positioning of temperature anomaly areas in the complex combustion environment of a boiler. After the above processing, the previously blurry in-focus areas or high-temperature abnormal areas in the original image now exhibit clear texture structures and well-defined boundaries, while the surrounding background radiation and noise interference are effectively suppressed, significantly improving the engineering applicability of the image.

[0099] See Figure 6 , Figure 6 This is an example of an infrared image enhancement technique used in an embodiment of this application to enhance local targets within a boiler furnace combustion environment. Figure 6 As shown, Figure 6 (a) is the original infrared image. The local target is the coking area on the boiler heating surface. Under the conditions of high temperature fly ash, flue gas radiation and complex background interference, the gray level / radiation contrast between the local target and the surrounding background in the original infrared image is low and the local structural features are not obvious, making it difficult to directly distinguish the local target. Figure 6(b) is the background image obtained by spatial scale separation modeling. By using large-scale structuring elements to model the background of the original image, the local small-scale target response can be effectively suppressed, while the slow-changing large-scale background radiation distribution characteristics in the furnace are preserved, thus obtaining a background image that does not contain significant coking target details. Figure 6 (c) is the background compensation image obtained by background compensation processing. Background compensation processing can effectively reduce the influence of large-scale non-uniform background radiation in the furnace on target detection, improve the local contrast between the coking area and the background, and make the structural outline of the coking area clearer. Figure 6 (d) shows the output after image enhancement processing. After image enhancement processing, the morphological structure and local details of the coking area are further highlighted, thereby improving the visualization and recognition effect of the boiler coking area under complex high-temperature background. According to actual display requirements, different pseudo-color mapping methods can also be used to display the enhancement results to further improve the display and recognition of local targets.

[0100] Based on the above description, the infrared image local target enhancement method for boiler radiation fields provided by this application reconstructs the grayscale relationship between the background and local targets in the image, suppresses continuously slowly varying large-scale background components, and avoids the situation where background radiation and noise are simultaneously amplified during direct enhancement. This effectively solves the problem that weak local targets are difficult to clearly identify and accurately locate when they are obscured by strong backgrounds in complex radiation scenarios, ensuring that the enhancement operation is mainly used for local abnormal signals that deviate from the background. This significantly improves the stability and engineering applicability of image processing and provides highly reliable visualization evidence for target detection under complex working conditions.

[0101] The above describes a method for enhancing local targets in infrared images oriented towards boiler radiation fields, provided by embodiments of this application. The following describes the apparatus for performing the above-described method for enhancing local targets in infrared images oriented towards boiler radiation fields.

[0102] See Figure 7 , Figure 7 This is a schematic diagram of an infrared image local target enhancement device for a boiler radiation field, provided as an embodiment of this application. Figure 7 As shown in the figure, an infrared image local target enhancement device for boiler radiation field provided in this application embodiment includes:

[0103] The image preprocessing module 701 is used to acquire an infrared image sequence under the target radiation scene and perform time-series preprocessing on the infrared image sequence to obtain an input image;

[0104] Background separation module 702 is used to model a background image that characterizes the background radiation distribution by performing spatial scale separation between the background and local targets in the input image.

[0105] The background compensation and image enhancement module 703 is used to perform background compensation processing on the input image and the background image to obtain a background compensation image, and to perform image enhancement processing on the background compensation image to improve the contrast of local targets.

[0106] In one possible implementation, the image preprocessing module 701, used for temporal preprocessing of the infrared image sequence to obtain the input image, is specifically used for:

[0107] The process involves identifying multiple infrared images within a preset time window and then statistically calculating the grayscale values ​​corresponding to the same pixel position in each infrared image to obtain the input image.

[0108] In one possible implementation, the background separation module 702 is specifically used for:

[0109] Morphological erosion is performed on the input image using a structuring element, the size of which is larger than the typical projection size of the local target in the infrared image sequence. The morphological erosion result is used as the labeled image, and the input image is used as the constraint image for morphological reconstruction to obtain the background image.

[0110] In one possible implementation, the background compensation and image enhancement module 703, used to perform background compensation processing on the input image and the background image to obtain a background-compensated image, is specifically used for:

[0111] The initial compensation image is obtained by performing pixel-by-pixel difference operation between the input image and the background image; the background compensation image is obtained by constraining the gray values ​​in the initial compensation image.

[0112] In one possible implementation, the background compensation and image enhancement module 703, used for image enhancement processing of the background-compensated image, is specifically used for:

[0113] The background compensation image is divided into multiple local regions; histogram equalization is performed on each local region, and during the histogram equalization process, histogram peaks exceeding a preset threshold are cropped and the cropped portions are redistributed to various gray levels; the histogram equalization results of each local region are smoothly fused using a difference algorithm.

[0114] In one possible implementation, the image preprocessing module 701 is further configured to:

[0115] The input image undergoes denoising preprocessing.

[0116] It should be noted that the detailed functions of each module in the embodiments of this application can be found in the corresponding disclosure of the above-mentioned embodiment of the method for enhancing local targets in infrared images of boiler radiation fields, and will not be repeated here.

[0117] This application also provides an electronic device in its embodiments. See also... Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device in this embodiment may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 8 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0118] like Figure 8 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage device 808 into a random access memory (RAM) 803. When the electronic device is powered on, the RAM 803 also stores various programs and data required for the operation of the electronic device. The processing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0119] Typically, the following devices can be connected to I / O interface 805: input devices 806 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 807 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 808 including, for example, memory cards, hard drives, etc.; and communication devices 809. Communication device 809 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0120] This application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the infrared image local target enhancement methods for boiler radiation fields provided in this application.

[0121] This application also provides a computer-readable storage medium carrying one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the infrared image local target enhancement methods for boiler radiation fields provided in this application.

[0122] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0124] In the above embodiments, the implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, in the form of a computer program product.

[0125] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for enhancing local targets in infrared images of boiler radiation fields, characterized in that, The infrared image local target enhancement method for boiler radiation field includes: Acquire an infrared image sequence under the target radiation scene, and perform time-series preprocessing on the infrared image sequence to obtain an input image; By performing spatial scale separation on the background and local targets in the input image, a background image is modeled to characterize the background radiation distribution. The input image and the background image are subjected to background compensation processing to obtain a background compensation image, and the background compensation image is subjected to image enhancement processing to improve the contrast of the local target.

2. The method for enhancing local targets in infrared images oriented towards boiler radiation fields according to claim 1, characterized in that, The step of performing time-series preprocessing on the infrared image sequence to obtain the input image includes: Determine multiple infrared images of the infrared image sequence within a preset time window; The input image is obtained by statistically calculating multiple grayscale values ​​corresponding to the same pixel position in the multiple infrared images.

3. The method for enhancing local targets in infrared images oriented towards boiler radiation fields according to claim 1, characterized in that, The step of modeling a background image to characterize the background radiation distribution by performing spatial scale separation between the background and local targets in the input image includes: Morphological erosion is performed on the input image using a structuring element, wherein the size of the structuring element is larger than the typical projection size of the local target in the infrared image sequence. The background image is obtained by using the morphological erosion operation result as the labeled image and the input image as the constraint image for morphological reconstruction.

4. The method for enhancing local targets in infrared images oriented towards boiler radiation fields according to claim 1, characterized in that, The step of performing background compensation processing on the input image and the background image to obtain a background compensated image includes: An initial compensation image is obtained by performing a pixel-by-pixel difference operation between the input image and the background image; The background compensation image is obtained by constraining the grayscale values ​​in the initial compensation image.

5. The method for enhancing local targets in infrared images oriented towards boiler radiation fields according to claim 1, characterized in that, The image enhancement processing of the background-compensated image includes: The background compensation image is divided into multiple local regions; Histogram equalization is performed on each local region separately. During the histogram equalization process, histogram peaks exceeding the preset threshold are cropped and the cropped portions are redistributed to each gray level. The difference algorithm is used to smoothly fuse the histogram equalization results of each local region.

6. The method for enhancing local targets in infrared images oriented towards boiler radiation fields according to claim 1, characterized in that, The infrared image local target enhancement method for boiler radiation fields also includes: The input image is preprocessed to remove noise.

7. A local target enhancement device for infrared images oriented towards the radiation field of a boiler, characterized in that, The infrared image local target enhancement device facing the boiler radiation field includes: An image preprocessing module is used to acquire an infrared image sequence under the target radiation scene and perform time-series preprocessing on the infrared image sequence to obtain an input image; The background separation module is used to model a background image that characterizes the background radiation distribution by performing spatial scale separation between the background and local targets in the input image. The background compensation and image enhancement module is used to perform background compensation processing on the input image and the background image to obtain a background compensation image, and to perform image enhancement processing on the background compensation image to improve the contrast of the local target.

8. A computer program product, characterized in that, It includes computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the method for local target enhancement of infrared images oriented towards boiler radiation fields as described in any one of claims 1 to 5.

9. An electronic device, characterized in that, It includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program to enable the electronic device to implement the infrared image local target enhancement method for boiler radiation fields as described in any one of claims 1 to 5.

10. A computer storage medium, characterized in that, The storage medium carries one or more computer programs that, when executed by an electronic device, enable the electronic device to implement the method for local target enhancement of infrared images oriented towards boiler radiation fields as described in any one of claims 1 to 5.

Images