Image correction method based on infrared area scanning device

Abstract

The present application provides an image correction method based on infrared area array scanning equipment, which corrects the four surrounding dark corner problems caused by non-uniform of the infrared image collected by the equipment, and the method is also applicable to the image spot problem caused by dust on the detector. The method uses the imaging characteristics of the infrared area array search system: the dark corner distribution and the spot distribution between the continuous frames have consistency in space and continuity in time domain. After the normalized processing of the image integration of each circle, an adaptive matrix is obtained, which quickly solves the problems of infrared image dark corner and spot. It lays a good foundation for the following infrared panoramic stitching, target detection and identification.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology. Its main content is to correct the image quality degradation caused by non-uniformity and detector dust in infrared array scanning systems, as well as the presence of vignetting and spotting in the images. Background Technology

[0002] Because infrared wavelengths cover a wider range than visible light and possess a thermal effect (any heat-generating object emits infrared radiation), and because they can be observed at night, infrared imaging systems play an irreplaceable role in target detection and are widely used in security, defense, and remote sensing. With the rapid development of computer vision in recent years, the functions of infrared array search systems have been continuously improved and greatly enhanced, making them a crucial component of modern defense systems.

[0003] An infrared array detector maintains a constant elevation while rotating at a uniform azimuth speed of 360 degrees per second, continuously acquiring 100 frames of images, each frame measuring 640×513 pixels. Due to external factors, infrared detectors exhibit varying output responsivity across pixels even with constant infrared radiation, resulting in widespread non-uniformity. Furthermore, the underlying mechanisms of non-uniformity vary among different types and materials of infrared detectors, making it difficult to establish a unified and comprehensive mathematical model for non-uniformity.

[0004] To address the non-uniformity problem, most current methods fall into two categories: calibration-based non-uniformity correction methods and scene-based non-uniformity correction methods.

[0005] Calibration-based correction methods are simple and easy to implement, with two-point calibration being the most common. This involves using two blackbodies at different temperatures and a linear fitting algorithm to calculate the offset and gain of each detector. However, infrared thermal imagers are typically calibrated before leaving the factory. After delivery, due to unavoidable changes in the operating environment, such as drift caused by temperature and voltage variations, the non-uniformity in the image gradually increases, sometimes even severely affecting its usability. Therefore, periodic calibration is necessary.

[0006] Scene-based calibration methods (SBNUC) are diverse and do not require periodic recalibration of calculation parameters. These include methods based on minimum mean square error, methods based on jitter scanning mechanisms to determine motion position, methods based on mean and variance, and Bayesian methods. These methods can adaptively estimate the offset and gain of the infrared focal plane detector, but they suffer from artifacts and ghosting issues.

[0007] In summary, the images corrected by the two common non-uniformity correction algorithms still have some problems, affecting the imaging effect and subsequent processing. To address these issues, after hardware non-uniformity correction, the images acquired by the infrared array scanning device are preprocessed before image processing of the acquired image sequence to better resolve the problems caused by non-uniformity. Summary of the Invention

[0008] To address the shortcomings of existing infrared imaging technologies, such as image spots and uneven brightness, and to prevent them from affecting subsequent image stitching algorithms and target detection, tracking, and recognition algorithms, this invention proposes a new image correction method for infrared array scanning systems, taking advantage of their characteristics. This method corrects the image quality degradation caused by image non-uniformity and detector dust, as well as the presence of dark corners and spots in the image.

[0009] The technical solution adopted by this invention to solve the technical problem is as follows: Through comparative analysis, it was found that the non-uniformity of images between consecutive frames has the same distribution, exhibiting consistency in space and continuity in the temporal domain. Through cumulative normalization processing, an adaptive correction matrix is ​​obtained, which can solve the problems of vignetting and glare, eliminating the deficiencies in the device's imaging.

[0010] The method includes the following steps:

[0011] The detector of the infrared array search system rotates at a constant speed in the horizontal direction, continuously acquiring 100 frames of infrared images. The acquired infrared images are grayscale images, and each frame has M*N pixels, where the grayscale value of each pixel can be represented by g. 11 ,g 12 ,...,g MN If we represent the image matrix G, then... i Represented as:

[0012] For image sequence G i (where i = 1, 2, 3, ..., 100) are accumulated to obtain the accumulation matrix T.

[0013] Find the maximum gray value P in the cumulative matrix T. max The accumulated matrix is ​​normalized, and the correction matrix η obtained after normalization is:

[0014]

[0015] The infrared array detector developed in the laboratory rotates once every 1.5 seconds. In order to prevent poor imaging effect and new problems such as light spots and dark corners caused by external interference (such as temperature and voltage changes) during long-term operation, the correction matrix η is updated once every hour to prevent resource waste caused by real-time updates.

[0016] 3. After obtaining the correction matrix η, the acquired image sequence is corrected to obtain the corrected i-th frame image as G'. i :

[0017] G i '=G i *η (4)

[0018] The corrected image eliminates issues such as light spots and vignetting, making it easier to perform subsequent image processing operations, such as image stitching and target recognition.

[0019] The beneficial effects of this invention are that the algorithm has low computational complexity, saves computation time, consumes fewer resources, and can well meet real-time requirements. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Figure 1 This is a flowchart illustrating the implementation process of the present invention.

[0022] Figure 2 The problem of lens flare caused by dust.

[0023] Figure 3 The effect of correction on the light spot problem. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Several parameters are involved, and these parameters need to be adjusted for specific processing environments to achieve good performance. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Scope of application: Infrared images acquired by an infrared array scanning system, or infrared images acquired continuously based on the same detector.

[0026] Simulation environment: Matlab 2016a

[0027] Images: The images used in the experiment were 16-bit grayscale images.

[0028] Test steps:

[0029] (1) Applicable area scan system that acquires 100 frames of images, see Figure 2 Its grayscale matrix is:

[0030]

[0031] (2) Utilizing spatial consistency and temporal continuity, the non-uniformity of gray values ​​between consecutive frames has the same distribution, for image sequence G i After accumulation, the accumulation matrix is ​​obtained;

[0032]

[0033] (3) The maximum value in the accumulation matrix:

[0034] p max =1280601

[0035] The accumulated matrix is ​​normalized to obtain the correction matrix;

[0036]

[0037] (4) Each frame of the image sequence is multiplied by a correction matrix to obtain a new image, which effectively solves the problems of vignetting and glare. See [link to image analysis]. Figure 3 ;

[0038]

[0039] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. An image correction method based on an infrared array scanning device, the method comprising the following steps: 1) The detector of the infrared array search system rotates at a constant speed in the horizontal direction, continuously acquiring 100 frames of infrared images. The acquired infrared images are grayscale images, and each frame has M*N pixels, where... The grayscale value of each pixel can be expressed in g. 11 ,g 12 ,...,g MN The image matrix G is represented as follows. i Represented as: For image sequence G i (where i = 1, 2, 3, ..., 100) are accumulated to obtain the accumulation matrix T: 2) Find the maximum gray value P in the cumulative matrix T. max The accumulated matrix is ​​normalized, and the correction matrix η obtained after normalization is: 3) After obtaining the correction matrix η, the acquired image is corrected to obtain the corrected image G'. i : G′ i =G i *h (4).

Images