阅读说明：本技术 一种暗弱目标红外微分探测方法 (Infrared differential detection method for dark and weak targets ) 是由 饶鹏 陈略 贾良杰 于 2021-07-15 设计创作，主要内容包括：本发明公开了一种暗弱目标红外微分探测方法,包括建立不同的红外微分探测模型,如典型目标和背景的光谱特性模型、偏振特性模型和运动特性模型,通过在不同维度的微分探测区间内对目标特性模型的比较分析,在物理因素层面筛选探测维度；提出微分探测因子选择优化方法,对筛出的维度进行优化,选出最优的红外微分维度和区间作为微分探测因子,使得输出的微分探测因子流中目标分量与杂波、噪声分量差异最大。红外微分探测方法为解决对暗弱目标检测效果差和难以识别的难题提供基础。(The invention discloses a dark and weak target infrared differential detection method, which comprises the steps of establishing different infrared differential detection models, such as a spectral characteristic model, a polarization characteristic model and a motion characteristic model of a typical target and a background, and screening detection dimensions on the level of physical factors by comparing and analyzing the target characteristic models in differential detection intervals with different dimensions; and (3) providing a differential detection factor selection optimization method, optimizing the screened dimensions, and selecting the optimal infrared differential dimensions and intervals as differential detection factors so as to enable the difference between target components in the output differential detection factor flow and clutter and noise components to be maximum. The infrared differential detection method provides a basis for solving the difficult problems of poor detection effect and difficult identification of the dark and weak targets.)

1. A dark and weak target infrared differential detection method is characterized in that: the method comprises the following steps:

the method comprises the following steps: constructing an infrared signal statistical model:

for analyzing and screening infrared differential detection factors, researching remote sensing detection action factors aiming at a typical target, and constructing a spectral characteristic model L (lambda, L, T), a polarization characteristic model P (theta, L) and a motion characteristic model G (x, y, T) of the typical target and a background;

on the basis of constructing a target and background spectrum model, carrying out noise-clutter combined statistical characteristic analysis, constructing a motion characteristic model G (x, y, t), and counting the response of the target on a detection system and the rule of space position and time;

respectively simulating and calculating infrared differential probe measurement of a target and a background by using a radiation transmission mode; carrying out target detection on the analog detection by using a traditional detection method and an infrared differential detection method, and calculating a distinguishing coefficient of the traditional detection method and the infrared differential detection method; finally, comparing the distinguishing coefficients, and determining available infrared differential detection dimension information of the typical target and the scene;

screening detection dimensions on the physical factor level through comparative analysis of an infrared differential detection mechanism model and a target characteristic model;

step two: infrared differential detection model construction and differential detection factor selection:

selecting the optimal infrared differential dimension and interval as differential detection factors according to the detection dimension output in the step one, and optimizing the obtained factors to enable the difference between dark and weak target components in the output infrared signals and clutter and noise components to be maximum;

the mathematical model of the infrared differential detection model is summarized by the following formula:

2. the infrared differential detection method for the dark and weak targets as claimed in claim 1, characterized in that: the action factors of remote sensing detection in the first step comprise wave bands, polarization states and space-time characteristics.

3. The infrared differential detection method for the dark and weak targets as claimed in claim 1, characterized in that: in the first step, G is a gray value, P is a polarization degree, L is a spectrum intensity, lambda is a wavelength, L is a detection distance, T is a radiation source temperature, (x, y) are image coordinates, and T is an imaging time.

4. The infrared differential detection method for the dark and weak targets as claimed in claim 1, characterized in that: in the second step, F is a method for solving the difference, y₁，y₂，…，y_mIs a response in a different dimension x, x₁，x₂,…,x_nFor n different infrared micro-divisions in the x dimension,representing an optimal set of infrared differential factors.

Technical Field

The invention belongs to the technical field of infrared remote sensing and infrared space, and particularly relates to an infrared differential detection method for dark and weak targets.

Background

At present, the space-based infrared detection is mainly used for detecting dark and weak targets under a complex ground background, and the limitations of low image signal-to-noise ratio/signal-to-clutter ratio exist, so that the perception of the targets only stays at the detection and tracking level, false alarm, missing detection and error tracking are very easy to cause due to noise and clutter, the system stability is difficult to guarantee, and the capacity requirement of the space-based infrared detection cannot be met. The effective detection of the infrared weak target is a technical bottleneck to be broken through urgently, and the realization of the detection of the infrared weak target has great value and practical significance for the development of a space-based infrared detection system. Under the existing infrared imaging system, the multi-dimensional information fusion can improve the detection and perception capability to a certain extent, but the difference between a target and a background cannot be effectively distinguished due to the adoption of the traditional energy integration-based method. The method is characterized in that in the process of acquiring, amplifying and quantizing signals of all dimensions, more noise and quantization errors are introduced, the signal-to-noise ratio is affected and even seriously degenerated, the digital fusion information processing is difficult to achieve expectation due to the fact that the infrared target signals, the noise and the background clutter cannot be selectively amplified, and the infrared detection capability is seriously restricted. This problem is also known as non-selective perception defects of infrared targets. The idea for solving the defects is to find the difference of the infrared target and the background in different dimensionality information and distinguish the difference to realize selective amplification, so that the selective perception of the target is fundamentally realized.

Disclosure of Invention

In order to overcome the problems, the invention provides an infrared differential detection method for dark and weak targets.

The technical scheme adopted by the invention is as follows:

a dark and weak target infrared differential detection method comprises the following steps:

the method comprises the following steps: constructing an infrared signal statistical model:

for analyzing and screening infrared differential detection factors, researching remote sensing detection action factors aiming at a typical target, and constructing a spectral characteristic model L (lambda, L, T), a polarization characteristic model P (theta, L) and a motion characteristic model G (x, y, T) of the typical target and a background;

on the basis of constructing a target and background spectrum model, carrying out noise-clutter combined statistical characteristic analysis, constructing a motion characteristic model G (x, y, t), and counting the response of the target on a detection system and the rule of space position and time;

respectively simulating and calculating infrared differential probe measurement of a target and a background by using a radiation transmission mode; carrying out target detection on the analog detection by using a traditional detection method and an infrared differential detection method, and calculating a distinguishing coefficient of the traditional detection method and the infrared differential detection method; finally, comparing the distinguishing coefficients, and determining available infrared differential detection dimension information of the typical target and the scene;

screening detection dimensions on the physical factor level through comparative analysis of an infrared differential detection mechanism model and a target characteristic model;

step two: infrared differential detection model construction and differential detection factor selection:

selecting the optimal infrared differential dimension and interval as differential detection factors according to the detection dimension output in the step one, and optimizing the obtained factors to enable the difference between dark and weak target components in the output infrared signals and clutter and noise components to be maximum;

the mathematical model of the infrared differential detection model is summarized by the following formula:

and the action factors of remote sensing detection in the first step comprise wave bands, polarization states and space-time characteristics.

In the first step, G is a gray value, P is a polarization degree, L is a spectral intensity, lambda is a wavelength, L is a detection distance, T is a radiation source temperature, (x, y) are image coordinates, and T is an imaging time.

Wherein, F in the second step is a method for solving the difference, y₁,y₂,…,y_mIs a response in a different dimension x, x₁，x₂，…，x_nFor n different infrared micro-divisions in the x dimension,representing an optimal set of infrared differential factors.

The invention has the following advantages:

the infrared differential detection establishes a distinguishing factor through the change rate of multimode detection data, and the biggest advantage of the infrared differential detection is that the distinguishing capability is improved for a scene that a target is close to a background during single-mode detection through simple analysis, so that the numerical analysis of the infrared differential detection is established by taking the target distinguishing degree as a comparison reference when the target distinguishing degree is 0 during single-mode detection, the dimensionality and the differential division which can enable the difference between an infrared dark and weak target, the background and noise to be the biggest are determined, and the practical application guidance of the infrared dark and weak target detection is provided.

Drawings

FIG. 1 is a block diagram of an implementation flow of the present invention;

FIG. 2 is a flow chart of the present invention for analyzing and screening useful infrared differential factors.

Detailed Description

The present invention will be further described below, but the present invention is not limited to these.

A dark and weak target infrared differential detection method comprises the following steps:

the method comprises the following steps: constructing an infrared signal statistical model:

for analyzing and screening infrared differential detection factors, aiming at a typical target, researching remote sensing detection action factors including wave bands, polarization states, space-time characteristics and the like, and constructing a spectral characteristic model L (lambda, L, T), a polarization characteristic model P (theta, L) and a motion characteristic model G (x, y, T) of the typical target and a background, wherein G is a gray value, P is a polarization degree, L is a spectral intensity, lambda is a wavelength, L is a detection distance, T is a radiation source temperature, (x, y) is an image coordinate, and T is an imaging moment;

on the basis of constructing a target and background spectrum model, carrying out noise-clutter combined statistical characteristic analysis, constructing a motion characteristic model G (x, y, t), and counting the response of the target on a detection system and the rule of space position and time;

respectively simulating and calculating infrared differential probe measurement of a target and a background by using a radiation transmission mode; carrying out target detection on the analog detection by using a traditional detection method and an infrared differential detection method, and calculating a distinguishing coefficient of the traditional detection method and the infrared differential detection method; finally, comparing the distinguishing coefficients, and determining available infrared differential detection dimension information of the typical target and the scene;

screening detection dimensions on the physical factor level through comparative analysis of an infrared differential detection mechanism model and a target characteristic model;

step two: infrared differential detection model construction and differential detection factor selection:

selecting the optimal infrared differential dimension and interval as differential detection factors according to the detection dimension output in the step one, and optimizing the obtained factors to enable the difference between dark and weak target components in the output infrared signals and clutter and noise components to be maximum;

the mathematical model of the infrared differential detection model is summarized by the following formula:

f is a method of finding the difference, y₁,y₂,…,y_mIs a response in a different dimension x, x₁,x₂,…,x_nFor n different infrared micro-divisions in the x dimension,representing an optimal set of infrared differential factors.

The construction process of the selective infrared differential detection mechanism-based model is a process of specifically deducing and optimizing the formula (1) and determining a differential factor set. The specific derivation and optimization process is as follows:

the response integral of the target in a certain dimension satisfies:

wherein y (x) is the response relation of the remote sensing instrument to the dimension information, and is obtained according to the signal statistical model in the step one, and for small target detection, the target response integral R_AIMAnd background integral R_BGRespectively comprises the following steps:

wherein y is_AIM(x) And y_BG(x) Respectively, the response of the remote sensing instrument to target and background information in the spectral range dimension. To distinguish the targets, it is clear that R_AIMAnd R_BGThere is a distinction. Assuming a differential mode is used, let

When R is_AIMAnd R_BGWhen the difference is large, namely the difference between the target and the background in the response dimension is large, the target and the background can be obviously distinguished in both the delta R mode and the delta K mode, and the difficulty is that when the R mode is large_AIMAnd R_BGWhen the difference is small, neither Δ R nor Δ K can be distinguished.

The invention introduces an infrared differential detection concept, and an infrared differential formula is expressed as follows:

the formula indicates from a theoretical level that infrared differential detection amplifies the difference between the target and the background. For the infrared differential detection formula, the distinguishing coefficient delta K' is compared with the traditional detection distinguishing coefficient delta K, and obviously when the target R is_AIMAnd background R_BGWith less difference, Δ K ≈ 1, so Δ K 'is required when dual-mode infrared differential detection is performed'<<1 or delta K'>>1, compared with the traditional detection distinguishing coefficient, the distinguishing function of the new distinguishing coefficient is obvious.

In factor setIn the aspect of solving, a feasible method based on the order optimization theory is preliminarily used. The method mainly comprises the following steps:

1) extracting N feasible solutions from the solution set according to uniform distribution;

2) evaluating the feasible solution by using the Euclidean distance to construct an feasible solution sequence curve (OPC);

3) determining a detection set S according to the OPC curve type;

4) selecting an optimal solution from the S by using an optimization model, and obtaining a corresponding fitness error value;

examples

Simulation environment: matlab2020 b;

test input: basic parameters of different types of infrared targets, radiation parameters of different types of backgrounds, parameters of the analog detection system such as track height, read noise, quantization digit, wave band range and the like, and different background parameters.

It is noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.