CN101980283A - A Dynamic Blind Element Compensation Method - Google Patents

Abstract

The invention discloses a method for dynamically compensating a blind pixel. The method comprises the following steps of: arranging pixel gray values of a 3*3 window matrix dot picture element from small to large according to the remarkable abnormity of the blind pixel in a local window; if the absolute value of the difference value between a medium value and other picture element gray value is greater than a preset threshold value, determining the picture element as the blind pixel; detecting all picture elements of an M*N array by using a medium gray value instead of a blind pixel gray value; and compensating all blind pixels. Therefore, a high-quality image is obtained in a display device.

Description

A Dynamic Blind Element Compensation Method

technical field

The invention relates to the technical field of uncooled infrared detection, in particular to an optimized dynamic blind element compensation method in infrared image processing.

Background technique

As the latest generation of detectors, Infrared Focal Plane Array (IRFPA) has been widely used in various military and civilian fields. Due to the influence of manufacturing process, materials and other factors, the IRFPA device inevitably has the problem of blind elements, which affects the signal-to-noise ratio of the image output by the infrared imaging system. Blind cells, or invalid cells, refer to the cells whose response is too high or too low in the IRFPA device. The number and distribution of blind pixels seriously affect the output image quality of infrared imaging systems. Blind pixels appear as bright or dark spots in the image. Detection and compensation of blind pixels in the imaging stage can help improve image quality.

式中i＝1～M，j＝1～N，V_s(i，j)为第(i，j)像元对应于辐照功率P的响应电压，P为第(i，j)像元所对接受的辐照功率。For M×N infrared focal plane array detectors, the pixel responsivity R _{(i, j)} is the output signal voltage generated by the pixel for each unit of irradiation power under the condition of one frame period and a certain dynamic range of IRFPA,

In the formula, i=1~M, j=1~N, V _s (i, j) is the response voltage of the (i, j)th pixel corresponding to the irradiation power P, and P is the (i, j)th pixel The received radiation power.

式中M和N分别是IRFPA像元的行数和列数；d和h分别是死像元数和过热像元数。实际测量中，d和h是经过多次迭代计算得到。The average value of the response rate of each effective pixel of IRFPA,

In the formula, M and N are the number of rows and columns of IRFPA pixels, respectively; d and h are the number of dead pixels and overheated pixels, respectively. In actual measurement, d and h are calculated through multiple iterations.

Blind pixels include dead pixels and overheated pixels. According to the national standard GB/T17444-1998 "Technical Standards for Infrared Focal Plane Acceptance Test", dead pixels are pixels whose response rate is lower than 1/10 of the average response rate, and overheated pixels A cell is a cell with a response rate 10 times higher than the average response rate. Generally speaking, the response imaged by the normal detection unit on the IRFPA changes linearly with the external temperature within a certain dynamic range, as shown in curve 2 in Figure 1; the blind unit is different, its response principle is the normal dynamic range, and Generally do not change with the external environment. In addition, according to the size of the response value, blind pixels include dead pixels and overheated pixels, as shown in curve 1 and curve 3 in Figure 1, respectively.

The processing of blind cells includes two aspects of blind cell detection and blind cell compensation. Blind element detection is the premise and foundation of blind element compensation, and improper detection will bring additional noise to infrared images. Blind pixel compensation is the process of predicting and replacing the information of the blind pixel position using the effective image information around the blind pixel or the image information of the previous and subsequent frames. Therefore, the idea of blind pixel compensation has two directions: first, time compensation, that is, using Inter-frame correlation of sequential images, obtaining compensation information from adjacent frames. Its advantage is that it can keep the edge of the image very well, but its disadvantage is that it has a strong dependence on the front and back frames; second, spatial compensation, which uses the pixel information around the blind pixel to compensate it. Such as linear interpolation compensation, median filter, etc. The advantage of this method is that the process is simple and operable.

Contents of the invention

The problem to be solved by the present invention is: how to provide a dynamic blind element compensation method, which has strong operability and good versatility, and can effectively detect and compensate blind elements.

The technical problem proposed by the present invention is solved like this: provide a kind of dynamic blind element compensation method, for the infrared focal plane array of M * N, suppose S (i, j) be that the center is (i, j), and size is 3 ×3 window matrix, where i ∈ (1, M), j ∈ (1, N), the pixel grayscale of each pixel in the window is recorded as S _K (i, j), k=1, 2, .. .9, S _K (i, j) is arranged from S ₁ (i, j) to S ₉ (i, j) from small to large, and the pixel grayscale of the median value is S ₅ (i, j), including The following steps:

Step 1: Move the 3×3 window matrix S(i, j) in the row direction along the image data, that is, the center point (i, j) traverses all points from point (2, 2) to (2, j), j=2,3...N-2;

Step 2: Move one point each time, arrange the pixel gray values of all pixels in S(i, j) from small to large, and take the middle value S ₅ (i, j);

Step 3: Calculate the difference ΔS _k (i, j) _and ΔS _k (i, j) between other pixel gray values S k (i, j) and the median S ₅ (i, j) )=|S ₅ (i, j)-S _k (i, j)|;

Step 4: Compare all ΔS _k (i, j) with the preset threshold δ, if ΔS _k (i, j) > δ, then the point is a blind element, and replace the pixel of this point with S ₅ (i, j) Gray value S _k (i, j);

Step 5: Return to step 1 until j=N-1, skip to the next row, and start traversing a new row;

Step 6: Traverse all non-edge point pixels until (i, j),

S = {(i, j) | i ∈ [2, M-1], j ∈ [2, N-1]}.

The infrared imaging system is mainly for real-time dynamic imaging of the scenery. The present invention provides an optimized median filtering blind element compensation technology, which has a simple algorithm and can detect and compensate dynamic blind elements. The beneficial effect lies in that the algorithm is simple, easy to realize by hardware, has good versatility, can effectively detect blind elements, and maintain image edge details. The combination of blind pixel detection and blind pixel compensation can be realized, and blind pixel compensation can be effectively performed while quickly detecting blind pixels without marking the position of blind pixels. It is a real-time processing dynamic blind element compensation algorithm, which can detect and compensate the fixed blind element of the infrared focal plane array by using the significant difference in the response of the effective pixel and the blind element in the local window during the scene movement. , can also detect and compensate random blind pixels of images.

Description of drawings

Figure 1 is a schematic diagram of the pixel response curve of the infrared focal plane array detector, curve 1 is a dead pixel, curve 2 is a normal pixel, and curve 3 is an overheated pixel;

Fig. 2 is a marker diagram of the pixel gray values of each pixel in the 3×3 window matrix S(i, j) arranged from small to large, and k=1, 2...9 in the box represent the pixels Pixel gray value S ₁ (i, j), S ₂ (i, j)...S ₉ (i, j), where the gray median value is always marked as S ₅ (i, j);

The gray box 2 in the middle of Fig. 3 is the pixel point set S={(i, j) | ∈ [2, M-1], j ∈ [2, N-1]}.

Detailed ways

The present invention will be further described below in conjunction with accompanying drawing:

Detailed technical scheme of the present invention is:

For an M×N infrared focal plane array, let S(i, j) be a window matrix with a center at (i, j) and a size of 3×3, where i∈(1,M), j∈(1,N ), the pixel grayscale of each pixel in the window is denoted as S _k (i, j), k=1, 2K 9 . S _k (i, j) is arranged from small to large from S ₁ (i, j) to S ₉ (i, j), and the gray level of the median pixel is S ₅ (i, j), as shown in Figure 2.

The process of median filter blind element compensation algorithm is as follows:

(1). Move the 3×3 window matrix S(i, j) along the row direction of the image data, that is, the center point (i, j) traverses all points (2, j) from point (2, 2), j =2,3...N-2;

(2). Move one point each time, arrange the pixel gray values of all pixels in S(i, j) from small to large, as shown in Figure 2, take the middle value S ₅ (i, j);

(3). Calculate the difference ΔS _k (i, j) and ΔS _k (i, j) between other pixel _gray values S k (i, j) and the median S ₅ (i, j) except the median value j)=|S ₅ (i, j)-S _k (i, j)|

(4). Compare all ΔS _k (i, j) with the preset threshold δ, if ΔS _k (i, j) > δ, then the point is a blind element, and replace this point with S ₅ (i, j) Pixel gray value S _k (i, j)

(5). Return to step (1), until j=N-1, skip to the next row, and start traversing a new row;

(6). Until (i, j) traverses all non-edge point pixels, S={(i, j)|i∈[2, M-1], j∈[2, N-1]}, such as image 3.

According to the local significant abnormality of the blind cells, use the window matrix S(i, j) whose center is (i, j) and whose size is 3×3, to traverse and detect the blind cells of the image formed by M×N infrared focal planes. First arrange the gray values of all pixels in S(i, j) from small to large, and mark them as S ₁ (i, j), S ₂ (i, j)...S ₉ (i, j), The median value is S ₅ (i,j). Take the difference between the median value and other gray values, the absolute value of the difference ΔS _k (i, j)=|S ₅ (i, j)-S _k (i, j)|, compare ΔS _k (i, j) j) and the size of the preset threshold δ, if ΔS _k (i, j) > δ, then the point is a blind element, and S ₅ (i, j) is used to replace the gray value of the point pixel S _k (i, j) . Use the window matrix S(i, j) to detect and compensate all blind pixel points of the M×N pixel array along the row direction.

Claims (1)

1. dynamic blind element compensation method, for the infrared focal plane array of M * N, establish S (i, j) be the center be (i, j), size is 3 * 3 window matrix, wherein (1, M), (1, N), the pixel grey scale of each pixel is designated as S to j ∈ to i ∈ in the window _k(i, j), k=1,2 ... 9, S _k(i is j) from S ₁(i j) begins to S ₉(wherein the pixel grey scale of intermediate value is S for i, j) ascending arrangement ₅(i j), is characterized in that, may further comprise the steps:

Step 1: with 3 * 3 window matrix S (i j) carries out line direction along view data and moves, promptly central point (i, j) traversal from point (2,2) begin to (2, j) all point, j=2,3...N-2;

Step 2: move a point, (i, j) grey scale pixel value of Nei all pixels is got intermediate value S by arranging from small to large with S at every turn ₅(i, j);

Step 3: calculate other pixel gray-scale values S except that intermediate value _k(i is j) with intermediate value S ₅(i, difference DELTA S j) _k(i, j), Δ S _k(i, j)=| S ₅(i, j)-S _k(i, j) |;

Step 4: all Δ S relatively _k(i, j) and the size of predetermined threshold value δ, if Δ S _k(i, j)＞δ, then this point is a blind element, uses S ₅(i j) replaces this grey scale pixel value S _k(i, j);

Step 5: return step 1,, jump to next line, begin to travel through new delegation up to j=N-1;

Step 6: up to (i j) travels through the some pixel at all non-edges,

S＝{(i，j)|i∈[2，M-1]，j∈[2，N-1]}。

Images