Infrared weak target detection method based on foreground weighted local contrast
A local contrast, weak target technology, applied in the field of image processing, can solve problems such as detection performance degradation, and achieve the effects of strong suppression, fast calculation speed, and strong practicability
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 3
[0094] Embodiment 3, a single-scale acceleration algorithm, including: calculating the local contrast of a pixel:
[0095] The neighborhood structure is used to slide the window from top to bottom and from left to right in the original image; where the neighborhood structure includes the target area in the center and the surrounding area, the target area is a square, containing one or more pixels, assuming white The area is the target area where the pixel I(x, y) is located, containing M pixels in total; the gray area is the surrounding area of the target, containing N pixels in total, and the gray level average of the target area of I(x, y) and the surrounding area The values are:
[0096]
[0097] I j is the gray value of the pixel I(x,y) in the target area, I k is the gray value of the pixel I(x,y) in the surrounding area; then, the local contrast of the pixel I(x,y) is defined as:
[0098] D(x,y)=|m t (x,y)-m s (x,y)|
[0099] When D(x,y)≥T D , calculate ...
Embodiment 4
[0107] Embodiment 4, multi-scale acceleration algorithm, including:
[0108] Compute the local contrast of a pixel:
[0109] The neighborhood structure is used to slide the window from top to bottom and from left to right in the original image; the neighborhood structure includes the target area in the center and the surrounding area, and the target area is square and contains one or more pixel units. Suppose The white area is the target area where the pixel I(x,y) is located, which contains M pixels in total; the gray area is the surrounding area of the target, which contains N pixels in total, and the grayscale of the target area of I(x,y) and the surrounding area The average values are:
[0110]
[0111] I j is the gray value of the pixel unit I(x,y) in the target area, I k is the gray value of the pixel unit I(x,y) in the surrounding area; then, the local contrast of the pixel I(x,y) is defined as:
[0112] D(x,y)=|m t (x,y)-m s (x,y)|
[0113] When D(x,y...
experiment example
[0125] This application selects 4 representative video sequences containing weak infrared targets from practical applications. First, one person labels all the images and all objects in the video sequence, and then another person checks, checks, and modifies them frame by frame and object by object to ensure the correctness of the annotations. Finally, the annotation results are used as the evaluation benchmark for the performance of all algorithms (true value). The characteristics of different video sequences are shown in Table 1, and typical images in video sequences are as follows Figure 17-20 shown.
[0126]
[0127] Table 1 Details of infrared video sequences in different scenarios
[0128] First, all algorithms are tested on all video sequences. Calculate the detection rate P of all algorithms on each video sequence d and the false alarm rate F a , the obtained ROC curve is as Figure 21-24 shown.
[0129] Depend on Figure 21-24 It can be seen that ours is t...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More