Optical and Quantum Electronics (2021) 53:396

[Link]

An infrared dim target detection algorithm based on density

peak search and region consistency

Baohua Zhang1,3 · Jinhui Zhu1 · Xiaoqi Lu2,3 · Yu Gu1,3 · Jianjun Li1,3 · Xin Liu1,3 ·
Ming Zhang1,3

Received: 9 March 2021 / Accepted: 8 June 2021 / Published online: 14 July 2021
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021

Abstract
To suppress background clutter and improve detection accuracy, we propose a dim target
detection algorithm based on density peak search and region consistency. A density peak
search algorithm is first applied to extract candidate targets, and these are then classified
and marked according to the local mosaic probability factor, which is important in order to
suppress the backgroundsssss clutter and accurately strip the candidate target region from
the background. Based on the regional stability of the dim targets, local mosaic gradient
factors are used to screen real targets from candidates, and a facet kernel filter is used to
extract the irregular contours of dim targets with the aim of enhancing them. Our experi-
mental results show that compared with existing algorithms, the proposed method has bet-
ter detection accuracy and robustness in various complex scenarios.

Keywords Small target detection · Density peak search · Local mosaic model · Infrared
images

1 Introduction

An infrared search and tracking system (IRST) has the advantages of good concealment
and robust anti-interference ability, and this approach is widely used in many applications
such as military early warning systems, precision guidance and remote sensing (Zeng et al.
2006; Wang et al. 2019). The method of infrared target detection is a key aspect of this
technology (Chen et al. 2014). However, there are many challenges associated with dim
target detection; for example, the target may only contain dozens of pixels in each frame, or
may lack texture and shape information, and the signal-to-clutter ratio in each frame may

* Jinhui Zhu
zjh_9277@[Link]
1
School of Information Engineering, Inner Mongolia University of Science and Technology,
Baotou 014010, Inner Mongolia, China
2
School of Information Engineering, Mongolia Industrial University, Huhehaote 010051,
Inner Mongolia, China
3
Inner Mongolia Key Laboratory of Pattern Recognition and Intelligent Image Processing,
Baotou 014010, Inner Mongolia, China

13
Vol.:(0123456789)
396 Page 2 of 18 B. Zhang et al.

be very low (Ye et al. 2017; Huang et al. 2019b, a). Dim targets are often submerged in
complex backgrounds (such as cloud edges, ocean waves, and high-brightness noise).
Infrared dim target detection methods can be divided into two types: single-frame (Dai
and Wu 2017) and multi-frame methods (Lv et al. 2018; Wan et al. 2016; Zhang et al.
2019). To detect such targets, Lv used a multi-frame iteration process to estimate the back-
ground. However, optimisation of the coefficients of the filter required a high number of
iterations, so the efficiency of this algorithm was poor (Lv et al. 2018). Wan proposed a
detection method based on the inter-frame difference, in which the differences between
adjacent frames were used to locate the target, but the contours extracted for the target were
incomplete, and the interval between the frames severely affected the detection accuracy
(Wan et al. 2016). Zhang used a quaternion discrete cosine transform to extract the features
of dim targets from multi-frame images, and this approach made full use of the spatiotem-
poral features of four channels. However, target misalignment between multiple frames
reduced the detection accuracy, and the algorithm had a high level of complexity and poor
real-time performance (Zhang et al. 2019). Single-frame methods have many advantages,
such as low complexity, high execution efficiency and strong real-time performance, and
have therefore attracted more attention.
Most traditional single-frame methods, such as those involving morphological filter-
ing (Zeng et al. 2006) or probability statistical filtering (Deshpande et al. 1999; Soni et al.
1993; Tomasi and Manduchi 1998; Barnett, 1989), detect targets by suppressing the back-
ground. Although these methods are easy to implement in real time, their performance
is greatly reduced by clutter and interference when a complex background is present. By
exploiting the characteristics of the human visual system, Chen proposed a local contrast
measure (LCM) that detected a target based on the difference in brightness between the
target and the adjacent areas (Chen et al. 2014). This method was not suitable for scenes
with a high level of background brightness, and a large number of improved methods were
subsequently developed, such as those based on multiscale patch-based contrast measures
(Wei et al. 2016; Nie et al. 2018). These methods were unable to extract the features of
the target, and yielded high false positive rates caused by high-intensity clutter and high-
brightness noise.
Huang proposed a dim target detection method based on maximum grey region grow-
ing, which segmented the target and background by selecting seed growing points (Huang
et al. 2019b, a). However, this method was only suitable for scenarios in which the num-
ber of targets was much smaller than the number of candidates. Xia combined a random
walk algorithm with the local contrast features to achieve efficient target detection, but the
selected seed points in the algorithm needed to be labelled and classified, which was inef-
ficient (Xia et al. 2018). Qin introduced the concept of facet kernel filtering to suppress
clutter and enhance the targets(Qin et al. 2019), the selection of candidate points in this
method was simple, but due to its dependence on a single characteristic of the target, the
false alarm rate was high for complex backgrounds such as those containing strong clutter
and bright edges.
In order to achieve robust detection of dim targets in images with complex backgrounds,
we propose an infrared dim target detection algorithm based on a density peak search and
region consistency. The main contributions of our algorithm are as follows.
We approach the problem of dim target detection as an abnormal point detection problem,
and candidate target points are initially screened out using an unsupervised method based on
an improved random walker (RW) algorithm. Local mosaic template regions are then con-
structed with the candidate target point as the centre pixel, and candidate targets are classified
and labelled locally to suppress clutter. Based on the gradient difference and the directional

13
An infrared dim target detection algorithm based on density… Page 3 of 18 396

consistency of dim infrared targets, a local mosaic probability factor (LMPF) is used to sepa-
rate the target from noise and clutter, and targets are enhanced using a local mosaic gradi-
ent factor (LMGF). Our experimental results show that compared with existing algorithms,
the proposed detection algorithm yields better performance for images containing complex
scenes.

2 Proposed method

The details of the proposed method are as follows. Candidate targets are first identified based
on their abnormal attributes, using a density peak search algorithm. They are then selected as
the central point, and an improved RW algorithm is then applied to construct the local mosaic
template region, thus reducing the feature search range. To suppress clutter, the LMPF is used
to classify and mark candidate targets, which is a critical step in separating targets from clutter.
Based on the direction consistency of the dim targets, the LMGF is then used to enhance the
edges of the targets, which allows them to be extracted from the background. Figure 1 shows a
flow chart for the proposed method.

2.1 Density peak search algorithm

The density peak search algorithm (DPeak) is a granular computing model for which theoreti-
cal basis rests on two assumptions: (1) the central point is surrounded by neighbouring data
points with a lower local density; and (2) two central points are not adjacent to each other
(Wang et al. 2017).
For any pixel i, the DPeak model needs to calculate the local density 𝜌i and the distance to
the nearest correlation distance 𝛿i
𝜌i = fi , (1)

𝛿i = min dij , (2)

i∶𝜌j >𝜌i

where fi represents the gray value of the pixel, the highest density point 𝛿i = max(dij ),
j
| | | |
dij = |xi − xj | + |yi − yj |, (3)
| | | |

Υ = 𝜌 × 𝛿, (4)

Fig. 1  The entire flow chart of the proposed method

13
396 Page 4 of 18 B. Zhang et al.

where dij is the Manhattan distance, and then calculate the joint feature factor Υ pixel by
pixel, and put them in the queue Q in descending order, and extract the first m pixels as
candidate target points.
The grey value of the target is normally higher than that of the local neighbouring pix-
els, meaning that the two assumptions of the DPeak model are applicable to the character-
istics of dim targets. The DPeak model can therefore reduce the search range of the target
and reduce the calculation time.

2.2 Local mosaic probability factor (LMPF) and local mosaic gradient factor (LMGF)

The textural structure of the dim target is essentially different from the clutter. Figure 2
shows a comparison of the gradient distributions: Fig. 2a shows an infrared image, and
Fig. 2b shows an enlarged view of the target area in Fig. 2a. The upper image shows the
gradient distribution of the background clutter, and the lower image shows the gradient
distribution of the target and its neighbourhood. It can be seen that the gradient direction
of the target neighbourhood is consistent and the amplitude is uniform; in comparison,
the gradient direction for the background clutter is messy, and the amplitude difference
is large. Figures2c and d show three-dimensional gradient distribution maps for the back-
ground clutter and the target, respectively. It can be seen that the dim target has different
characteristics, such as higher grey values, a higher gradient and a higher density than the
adjacent background area. The aim of candidate point screening is to judge whether each
point represents a target, based on these local attributes.
The roaming window of the RW algorithm is a rectangular frame, and the outermost
pixels are usually classified as background. In a multi-target scene, if the targets are adja-
cent or touching, these neighbouring targets will be classified as background (Qin et al.
2019), meaning that parts of them will be missed. In order to accurately extract the tar-
get from the background, we improve the RW algorithm by introducing a more efficient
marking model. Unlike the global traversal method, our improved model traverses only the
candidate target area. The roaming window is also changed to an adaptive geometric win-
dow, which more closely reflects the contours of the target. On this basis, the outer area of
the geometric window (background) is marked as 0-type, and the central area (target) is
marked as 1-type, which is useful in terms of accurately segmenting the target. In Fig. 3a,
the red frame indicates the central region, and the area between the yellow and red frames
represents the transition region. The outer area lies between the blue and yellow frames.
The candidate target is used as the central pixel of the local mosaic model, and is marked
as a 1-class pixel, while pixels in the outer area of the mosaic model are marked as 0-class.

Fig. 2  Background clutter, target and its three-dimensional gradient map in infrared image

13
An infrared dim target detection algorithm based on density… Page 5 of 18 396

Fig. 3  Local mosaic model. a The marking strategy in the local mosaic model; b area division diagram of
local mosaic model

Since most dim targets have dimensions of less than 9 × 9 pixels, the size of the template
is set to 11 × 11 pixels. The shape of the marked pixels in the outer region is close to the
target contour, and can accurately approximate the true distribution of the target. The tran-
sition region and the outer region are defined as follows:
MR = (CR ⊕ shape) − CR, (5)

OR = (MR ⊕ shape) − MR, (6)

CR, MR and OR refer to the central region, the transition region and the outer region
respectively. ⊕ refers to the expansion operation, and shape refers to the diamond-shaped
structural element.
In the local mosaic model, only the central pixel is marked as 1-class, while pixels close
to the outer region are marked as 0-class. The pixels in the transition region are classified
based on their similarity to the grey level of the central pixel.
To create a new local mosaic region, the transition region and the outer region are
updated using Eqs. (5) and (6) when the candidate target points have been obtained. In our
approach, the outer area is selected as the background, and the targets are screened using
the LMPF, an indicator that compares the probability value for the candidate target region
with the probability value for the background.
{
MVP(pixels)
, if pixels ⊄ Q and lable = 1
LMPF = MVP(OR) , (7)
0, otherwise

In Eq. (7), MVP (mean value of probability) is the average of pixel probabilities, and
pixels are pixels expanded by candidate target pixels in the central area. LMPF is the gray
value of the central area.
It is difficult to use the LMPF to achieve robust detection of dim targets in highlight
scenes, especially in low-contrast images. We therefore introduce the LMGF to expand the
difference between the target and the background and improve the accuracy of our method.
As shown in Fig. 3b, the template is divided into eight parts based on a three-level nested
model: these are labelled as S1, S2, S3, S4 (where S represents the central area) and T1,
T2, T3, T4 (where T represents the outer area). Here, the LMGF is defined as follows:

13
396 Page 6 of 18 B. Zhang et al.

1 ∑‖
Nsi
gradsi ‖
2
gsi = ‖ , (8)
Nsi k=1 ‖

Gs max = max gsi , (9)

1≤i≤4

Gs min = min gsi , (10)

1≤i≤4

in Eq. (8), gradsi is the gradient element in the area of Si (i = 1, 2, 3, 4), Nsi is the number
of gradients in Si , gsi is the average magnitude of the gradient in Si . Gs max and Gs min are the
maximum and minimum values of the average gradient of Si (i = 1, 2, 3, 4).

1 ∑‖
Nti
gradti ‖
2
gti = ‖ , (11)
Nti j=1 ‖

Gt max = max gti , (12)

1≤i≤4

Gt min = min gti , (13)

1≤i≤4

where, gradti is the gradient element in the area of Ti (i = 1, 2, 3, 4), Nti is the number of
gradients in Ti , gti is the average magnitude of the gradient in Ti . Gt max and Gt min are the
maximum and minimum values of the average gradient of Ti (i = 1, 2, 3, 4).
�∑ � �
4 G G
gti + gsi , if Gs max < Gt max
LMGF = i=1 s min t min , (14)
0, otherwise

where, the LMGF is the gray value of the central area. The direction of the gradient in the
neighbourhood of the dim target is consistent and the amplitudes are similar, meaning that
the gradient direction and amplitude of the clutter are significantly different. The target can
therefore be identified based on this information.

2.3 Target extraction and enhancement

The input image is filtered using 5 × 5 kernel matrix K (Qi et al. 2016) to get the enhanced
image,Pi is original image, as shown in Eq. (15):

⎛ −4 −1 0 −1 −4 ⎞
⎜ −1 2 3 2 −1 ⎟
K=⎜ 0 3 4 3 0 ⎟, Pe = Pi ∗ K, (15)
⎜ ⎟
⎜ −1 2 3 2 −1 ⎟
⎝ −4 −1 0 −1 −4 ⎠
where, the target map Pm is obtained according to Eq. (16), Pm and Pe are calculated by the
Hadamard product to get a weighted image Pw,Pw is called “Weighted”,
Pm = LMPF ∙ LMGF, (16)

13
An infrared dim target detection algorithm based on density… Page 7 of 18 396

Pw = Pm ◦Pe , (17)

Th = 𝜆 ∙ max(Pw ), (18)
in the enhanced image Pe, both the targets and the clutter are enhanced, while Pm only
maps the target area in the proposed method. Therefore, the weighted image Pw is more
approximate the target contour, indicating that the proposed algorithm can effectively sup-
press the clutter and enhance the targets. Th is a simple threshold the value range of 𝜆 is
0.6 ~ 0.9.

2.4 Proposed detection method

As shown in Algorithm 1, the proposed method selects candidate targets by constructing a

feature space for the local density-closest correlation distance, and then constructs the local
mosaic template area with the candidate target point as the central pixel. In order to sup-
press clutter, the candidate target regions are classified and labelled based on the LMPF,
and the real targets are finally selected from the candidate targets based on the LMGF.

13
396 Page 8 of 18 B. Zhang et al.

3 Experiments and analysis

In this section, the experiments are performed to verify the effectiveness of the proposed
method. The experimental computing environment in this paper is 3.40 GHz Intel i7-3770
CPU processor, 8 GB memory, and the test software is MATLAB 2018b.

3.1 Experimental setup

(1) Datasets: To verify the robustness and accuracy of the proposed method, six sets of
infrared image sequences with different complex backgrounds were selected. Table 1
gives detailed information on these six sequences. Each sequence has a different back-
ground: Frame 1 contains the edge of the sky and irregular clouds; Frame 2 has a glare

13
An infrared dim target detection algorithm based on density… Page 9 of 18 396

Table 1  Details of six infrared target datasets

Frame Frame number Target size Scene Clutter

1 100 3×3 Cloudy sky Strong edges of cloudy shape

2 66 2×5 Sky bright and heavy noise
3 80 3×4 Sky a circular clutter
4 60 4×5 Cloudy sky Complicated cloud clutters
5 67 4 × 9 and 4 × 12 Sea Bright noise and lined shape clutters
6 85 5 × 7, 5 × 6, 5 × 5 and 4 × 4 Sea and sky Heavy noise

background(Huang et al. 2019b, a); Frame 3 contains a large amount of ring-shaped

point clutter; in Frame 4, the target is surrounded by clouds, which form a great deal of
clutter(Qin et al. 2019); Frame 5 shows a dual-target image sequence with high-intensity
sky and sea wave clutter as the background; and Frame 6 is a four-target image sequence
with low intensity(Xia et al. 2018) (Table 2).
(2) Baseline Methods: This paper selects the existing 9 methods as comparison meth-
ods, including Top-Hat (Zeng et al. 2006), Fast-Saliency (Qi et al. 2016), VARD (Nasiri
and Chehresa, 2017), LCM (Chen et al. 2014), MPCM (Wei et al. 2016), AAGD (Deng
et al. 2016), NRAM (Zhang et al. 2018a, b),Max-Median (Deshpande et al. 1999) and LIG
(Zhang et al. 2018a, b).
(3) Evaluation metrics:
𝜎in
BSF =
𝜎out
, (19)

Cout
CG = , (20)
Cin
𝜎in and 𝜎out represent the intensity standard deviation of the source image and the processed
image respectively; Cin and Cout are the contrast between the source image and the pro-
cessed image. The contrast is defined as Eq. (21):

Table 2  Parameter settings of different methods

No. Methods Parameter settings

1 Top-Hat Local window size: 5 × 5

2 Fast-Saliency Gaussian filter size: 5 × 5, σ = 1.5
3 VARD External, Guard and Internal Patch size: 15 × 15, 11 × 11, 7 × 7
4 LCM Local window size: 8 × 8, sliding steps of row and column: 4
5 MPCM L = 3, Local window size: 3 × 3, 5 × 5, 7 × 7
6 AAGD Internal and external window size:3 × 3,19 × 19
�√
7 NRAM Patchsize:50 × 50,slidingstep:10,𝜆 = 1� min (m, n)
√ √
,𝜇0 = 3 min (m, n),c = min (m, n) 2.5
8 Max-Median Local window size: 15 × 15
9 LIG K = 0.2,n = 19
10 Proposed window size: 11 × 11, 𝜇 = 100

13
396 Page 10 of 18 B. Zhang et al.

C = ||ft − fb ||, (21)

where, ft is the maximum intensity value of the target area and fb is the average value of
the background area. In this paper, we use BSF and CG as evaluation indicators. BSF can
be used to evaluate the ability of the algorithm to suppress clutter globally, and its value is
proportional to the suppression effect, while CG can be used to evaluate the effect of target
enhancement, and its value is proportional to this effect.
number of true target detection
TPR = , (22)
total number of actual targets

number of detected false targets

FPR = , (23)
total number of pixels in images

TPR and FPR represent the true positive and false positive rates, respectively, and are
used to draw the ROC curve and evaluate the algorithm: the closer the position of this
curve to the upper left corner of the graph, the better the ability of the algorithm.

3.2 Qualitative comparisons

Figure 4 shows a comparison of the detection results from the source images based on the
LMPF, the LMGF and the weighted method, the weighted method represents Pw. It can be
seen that for Frames 1, 2 and 4, the detection results based on the LMPF still contain point
shadow clutter. This indicates that the clutter suppression ability of the LMGF is better
than that of the LMPF. However, in terms of target edge retention and target enhancement,
the LMPF yields better performance than the LMGF, except for Frame 5; due to the influ-
ence of the strong light and sea clutter, the effectiveness of the LMPF is slightly lower than
that of the LMGF in this case.
In order to verify the performance of this algorithm, we used six datasets in our detec-
tion experiments, including both single-target and multi-target image sequences. To quali-
tatively analyse the effect of each step of the proposed method, each source image was
processed using LMPF, LMGF and the weighted model, respectively, and the BSF and CG
were calculated for the outputs. Table 3 shows the results. It can be seen that the results
of the weighted model were better than those of the LMPF or the LMGF models, mean-
ing that the weighted model can enhance the target and suppress background clutter more
effectively.
In order to evaluate the performance of our algorithm, six datasets were used as experi-
mental data, including both single-target and multi-target image sequences. Nine alterna-
tive methods were selected for comparison: the top-hat, max-median, fast-saliency, LCM,
MPCM, VARD, AAGD, NRAM and LIG methods. Figures 5 and 6 show the detec-
tion results for these six datasets. The four datasets in Fig. 5 all contained single-target
sequences, and their backgrounds included high-intensity light, cloud clutter and complex
irregular clutter. The two datasets shown in Fig. 6 were two-target and four-target image
sequences, respectively, and were used to verify the multi-target detection ability of our
algorithm.
The source images in Fig. 5 are four sets of image sequences showing areas of sky.
In Frame 1, the background is the sky and irregular cloud edges, forming a lot of clut-
ter; Frame 2 has a light background with strong intensity, and the background brightness

13
An infrared dim target detection algorithm based on density… Page 11 of 18 396

Fig. 4  The detection results of each stage of the algorithm (each image uses a red rectangular frame to dis-
play the real target area and an enlarged view of the target area in the lower left corner of the image)

is similar to that of the target; the background of Frame 3 contains a great deal of ring-
shaped point clutter; and the target in Frame 4 is surrounded by clouds, which create clutter
interference.
The detection results from the top-hat and the max-median methods contain a great
deal of clutter, indicating that they have poor effectiveness in terms of background sup-
pression, and the top-hat method loses the target in Frame 2. The detection results from
the LCM and the MPCM methods not only contain background clutter, but also have

13
396 Page 12 of 18 B. Zhang et al.

Table 3  BSF and CG values of Methods LMPF LMGF Weighted

each step of the method in all
image frames
Frame 1 5.640 5.754 9.480
Frame 2 4.955 9.277 12.370
BSF Frame 3 8.396 9.069 12.003
Frame 4 4.474 5.328 7.112
Frame 5 13.748 14.020 19.690
Frame 6 7.385 9.981 9.983
Frame 1 2.486 2.470 2.486
Frame 2 8.121 8.121 8.121
CG Frame 3 2.671 2.457 2.682
Frame 4 5.250 5.248 5.257
Frame 5 1.563 2.013 2.252
Frame 6 3.344 2.297 3.564

a high false alarm rate. In contrast, there are few background clusters in the results of
the fast-saliency method. The other comparison methods also show some problems: the
AAGD method gives a small amount of linear clutter in Frames 1 and 4; the NRAM
method loses the target in Frame 1; and the LIG method gives a small number of back-
ground clusters in Frame 4. In general, our proposed method effectively suppresses the
background clutter and preserves the contours of the target, and gives the best detection
performance.
The experimental dataset shown in Fig. 6 contains two-target and four-target image
sequences with the sea and sky as background, and these are used to verify the multi-target
detection performance of our approach. Frame 5 is a dual-target image sequence with a
strongly lit sky and clutter from sea waves. The background suppression effects from the
top-hat, fast-saliency, VARD, AAGD, NRAM and LIG methods are good. Since there is
a great deal of clutter in the detection results from the LCM, MPCM and max-median
methods, this means that the background suppression effects are poor. Frame 6 is a four-
target image sequence with low brightness. The fast-saliency and VARD methods lose the
target, and the LCM, max-median and MPCM methods have poor background suppres-
sion effects. The targets detected by the NRAM and the LIG methods are blurry, while
the contours extracted with the top-hat and AAGD methods are clear. Compared with the
alternative methods, our approach has obvious advantages in terms of background clutter
suppression and target enhancement, and yields better performance for all of the datasets.

3.3 Quantitative comparisons

The BSF, CG, and ROC curves and the average running time of the algorithm were used
to further evaluate the detection performance of our method. The BSF was used to evalu-
ate the background suppression capability, while the CG was applied to evaluate the target
enhancement capability of the algorithm. Table 4 shows the results for the BSF, CG and
average running time for our algorithm.
In Table 4, the BSF of the proposed method is best for Frames 1, 2, 4, and 5, while the
NRAM method performs best for Frames 3 and 4. This shows that the background suppres-
sion ability of our method is stronger than that of the alternative methods.

13
An infrared dim target detection algorithm based on density… Page 13 of 18 396

Fig. 5  Four original image sequences and detection results under different methods (each image uses a red
rectangular frame to display the real target area and an enlarged view of the target area in the lower left cor-
ner of the image)

CG is the target local contrast gain. The larger the value of this parameter, the better
the background suppression and the target enhancement performance. The CG values
from the proposed method were the highest for all image sequences.
In terms of the running time, the fast-saliency method required the shortest time,
and AAGD took the shortest time in the remaining five image sequences. The NRAM
method required the longest time for all image sequences. The running time for our
method was between these two extremes, taking into account the detection accuracy and
computing efficiency.
Figure 7 shows a comparison of the ROC curves for these algorithms. It can be
clearly seen that the ROC curve for our method is the highest.

13
396 Page 14 of 18 B. Zhang et al.

Fig. 6  Two original image sequences and detection results under different methods (each image shows the
real target area with a red rectangular frame)

3.4 Parameter Sensitivity Analysis

In our method, the target detection problem is treated as a clustering problem with unsu-
pervised learning, meaning that the selection of the number of candidate points is the main
parameter that determines the performance. The weights between adjacent pixels strongly
affect the location of the target contour, and we use the parameter 𝜇 to determine these
weights. In this section, we report the results of experiments performed to analyse the
influence of this parameter on the test results.
Since dim targets were often submerged in clutter, we assumed that most of the seed
points were false targets and a few were target points, meaning that the target points formed
a set of abnormal points. Figure 8 shows TPR diagrams for varying numbers of candidate
points m for six groups of image sequences. According to the experimental results, this
proposed method sets m to 12.
Figure 9 shows the distribution of the ROC curve for different values of the parameter
𝜇 for the proposed method. It can be seen that if this parameter is set to 100, the results for
the ROC curve are best for all the image sequences. If it is set to 200, the curve is slightly
lower, and it is set to zero or 300, the target may be missed or false targets may be counted,
causing poor detection results. We therefore set this parameter to 100.

4 Conclusion

In this paper, a density peak algorithm is used to traverse each pixel in an image to
establish a density-distance spatial decision map. Candidate points are then selected
based on a proportional relationship between noise points and dim targets. A local
mosaic model is built with the candidate pixels as the central points, and the targets can
be screened from the candidate targets based on the mapping relationship between the
LMPF and the LMGF, which narrows the target search range and reduces the amount

13
 Table 4  BSF, CG and average running time of different algorithms under each image frame
Methods Proposed Top-hat Fast-saliency VARD LCM MPCM AAGD NRAM Max–median LIG

BSF Fra1 9.480 3.754 2.726 8.718 1.356 5.947 5.111 7.432 6.075 8.842
Fra2 12.370 5.190 1.686 6.229 0.414 3.218 4.090 7.322 1.045 6.107
Fra3 12.003 1.648 2.271 5.634 1.339 2.514 2.141 24.345 2.313 19.925
Fra4 10.099 0.723 2.613 7.578 0.709 2.080 1.656 7.112 1.227 3.499
Fra5 19.690 9.436 9.350 17.215 1.730 7.819 12.314 18.115 4.407 17.257
Fra6 9.983 7.683 10.346 13.714 1.065 5.487 5.638 17.825 4.364 16.589
CG Fra1 2.486 2.385 2.268 2.485 1.914 1.634 2.476 0.565 1.502 2.486
Fra2 8.121 0.948 7.555 8.105 3.368 6.090 8.050 8.089 4.907 8.093
Fra3 2.682 2.473 2.605 2.325 1.588 0.479 0.698 2.188 1.228 2.682
An infrared dim target detection algorithm based on density…

Fra4 5.257 4.738 1.334 5.239 1.714 1.752 2.511 3.999 2.088 5.244
Fra5 2.252 0.962 0.608 1.199 1.264 1.175 2.206 1.943 1.453 2.013
Fra6 3.564 3.497 1.701 3.005 1.893 0.862 2.444 1.486 1.100 1.571
Time Fra1 0.580 0.570 0.026 0.067 0.116 0.103 0.049 0.830 0.725 0.694
Fra2 1.140 0.520 0.093 0.073 0.159 0.137 0.059 3.810 3.234 2.499
Fra3 1.030 0.430 0.069 0.069 0.150 0.126 0.052 3.110 2.658 2.213
Fra4 0.820 0.510 0.078 0.080 0.152 0.131 0.066 2.040 1.983 1.716
Fra5 1.250 0.470 0.106 0.082 0.167 0.153 0.064 5.920 3.472 3.175
Fra6 0.840 0.420 0.075 0.076 0.146 0.133 0.059 4.090 2.577 2.562

The bold font indicates the best result

Page 15 of 18 396

13
396 Page 16 of 18 B. Zhang et al.

Fig. 7  ROC curves of different image frames

Fig. 8  TPR curves of different parameters m in this algorithm

of calculation required. The proposed method can also accurately extract targets from
multi-target image sequences. Experimental results show that our method performs bet-
ter than alternative approaches, as it not only suppresses clutter and enhances the target,
but also has a wide range of applications. In future work, we intend to develop a new
positioning method for the target region, reduce the running time further, and build a
robust feature space for screening real targets.

13
An infrared dim target detection algorithm based on density… Page 17 of 18 396

Fig. 9  TPR curves of different parameters 𝜇 in this algorithm

Acknowledgements The authors thank the anonymous reviewers and editors for the very constructive com-
ments. This work was supported by the National Natural Science Foundation of China(61962046,616630
36,61841204). Inner Mongolia Outstanding Youth Cultivation Fund (2018JQ02). Inner Mongolia Natural
Science Foundation (2015MS0604).

Declarations

Conflict of interest The authors declare no conflicts of interest.

References
Barnett, J.: Statistical analysis of median subtraction filtering with application to point target detection in
infrared backgrounds. Proc. SPIE. 1050, 10–18 (1989)
Chen, C.L.P., Li, H., Wei, Y.T., Xia, T., Tang, Y.Y.: A local contrast method for small infrared target detec-
tion. IEEE Trans. Geosci. Remote Sens. 52, 574–581 (2014)
Dai, Y., Wu, Y.: Reweighted infrared patch-tensor model with both nonlocal and local priors for single-
frame small target detection. IEEE J. Sel. Top. Appl. Earth Observ. Remote Sens. 10, 3752–3767
(2017)
Deng, H., Sun, X., Liu, M., Ye, C., Zhou, X.: Infrared small-target detection using multiscale gray differ-
ence weighted image entropy. IEEE Trans. Aerosp. Electron. Syst. 52, 60–72 (2016)
Deshpande, S.D., Er, M.H., Venkateswarlu, R., Chan, P.: Max-mean and max-median filters for detection of
small targets. Proc. SPIE. 3809, 74–83 (1999)
Huang, Z., Chen, L., Zhang, Y., Yu, Z., Fang, H., Zhang, T.: Robust contact-point detection from panto-
graph-catenary infrared images by employing horizontal-vertical enhancement operator. Infrared Phys.
Technol. 101, 146–155 (2019a)
Huang, S., Peng, Z., Wang, Z., Wang, X., Li, M.: Infrared small target detection by density peaks searching
and maximum-gray region growing. IEEE Geosci. Remote Sens. Lett. 16, 1919–1923 (2019b)
Lv, P.-Y., Sun, S.-L., Lin, C.-Q., Liu, G.-R.: Space moving target detection and tracking method in complex
background. Infrared Phys. Technol. 91, 107–118 (2018)
Nasiri, M., Chehresa, S.: Infrared small target enhancement based on variance difference. Infrared Phys.
Technol. 82, 107–119 (2017)

13
396 Page 18 of 18 B. Zhang et al.

Nie, J., Qu, S., Wei, Y., Zhang, L., Deng, L.: An infrared small target detection method based on multiscale
local homogeneity measure. Infrared Phys. Technol. 90, 186–194 (2018)
Qi, S., Xu, G., Mou, Z., Huang, D., Zheng, X.: A fast-saliency method for real-time infrared small target
detection. Infrared Phys. Technol. 77, 440–450 (2016)
Qin, Y., Bruzzone, L., Gao, C., Li, B.: Infrared small target detection based on facet kernel and random
walker. IEEE Trans. Geosci. Remote Sensing. 57, 7104–7118 (2019)
Soni, T., Zeidler, J.R., Ku, W.H.: Performance evaluation of 2-D adaptive prediction filters for detection of
small objects in image data. IEEE Trans. Image Process. 2, 327–340 (1993)
Tomasi, C., Manduchi, R.: Bilateral filtering for gray and color images. In: Sixth international conference on
computer vision (ICCV ’98), pp. 839–846. IEEE Computer Society (1998)
Wan, M., Gu, G., Cao, E., Hu, X., Qian, W., Ren, K.: In-frame and inter-frame information based infrared
moving small target detection under complex cloud backgrounds. Infrared Phys. Technol. 76, 455–467
(2016)
Wang, X., Yang, L.T., Li, H., Lin, M., Han, J., Apduhan, B.O.: NQA: a nested anti-collision algorithm for
RFID systems. ACM Trans. Embed. Comput. Syst. 18, 1–21 (2019)
Wang, G., Yang, J., Xu, J.: Granular computing: from granularity optimization to multi-granularity joint
problem solving. Granul. Comput. 2, 105–120 (2017)
Wei, Y., You, X., Li, H.: Multiscale patch-based contrast measure for small infrared target detection. Pattern
Recognit. 58, 216–226 (2016)
Xia, C., Li, X., Zhao, L.: Infrared small target detection via modified random walks. Remote Sens. 10, 2004
(2018)
Ye, Y., Shan, J., Bruzzone, L., Shen, L.: Robust registration of multimodal remote sensing images based on
structural similarity. IEEE Trans. Geosci. Remote Sens. 55, 2941–2958 (2017)
Zeng, M., Li, J., Peng, Z.: The design of top-hat morphological filter and application to infrared target detec-
tion. Infrared Phys. Technol. 48, 67–76 (2006)
Zhang, L., Peng, L., Zhang, T., Cao, S., Peng, Z.: Infrared small target detection via non-convex rank
approximation minimization joint l2, 1 norm. Remote Sens. 10, 1821 (2018a)
Zhang, P., Wang, X., Wang, X., Fei, C., Guo, Z.: Infrared small target detection based on spatial-temporal
enhancement using quaternion discrete cosine transform. IEEE Access 7, 54712–54723 (2019)
Zhang, H., Zhang, L., Yuan, D., Chen, H.: Infrared small target detection based on local intensity and gradi-
ent properties. Infrared Phys. Technol. 89, 88–96 (2018b)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

13