Author Affiliations
1Key Laboratory of Optoelectronic Science and Technology for Medicine of Ministry of Education, Fujian Normal University, Fuzhou, Fujian 350007, China2Fujian Provincial Engineering Technology Research Center of Photoelectric Sensing Application, Fujian Normal University, Fuzhou, Fujian 350007, China3College of Photonic and Electronic Engineering, Fujian Normal University, Fuzhou, Fujian 350007, China4Shenyang Institute of Automation·Guangzhou·Chinese Academy of Sciences, Guangzhou, Guangdong 510000 China;show less
Fig. 1. Sea-sky line of visible images of sea and air background. (a) Horizontal sea-sky line
[7]; (b)inclined sea-sky line
[16] Fig. 2. Visible panoramic images of sea and air background
[37-38]. (a) Panoramic image; (b) schematic image
Fig. 3. Four typical panoramic sea-sky lines in visual images of sea and air background. (a) Intact; (b) partially lost; (c) blurred; (d) extensively lost
Fig. 4. Infrared images of sea and sky background. (a) Gradient belt; (b) high contrast; (c) low contrast
Fig. 5. Interference images of sea and sky background. (a)(d) Sea clutter image and its infrared image; (b)(e) strong reflected light image and its infrared image; (c)(f) cloud interference image and its infrared image
Fig. 6. Low-quality images of sea and sky background. (a) Visible light image; (b) infrared image
Fig. 7. Test results of different methods in complex environment
Fig. 8. Test results of different methods in complex environment
Fig. 9. Test results of different methods in complex environment
Region | Location | Characteristic |
---|
Sky background region | Above | The reflection is strong and the brightness is large | Sea-sky line region | Middle | A horizontal or inclined line of pixel gradientmaximum runs through the entire image | Sea surface background region | Below | The reflection is weak and the brightness is low |
|
Table 1. Characteristics of ordinary visible images of sea and air background
Region | Characteristic |
---|
Panoramic equipment imaging region [34] | The complexity of the panoramic image backgroundincreases and the sea antenna is fractured | Sky background region | The reflection is strong and the brightness is large | Sea-sky line region | An ellipse or circle of pixel gradient maximum runsthrough the entire image (with a break) | Sea surface background region | The reflection is weak and the brightness is low |
|
Table 2. Characteristics of visible panoramic images of sea and sky background
Region | Location | Characteristic |
---|
Sky background region | Above | The reflection is strong and the brightness is large | Sea-sky line region | Middle | In the figure, the regions with large changes in gray level andthe sea-sky boundary zone with low brightness gradually runthrough the whole background infrared image (with blurred) | Sea surface background region | Below | The reflection is weak and the brightness is low |
|
Table 3. Characteristics of infrared images of sea and sky background
Image type | Characteristic |
---|
Image of ordinary visible sea-sky background | A horizontal or inclined line of pixel gradientmaximum runs through the entire image | Panorama image of ordinaryvisible sea-sky background | An ellipse or circle of pixel gradient maximumruns through the entire image (with a break) | Infrared image of sea-sky background | In the figure, the regions with large changes in gray level andthe sea-sky boundary zone with low brightness gradually run throughthe whole background infrared image (with blurred) |
|
Table 4. Characteristics of sea-sky line in different images
Interference characteristic | Interference factor | Suppression method |
---|
Linear edge characteristic | Sea clutter, strong light reflection, clouds | Nonlinear segmentation, region segmentation,double threshold segmentation, multi-scalewavelet transform, Gaussian smoothing filtering | Low imaging quality | Weather conditions (rain, snow, fog),optical distortion, USV system shaking | Laplace sharpening operator,multi-foot wavelet analysis | Others | Image occlusion of panoramic equipment region | Fractal dimension calculation,full resolution algorithm |
|
Table 5. Summary of disturbance characteristics, disturbance factors and suppression methods
Method | Advantage | Disadvantage | Environmentaladaptability |
---|
Linear fittingbased on RANSAN | Small amount of calculation andgood real-time performance | The selection of candidate points issusceptible to environmental interference | | Based on gradientsignificance | High detection efficiency | Poor noise resistance, notconsidering the sea antenna tilt | Poorenvironmentaladaptability | Based on OTSUsegmentation | Small computation and goodreal-time performance | The threshold value ismore sensitive and the abilityto resist noise is poor | | Based on the Houghtransform | High accuracy | The calculation amount is large,the real time is poor, the engineeringpracticality is low | Strongenvironmentaladaptability | Based onSeam Carving | High detection accuracy | Large amount of calculation | | Our method | Small amount of calculationand high accuracy | Due to environmental interference,the detected sea-sky lineis not strictly linear | Good environmentaladaptability |
|
Table 6. Summary of generation method of sea-sky line for linear model
Method | Advantage | Disadvantage | Environmental adaptability |
---|
Circle fitting | High detectionaccuracy | The prior radius is not easy to obtain,the calculation is large, and thereal-time is poor | Strong environmentaladaptability | Ellipse fitting | When the ellipticity is large, the algorithmfails, the computation is large, and thereal-time performance is poor |
|
Table 7. Summary of generation method of panoramic sea-sky line