A background separation method of nonuniform image segmentation

Liu, Junqiang; Gao, Jianming; Shen, Qingming; Tian, Junwei

doi:10.1109/iciea.2009.5138708

Cited by 6 publications

(3 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous studies [2], [62] have indicated that the edges and lines of objects exhibit a stronger association with high frequency components, while the frequency of the background and object surface tends to be relatively low [41]. With this understanding, we hypothesize that by emphasizing learning from high frequency components, the model can more effectively capture the semantic features of objects while mitigating the risk of learning spurious correlations.…”

Section: Two-step High-pass Filtermentioning

confidence: 80%

“…(1) Frequency restriction. Prior studies [2], [62] have revealed that the edges and lines of objects are more related to the high frequency components, while the frequency of background and object surface is relatively low [41]. Consequently, assuming that by learning more from the high frequency components, the model can better extract the semantic concepts of objects while avoiding learning spurious correlations between backgrounds and labels.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

School of Teacher, Nanjing Xiao Zhuang University, Nanjing, 210000, China

Wang¹

2017

Advances in Psychological Science

View full text Add to dashboard Cite

The nuclide 14 C can be produced in the atmosphere by high energy particles and γ-rays from high-energy phenomena. Through the carbon cycle, some of 14 CO 2 produced in the atmosphere can be retained in annual tree rings 1, 2 . Four events of rapid increase of the 14 C content occurred in AD 775, AD 994, BC 660 and BC 3371 were found 3-6 . Recently, the data of Jull et al. (2020) was inconsistent with our records around BC 3371. We measured our sample again and found the 14 C records are consistent with the value in Wang et al. (2017). Therefore, our 14 C records are robust. The inconsistency may be caused by the difference of calendar ages for the wood samples, or the physical origin of the event. First, crossdating on ring width can be performed only between trees whose growth has the same environmental conditions. Because the master tree-ring for dendrochronology is lack for Chinese trees. The master tree-ring from California has to be used. Therefore, the calendar ages derived from dendrochronology may be not precise. Second, the 14 C even may be not global. One evidence

show abstract

Section: Two-step High-pass Filtermentioning

confidence: 80%

Section: Introductionmentioning

confidence: 99%

School of Teacher, Nanjing Xiao Zhuang University, Nanjing, 210000, China

Wang¹

2017

Advances in Psychological Science

View full text Add to dashboard Cite

show abstract

“…This paper is aimed at inland ships recognition. Since the difference from the adjacent frames caused by low speed ship is not obvious, adopting the frame difference method would lead to the "cavity" phenomenon within ship image [1,2]. The optical flow is quite complex, and its anti-noise performance is poor without the special hardware devices, it cannot be used in the real-time processing of full-frame video stream [3].…”

Section: Introductionmentioning

confidence: 99%

Inland Ship Recognition Algorithm Based on Neighboring Static Region Substitution for Background Updating

Wang

Xie

Yan

et al. 2013

AMM

View full text Add to dashboard Cite

Considering the particular scene that the moving ships in Inland River are surrounded by the nearly static water, this paper presents a method that updates the background by adopting the neighboring static region to replace the dynamic areas. By making the difference between the current frame and the background image to get the foreground area, and using the region growing algorithm to shape a complete ship region, we can get the coordinate, the center point and the direction of its movement. To obtain its mean gray value to replace the foreground dynamic region, we choose a static adjacent background at front-end of the ship. Then we adopt a variable weight method to update background in order to prepare for the next image proceeding background difference calculation. As the loop goes on, we can make fast and accurate ship image identification from the surveillance streaming media.

show abstract

Dust Monitoring System Based on Video Image Processing

Mei

Ding

2012

Electrical, Information Engineering and Mechatronics 2011

View full text Add to dashboard Cite

A background separation method of nonuniform image segmentation

Cited by 6 publications

References 4 publications

School of Teacher, Nanjing Xiao Zhuang University, Nanjing, 210000, China

School of Teacher, Nanjing Xiao Zhuang University, Nanjing, 210000, China

Inland Ship Recognition Algorithm Based on Neighboring Static Region Substitution for Background Updating

Dust Monitoring System Based on Video Image Processing

Contact Info

Product

Resources

About