基于GN分裂的小目标检测区域推荐搜索算法 下载: 952次
An Algorithm of Small Object Detection Region Proposal Search Based on GN Splitting
图 & 表
图 1. 连通子图示意图
Fig. 1. Schematic of connected graph
下载图片 查看原文
图 2. Quick Shift算法分割示意图。(a)原图;(b)分割结果图
Fig. 2. Segment by Quick Shift algorithm. (a) Original image; (b)segmented result
下载图片 查看原文
图 3. 部分实验结果(Sk=3.5, dmax=13.5)。(a)飞机;(b)绵羊;(c)汽车;(d)轮船
Fig. 3. Part of experimental results with Sk=3.5, dmax=13.5. (a) Airplane; (b) sheep; (c) car; (d) buoy
下载图片 查看原文
图 4. 时间消耗与分割区域数量的关系
Fig. 4. Relationship between time consumption and number of segment regions
下载图片 查看原文
表 1算法流程
Table1. Algorithm flow
Algorithm 1: GN_RP |
---|
Input: (color) image | Output: Set of small object location boxes B | Obtain initial regions R={ri}using Quick Shift | Calculate histograms from different color space in ri | Foreach Neighbouring region pair (ri.rj) do | Calculate similarity S(i,j) as the weight wij | Generate graph G by neighbor node pair and similarity | Sort wij in an ascending order | While wij<wthr do | Delete eij from G | Extract set of connected subgraph G'from Gcut | Delete the maximum connected subgraph in G' | Extract object location boxes B from G' |
|
查看原文
表 2分割算法参数对结果的影响
Table2. Effects of segmentation parameters on the results
Types of different parameters | Recall/% | Number of proposals |
---|
Sk=5.0, dmax=20.0 | 70.04 | 12.06 | Sk=3.5, dmax=20.0 | 71.60 | 22.17 | Sk=5.0, dmax=13.5 | 81.26 | 23.38 | Sk=3.5, dmax=13.5 | 84.09 | 41.14 | Sk=2.5, dmax=13.5 | 78.44 | 50.49 |
|
查看原文
表 3不同算法的结果对比
Table3. Comparison of results using different algorithms
Types of different search | Recall/% | Time/s | Number of proposals |
---|
Exhaustion search (9900) | 80.55 | 0.004 | 9900 | SS(Sscale=80, Smin=50) | 61.84 | 0.113 | 111.86 | SS(Sscale=30, Smin=10) | 82.89 | 0.565 | 700.82 | GN_RP(Sk=3.5, dmax=13.5) | 84.09 | 0.068 | 41.14 |
|
查看原文
赵沛然, 吴新元, 汤新雨, 沈晓海, 许海燕, 李敏, 张学武. 基于GN分裂的小目标检测区域推荐搜索算法[J]. 光学学报, 2018, 38(9): 0915005. Peiran Zhao, Xinyuan Wu, Xinyu Tang, Xiaohai Shen, Haiyan Xu, Min Li, Xuewu Zhang. An Algorithm of Small Object Detection Region Proposal Search Based on GN Splitting[J]. Acta Optica Sinica, 2018, 38(9): 0915005.