WatershedSuperpixel资源-CSDN文库

共4个文件

mexa64：1个

pdf：1个

mexw64：1个

Superpixel,

Watershed

4星 · 超过85%的资源需积分: 46 174 浏览量 2015-10-20 21:02:37 上传评论 5 收藏 2.35MB ZIP 举报

资源推荐

资源详情

资源评论

收起资源包目录

Watershed-Superpixe_zwhoo_ICIP2015.zip （4个子文件）

Watershed-Superpixe_zwhoo_ICIP2015.pdf 2.32MB

SCoW.mexw64 61KB

SCoW.mexa64 22KB

demo.m 488B

WATERSHED SUPERPIXEL

Zhongwen Hu

Qin Zou

Qingquan Li

?,1

Shenzhen Key Laboratory of Spatial Smart Sensing and Service,

Shenzhen University, P.R. China

School of Computer Science, Wuhan University, P.R. China

{zwhoo@szu.edu.cn, qzou@whu.edu.cn, liqq@szu.edu.cn}

ABSTRACT

As a pre-processing tool, superpixel algorithms have been

popular used in many computer-vision applications. High

efﬁciency is a desired property of superpixel algorithms, e-

specially in real-time vision systems. In this paper, a novel

high-efﬁcient superpixel algorithm is developed based on the

watershed algorithm, namely the spatial-constrained water-

shed (SCoW). SCoW performs watersheding in a marker-

controlled manner, with a set of evenly placed markers.

To align superpixel boundaries to image edges, an edge-

preserving scheme is embedded into the SCoW which makes

a balance between the homogeneity and the compactness.

Without any complex computing, the proposed superpixel

algorithm is found to produce high quality superpixels as

traditional superpixel algorithms, while holding much higher

efﬁciency.

Index Terms— superpixel, watershed, spatial constraint,

image segmentation

1. INTRODUCTION

Superpixel segmentation, which over-segments an image

into a number of superpixel regions, is a widely used pre-

processing step for image- and vision-based systems. A

superpixel region is commonly deﬁned as a perceptually uni-

form and homogenous region in the image [1, 2]. With the

development of computer vision and multimedia techniques,

superpixel algorithms have becoming increasingly popular

for many applications, e.g., object detection and tracking [3],

image segmentation and modeling [4, 5, 6], salience detec-

tion [7, 8, 9, 10] and image and object classiﬁcation [11], etc.

This study is jointly supported by grants from Postdoctoral Science

Foundation of China under grants No. 2014M562206 and2012M521472,

Geographic National Condition Monitoring Engineering Research Cen-

ter of Sichuan Province under grant No. GC201513, National Natural

Science Foundation of China under grant No.61301277, Shenzhen Sci-

entiﬁc Research and Development Funding Program under grants No.

ZDSY20121019111146499 and No. JSGG20121026111056204, Shenzhen

Dedicated Funding of Strategic Emerging Industry Development Program

under grant No. JCYJ20121019111128765 and National Basic Research

Program of China (973 Program) under grant No. 2012CB725303. *Cor-

responding author: Dr. Q.Q. Li.

Fig. 1. Watershed Superpixels. (a) Traditional watershed, (b)

watershed with uniform markers but without spatial constrain,

One main advantage of using superpixel is to improve the

computational efﬁciency in the corresponding systems. With

superpixels, the later-stage processing would be speedup s-

ince the number of image primitives has been dramatically

reduced as compared to the original pixel representations.

Another advantage is that, the superpixels provide to investi-

gate the image in a larger scale, which can give better spatial

support to compute regional features.

In the past two decades, important efforts have been

made in the ﬁeld of image segmentation, e.g., watershed

segmentation [12], Normalized Cut [13], graph-based seg-

mentation [14], statistic region merging [15], etc. Generally,

image segmentation is used as a pre-processing step for vision

computing applications. However, with increasing demand

of high-performance pre-processing, the above segmentation

methods gradually show their limitations, and superpixel al-

gorithms are found to be more suitable as a pre-processing

tool. Exactly, a set of famous superpixel algorithms have

been proposed in the past decade [16, 17, 1], In [16], a

method named Turbopixels introduces spatial constraint to

level-set evolution to get dense over-segmented and compact

superpixels. While in SLIC [1], spatial constraint is embed-

ded into a k-mean clustering method to achieve superpixel

segmentation. These two methods can produce superpixels

effectively, however both requires a complex iterative process

to reﬁne the results. Some newly developed algorithms can

be found in [18, 19, 20]. In [19], superpixels are gained

by a structure sensitive over-segmentation technique which

exploits Lloyd’s k-means algorithm with a geodesic distance.

In [18], superpixels are produced in a energy-driven way,

where a hill-climbing algorithm is used to continuously re-

ﬁne the superpixels by modifying the boundaries. In [20],

superpixel segmentation is achieved in a lazy random walk

framework, which iteratively optimized the energy function.

Most of the above superpixel algorithms are computational

expensive, e.g., it may take a few seconds even minutes to

segment an image of a moderate size, e.g., 481×321 pixel-

s. Since the superpixel segmentation serves as a basis for

later-stage processing, higher efﬁcient algorithms are always

desired.

Note that, the watershed transform can be high-efﬁciently

performed. In this paper, we propose to develop a new su-

perpixel algorithm based on the watershed. However, without

spatial constraint, it produces superpixels of irregular shapes

and sizes, as shown in Fig. 1(a) and (b). Therefore, we present

to use spatial constrain to the watershed, which leads to a

spatial-constrained watershed superpixel algorithm (SCoW).

The proposed algorithm introduces spatial constraints and u-

niform markers to obtain compact and evenly distributed su-

perpixels, as shown in Fig. 1(c), while offering a signiﬁcant

boost in the speed. Moreover, an edge-preserving scheme is

used to align superpixel boundaries to the image edges.

The remainder of this paper is organized as follows. Sec-

tion 2 describes details of the proposed spatial-constrained

watershed algorithm. Section 3 reports the experiment result-

s, followed by a brief conclusion in Section 4.

2. SPATIAL-CONSTRAINED WATERSHED

2.1. Marker-controlled Watershed

The basic idea of marker-controlled watershed transform

is that ﬂooding the topographic surface from a previously

deﬁned set of markers. The gradient image is viewed as to-

pographic surface, thus, pixels with low gradient values will

be ﬂooded with high priority. Meyer proposed the marker-

controlled watershed by utilizing a set of ordered queues

(Fig.2) with different priorities [12]. In this method, mark-

er points are ﬁrstly put into the ordered queue, depending

on their priority; and then pixels with highest priorities are

pulled out and labeled; the unlabeled neighbors of this pixels

are then pushed into the queue. When the ordered queue is

empty, each pixel of the image is labeled. A watershed trans-

form is usually applied on gradient image, and the priority

inverse to the gradient magnitude. The details of implemen-

tation of this method can be seen in [12].

The pixels are labeled according to their priorities, which

is deﬁned as follows:

p(x, y) = g

max

− g(x, y) (1)

where p(x, y) denotes the priority of pixel located at (x, y) in

the image and g(x, y) is the gradient magnitude value; g

max

Fig. 2. An illustration of ordered queue with different priori-

ties.

stands for the maximum possible gradient magnitude in the

image; its value is equal to the number of the ordered queues.

The marker-controlled watershed starts with selecting a

set of markers. To make the superpixels have uniform sizes

and even distribution, the proposed method starts with placing

evenly distributed marker points in the image with a step w ×

w, where w is the grid interval. To avoid the markers located

on the image edges, the marker points are perturbed in an 3×3

neighborhood to the lowest gradient positions.

2.2. Spatial Constraint and Edge Preserving

Meyer’s method [12] processes pixels only depending on their

gradient values, which often leads to irregular shapes and

sizes. Thus, spatial constraint is introduced in this paper to

force the superpixels compact. The basic idea of spatial con-

straint is that the priority should not only depend on its gradi-

ent value, but also on the spatial pattern. The basic assump-

tion is that under similar conditions, the pixels closer to their

corresponding markers should be processed with higher pri-

orities. On the other hand, when the distances from the mark-

ers are the same, pixels with lower gradient values should be

given higher priorities. Then we deﬁne the pixel priority as

follows:

(x, y) = p

(x, y) + λp

(x, y) (2)

where p

(x, y) denotes pixel priority derived from the gra-

dient; and p

(x, y) denotes the pixel priority derived from

spatial constraint; λ is the user input parameter that balancing

two measurements.

The spatial constraint ensures that pixels near mark-

er points have higher priorities. Moreover, the superpixel

boundaries should align to the image edges. Thus, the edge-

preserving scheme should be considered to eliminate the

over-smoothed effect due to strong spatial constraint near

the edges. The pixel priority derived from spatial constraint

considering edge-preserving scheme is deﬁned as :

(x, y) = −d

x,y

· E (x, y) (3)

where d

x,y

denotes the Euclidean distance from pixel (x, y)

to its corresponding marker point; E(x, y) is the edge-

preserving function, which balances the spatial constraint

评论收藏

内容反馈

HEAD_L4

2019-08-30

可惜源代码封装处理了。

Guzenyel
上传者
2019-09-18

QQ群里有源代码的
laq79

2019-04-19

看不到原始码
A1036857413

2018-07-31

楼主都说不能运行，我先试试
川2017

2017-04-26

遗憾的是看不到源程序
gaolip77

2017-01-09

没有原始代码啊

前往

页

Guzenyel

粉丝: 149
资源: 2

Watershed Superpixel

WaterShed分水岭

watershed matlab code

watershed分水岭算法的源代码

超像素分割matlab代码

matlab superpixel图像解析

C++实现分水岭算法（Watershed Algorithm）

SCoW-Watershed Superpixel.zip_Scow_superpixel_watershed_图像分割算法

Linear Spectral Clustering Superpixel:Linear Spectral Clustering Superpixel-matlab开发

Texture Relative Superpixel Generation With Adaptive Parameters

rWeb:Watershed 的 Web Kiosk 的可定制版本 - rWeb

watershed-分水岭图像分割

Watershedalgorithm.zip_watershed algorithm_watershed c#_分水岭分割_图像

3_matlab_brain_watershed_

watershed算法

Watershed Delineation Tools流域划分插件

An Iterative Spanning Forest Framework for Superpixel Segmentation

Watershed-algorithm.zip_图形图像处理_matlab_

Watershed_matlab_分水岭分割_源码

SLIC超像素-MATLAB-Superpixel.zip

Watershed-Analysis:分水岭分析软件

watershed matlab 医学图像处理

watershed分水岭算法

watershed.rar_watershed matlab_分水岭_分水岭 matlab_分水岭matlab_分水岭分割

图像分水岭分割源码-euclidean_watershed_Segment.m

watershed分水岭图像分割方法C++代码_亲测可用

Watershed_Fun.zip_matlab例程_matlab_

Watershed分水岭分割算法

TurboPixels: Superpixel

最新资源