CN101900687A - Method for monitoring and early warning water bloom in small water area based on image processing - Google Patents

Method for monitoring and early warning water bloom in small water area based on image processing Download PDF

Info

Publication number
CN101900687A
CN101900687A CN 201010218511 CN201010218511A CN101900687A CN 101900687 A CN101900687 A CN 101900687A CN 201010218511 CN201010218511 CN 201010218511 CN 201010218511 A CN201010218511 A CN 201010218511A CN 101900687 A CN101900687 A CN 101900687A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
water
monitoring
bloom
early
warning
Prior art date
Application number
CN 201010218511
Other languages
Chinese (zh)
Other versions
CN101900687B (en )
Inventor
周伟
王楷
石为人
苏士娟
范敏
贾承晖
陈舒涵
陈露
雷璐宁
Original Assignee
重庆大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention provides a method for monitoring and early warning water bloom in a small water area based on image processing, which relates to a method for monitoring and early warning water bloom in a small water area. In the invention, a camera and a computer are used for realizing the purposes of monitoring and early warning the water bloom in the small water area through calculation according to programs. The invention has the characteristics of simple monitoring equipment, strong environment adaptive capacity, small algorithmic calculated amount, high speed, low requirements for hardware, flexibility, convenience, low cost, convenient maintenance and the like, can be directly embedded with the existing water surface video monitoring system or can be developed into embedded equipment or can directly operate on the computer, and can realize the purpose of quickly and effectively early warning the water bloom in real time. The invention can be widely used for monitoring and early warning the water bloom in small water areas such as upstream reservoirs of rivers, reservoirs used as sources of drinking water of towns, river reaches in water taking areas of towns, rivers, lakes or landscape water areas where the water bloom often occurs, and the like.

Description

一种基于图像处理的小型水域水华监测预警方法 Small blooms monitoring and early warning method based on image processing waters

技术领域 FIELD

[0001] 本发明属于水环境监测技术领域,具体涉及小型水域(即江、河上游水库、城镇饮用水源的水库、城镇取水区河段、常发生水华的江、河、湖泊区及景观水域等)的水华监测 [0001] The present invention belongs to the technical field of water environment monitoring, particularly relates to a small water (i.e. River, river upstream of the reservoir, the reservoir town drinking water, urban river water areas, rivers, lakes and landscape areas often occurs bloom waters, etc.) blooms monitoring

预警方法。 Early warning methods.

背景技术 Background technique

[0002] 近年来,随着工业、科技的飞速发展,大量的含有大量氮、磷的工、农业及生活废弃物排入水中,致使江、河、湖泊等水域的淡水出现富营养化而暴发水华(如绿藻水华、蓝藻水华等),我国的太湖、滇池、巢湖、洪泽湖都发生过水华,水华造成是最大危害是:①藻类毒素通过食物链间接影响人类的健康或藻类的产生的致癌物质直接威胁人类的健康和生存,污染水源;②自来水厂的过滤装置被藻类“水华”填塞,影响取水;③漂浮在水面上的“水华”影响景观,而且有难闻的臭味,污染环境等。 Freshwater [0002] In recent years, with the rapid development of industry, science and technology, a large number of contains large amounts of nitrogen, phosphorus industrial, agricultural and domestic waste into the water, resulting in the waters of rivers, lakes and outbreaks of eutrophication bloom (such as green algae blooms, blue-green algae blooms, etc.), China's Taihu, Dianchi and Chaohu, Hongze have occurred blooms, blooms caused the greatest harm is: ① algae toxins through the food chain indirectly affect human health or produced by algae carcinogen direct threat to human health and survival, pollute the water; ② water plant filtration devices are algae "blooms" tamponade, influence water; ③ floating on the water "blooms" affect the landscape, but also unpleasant odor, pollution of the environment. 可见水华现象频繁出现,不仅造成大面积水环境的破坏,而且也带来了巨大的经济损失。 Visible algal bloom phenomenon occurs frequently, not only caused damage to a large area of ​​water environment, but also brought huge economic losses. 因此,在江、河上游的源头水域、地方湖泊以及特定功能水域(城镇取水区、饮用水源和景观水域等)等淡水源头进行监测,对该水域水华的发生、变化及灾情趋势进行监测预警是非常有意义的。 Therefore, monitoring of source waters in the upper reaches of rivers, lakes and local specific features water (town water district, drinking water sources and the water landscape, etc.) and other freshwater sources, to monitor the occurrence of changes and trends in the disaster bloom waters early warning is very meaningful.

[0003] 现有水华监测预警方法,如2008年8月6日公开的,公开号为CN101236519A的“用于蓝藻监测及蓝藻水华预警的浮标”专利,公开的浮标由浮标载体、仪器舱、监测传感器集合体(由五种不同的传感器集合而成)、通讯天线、太阳能电池板构成。 [0003] bloom existing monitoring and early warning methods, such as the 2008 August 6 disclosure, a number of "for the blue-green algae and cyanobacteria bloom monitoring early warning buoys" patent CN101236519A, published by the buoy buoys carrier, equipment bay , monitoring sensors aggregate (a collection of five different sensors together), the communication antenna, solar panel configuration. 应用时,将一定数量的浮标布置在被测水域的特定位置,对水体的不同深度进行监测,并将监测的数据通过天线传输到监测中心,进行数据分析后实时预警。 Application, a certain number of buoys are arranged in a specific location of the test waters, the water at different depths are monitored and the monitored data is transmitted to the monitoring center via the antenna, real-time data analysis after warning. 该专利的主要缺点是: The main disadvantage of this patent is:

[0004] 1.监测的局限性大。 [0004] 1. Monitoring large limitations. 该专利只将一定数量的浮标布置在被测水域的特定位置,对水体中的不同深度进行监测(即水下监测),不能对较大面积的水域进行监测,也不能对水面污染物的扩散规律、分布范围、污染程度等进行监测;浮标在监测的时候需要与水体直接接触,而且占据一定的水域空间,不适合在一些特定功能的水域(如饮用水源、景观水域等)直接应用。 This patent only a certain number of buoys are arranged in a specific location of the test water for different water depths of monitoring (i.e. underwater monitoring), water can not monitor a large area, not the diffusion of water pollutant law, distribution, pollution degree monitoring; float require direct contact at the time of monitoring the water, and occupy a certain water space, is not suitable in the waters of some specific functions (such as drinking water, landscape water, etc.) directly. 因此,浮标监测的局限性较大。 Therefore, the limitations of the larger buoy monitoring.

[0005] 2.维护费用高,监测成本高。 High [0005] 2. maintenance costs, high monitoring costs. 该专利在实际应用中,需要专门的技术人员定期到监测现场对浮标进行维护,耗时耗人力,从而较大地增加了维护成本。 The patent in practical applications, the need for specialized technical staff on a regular basis to monitor the site for buoy maintenance, takes a lot of manpower, thus greatly increasing maintenance costs. 为了获取准确的水质信息,就必须用昂贵的高精度监测传感器,并且该专利的传感器量大,种类多,从而提高了监测成本。 In order to obtain accurate water quality information, it is necessary to use expensive high-precision monitoring sensor, and the sensor of the patent is large, many kinds, thereby increasing the cost of monitoring.

[0006] 3.普及性差。 [0006] 3. The popularity difference. 由于该专利的成本高,结构复杂,使用、操作和维护的专业性较强强, 不利于在管理部门、企业和科研单位普及使用。 Due to the high cost of the patent, complex structure, use, operation and maintenance of highly specialized and strong, is not conducive to the management departments, enterprises and research institutes universal use.

发明内容 SUMMARY

[0007] 本发明的目的是针对现有水华监测预警方法的不足,提供一种基于图像处理的小型水域水华监测预警方法,具有成本低,对环境适用能力强,能实时获取水面污染信息并能自动识别水华进行报警,便于推广应用等特点。 [0007] The object of the present invention is less than the existing method of monitoring and warning bloom, there is provided a method of monitoring and warning small water bloom water based on image processing, low cost, high capacity suitable for the environment, the surface contamination can obtain information in real time bloom and can automatically identify the alarm, and the like to facilitate application characteristics.

5[0008] 本发明的机理:本发明首先以被测水域的历史水华区域图像为样本图建立颜色先验模型,即将样本图的每个象素从三原色(RGB)空间转换到色、亮分离(HSV)空间,为了在各种光照情况下都能锁定水华区域,因此摒弃HSV色彩空间中的饱和度(S)和亮度(V)分量,只采用色调(H)分量。 5 [0008] The mechanism of the present invention: In the present invention first history image region blooms waters tested sample prior model established color chart, each pixel sample coming from the three primary colors of FIG conversion (RGB) color space to bright separation (HSV) space and can be locked to bloom in areas where a variety of lighting, thus abandon the saturation (S) HSV color space and brightness (V) component, using only the hue (H) component. HSV模型是最贴近人类对色彩的感知方式,而颜色信息对该空间的H(色调)分量比较敏感,且H(色调)分量很大程度地去除了光照的影响,因此建立的颜色先验模型不受环境光线变化的影响。 HSV model is the most close to the human perception of a color, and the color space information of the H (hue) component is more sensitive, and H (hue) component in addition to a large extent to influence of light, the color of prior model thus established It is not affected by changes in light conditions.

[0009] 先建立水华区域的颜色先验模型,设{x' jpu.i为历史样本图的水华区域的n 个像素,在颜色模型中,当H(色调)分量的离散化取值为i(i = 1,2..., 360)的色调概率为: [0009] Prior to the establishment of model bloom color region, n is set pixels {x 'jpu.i history of the sample of FIG bloom area, in the color model, when the discrete values ​​H (hue) component It is (i = 1,2 ..., 360) i is the probability of hue:

[0010] [0010]

[0011] 式(1)中:p是颜色模型,Pi是色调值为i的概率,5为5 (Kronecker delta)函数,函数b(x'」)为空间R2— {1,2…m}的索弓丨,即位于位置x'」的像素向直方图量化特征空间的索引。 [0011] Formula (1) where: p is the color model, color tone Pi is the probability value of i, 5 5 (Kronecker delta) function, the function b (x '') is a space R2- {1,2 ... m} cABLE bow Shu, i.e., at a position x '' pixel quantization index to the feature space histogram.

[0012] 再用连续自适应性均值漂移(Camshift)算法来检测水华色块,连续自适应性均值漂移算法是由均值漂移(MeanShift)算法演变而来,均值漂移(MeanShift)算法的过程就是通过核G上的采样均值平移矢量Mm(x)(即核K上的密度梯度的估计)更新核G中心的一个递归过程。 [0012] and then continuously adaptive mean shift (Camshift) algorithm to detect bloom color, continuous adaptive mean shift algorithm was developed by the mean shift (MeanShift) algorithm evolved, mean shift (MeanShift) algorithm is the process by sampling the mean shift vector the nucleus G Mm (x) (i.e., the density gradient is estimated on the core K) updating a recursive kernel of the process G center. 均值漂移算法应用在连续序列时,由核G的位置的初始中心开始不断地计算均值平移向量,迭代更新目标位置,直到收敛于最优匹配点。 Mean shift algorithm in a continuous sequence, the initial position of the center of the nucleus G began to calculate the mean of the translation vectors, iteratively updated target position, until convergence to the optimal matching point.

[0013] Mh,G(x)(即核K上的密度梯度的估计)计算公式如下: [0013] Mh, G (x) (i.e., the density gradient is estimated on the core K) is calculated as follows:

[0014] [0014]

[0015] 其中:h为带宽度,C为归一化常数,可见核G上的采样均值平移矢量为核K上的密度梯度的估计。 [0015] where: h is the width, C is a normalization constant, visible on the sample mean nucleus G translation vector estimated for the density gradient in the core K. }K(x)为核K上带宽度为h的多变量密度估计: } K (x) is the width of the core K of h multivariate density estimation:

[0016] [0016]

[0017] 集合{x'」}」=1,2.』是(1维欧氏空间炉的11个点,1^00表示该像素点的核函数: [0017] The set {x ' "}" = 1 "is (11 points dimensional Euclidean space of the furnace 1, 1 ^ 00 represents the kernel of pixels:

[0018] [0018]

[0019] 其中:cd为d维单位球体体积。 [0019] wherein: cd is the d-dimensional unit sphere volume.

[0020] 连续自适应性均值漂移算法就是在视频图像的处理区域内进行方向投影和均值漂移计算,以当前帧搜索窗口的质心和面积作为下一帧搜索窗口的初始值,再进行迭代以实现对目标的检测跟踪。 [0020] Continuous adaptive shift algorithm is performed to calculate the mean direction of the projection and the drift region in the video image processing, an area centroid of the current frame and the search window frame as an initial value for the next search window, and then iterating to achieve detection and tracking of targets.

[0021] 然后用改进的连续自适应性均值漂移算法过滤干扰色块,即在应用Camshift算法前,首先将概率分布图进行腐蚀、膨胀处理,去掉噪声块,然后检测目标色块的内外轮廓, 找到目标色块中的空洞进行填充。 [0021] and then a modified continuous adaptive mean shift filtered interference color, i.e. before application Camshift algorithm, the probability distribution for the first etching, the expansion process, block noise is removed, and then detecting internal and external contours of a target color, find the target color in the hole to fill. 然而进行以上预处理后,可能仍然存有干扰色块,所以在应用于水华检测的时候有必要改进一下Camshift算法的适应性。 However, after the above pretreatment, there may still be interference color, so when applied to bloom detect when it is necessary to improve the adaptability Camshift algorithm.

[0022] 从水面出现藻类聚集到水华全面爆发,通常需要15天左右的时间,所以在较短时间内,水面上出现绿色区域变化是缓慢的,相比而言,相邻两帧图像时间间隔是非常短的(几十毫秒),那么目标色块在帧间的面积变化是非常小的,基于这种变化规律,我们就能通过预测特征面积的方法来过虑干扰色块。 [0022] algae blooms gathered to full-blown, usually takes about 15 days, so in a relatively short period of time, change is slow green area appears on the surface, compared with the adjacent two images emerge from the water time interval is very short (tens of milliseconds), then the target color in the area of ​​change between frames is very small, based on this variation, we can be misplaced by the interference color method for predicting the characteristics of the area.

[0023] 设当前帧图像中色块的面积为Sn,下一帧的预测面积为S' n+1,那么 [0023] Set the current frame image in the color area of ​​Sn, a next frame is predicted area S 'n + 1, then

[0024] S' n+1 = Sn+A (5) [0024] S 'n + 1 = Sn + A (5)

[0025] 其中,常数A是一个经验值(根据具体情况设置),表示帧间目标色块面积的增量。 [0025] wherein A is an empirical constant value (for specifics), it represents the incremental target inter-patch area. 根据式(5)就能预测下一帧图像中目标色块的面积,如果满足 The formula (5) in the next frame can be predicted target color area, if satisfied

[0026] Sn+1-S' n+1| ^ I (6) [0026] Sn + 1-S 'n + 1 | ^ I (6)

[0027] Sn+1表示下一帧色块的实际测量面积,\为允许的误差范围,说明当前检测到的色块符合帧间色块面积的变化规律;如果|Sn+1_S' n+1|超出允许的误差范围内时,则认为下一帧实际检测到的色块面积不符合帧间水华区域面积的变化规律,将多余部分定义为干扰色块并将其滤掉。 [0027] Sn + 1 of the next frame indicates actual measurement patch area, \ allowable error range is described in line with the currently detected color variation inter patch area; if | Sn + 1_S 'n + 1 | within exceeds the allowable error range, then the next frame that is actually detected variation of color area does not meet the area of ​​the inter-bloom, the excess portion is defined as an interference color and was filtered off.

[0028] 实现本发明目的的技术方案是:一种基于图像处理的小型水域水华监测预警方法,利用摄像头和计算机,通过程序,依据历史数据建立先验颜色模型,运用改进的连续自适应性均值漂移(Camshift)算法过滤掉干扰色块并检测出水华区域,通过该区域面积的变化分析灾情趋势,并建立类似于气象灾害的预警模型,来监测并预警水华。 [0028] The purpose of the present invention, the technical solution is: A method for monitoring and warning small water bloom water based on image processing by the camera and the computer, through the program, establishing a priori model based on historical data of the color, the use of improved continuous adaptability mean shift (Camshift) algorithm to filter out interference detection and water color China region, by a change in trend analysis of the disaster area in the region, and the establishment of early warning model is similar meteorological disasters monitoring and early warning bloom. 具体步骤如下: Specific steps are as follows:

[0029] (1)建立颜色先验模型 [0029] (1) establishing a color prior model

[0030] 首先根据被测水域的历史水华区域的样本图或视频数据,建立颜色先验模型, 即: [0030] First, based on a sample history of blooms FIG area measured waters or video data, a priori model of color, namely:

[0031] 设{x'」}」=。 [0031] The set {x ''} '=. 』*样本图或视频数据的n个像素,将每个象素从RGB(三原色) 空间转换到HSV(色、亮分离)空间,将H(色调)分量的离散化取值为i (i = 1,2. . .,360), 把H(色调)分量的范围[0,360]缩放到W,255],以便取值范围的值能用一个字节(byte) 来表示。 n pixels' * of FIG sample or video data from each pixel RGB (three primary colors) to the HSV space conversion (color light separation) space, the discrete values ​​of H (hue) component is i (i = 1, 2,..., 360), the range of H (hue) component [0,360] scaled to W, 255], so that the value can be in the range of one byte (byte) is represented. 通过计算机按下式计算H分量的色调概率: H tone probability calculated as computer components:

[0032] [0032]

[0033] 其中:p是颜色模型,Pi是色调值为i的概率,5为5 (Kronecker delta)函数, 函数b(x'」)为空间R2— {1,2…m}的索弓丨,即位于位置x' j的像素向直方图量化特征空间的索引。 [0033] wherein: p is the color model, color tone Pi is the probability value of i, 5 5 (Kronecker delta) function, the function b (x '') is a space R2- {1,2 ... m} CABLE bow Shu , i.e., at a position x 'j the index space of the histogram of the pixel quantization characteristics.

[0034] (2)对水华区域进行检测 [0034] (2) the detection region to bloom

[0035] ①对小型水域进行监测 [0035] ① for small water monitoring

[0036] 第(1)步完成后,在小型水域(即江、河的上游水库或城镇饮用水源的水库或城镇取水区的河段或常发生水华的江、河、湖泊水域或景观水域等)的岸边架设摄像头,并通过视频传输线与计算机连接。 [0036] (1) after completion of step, (i.e., river, reservoir or river water urban area or town drinking water reservoir upstream of the river or water bloom occurs frequently in a small water rivers, lake water or landscape shore waters, etc.) of the camera set up and connected with the computer through the video transmission line. 摄像头架设的具体位置、数量及高度,根据被测水域的具体情况确定。 Camera set up a specific position, height and number, determined according to the specific circumstances of the test waters. 用以监测被测水域的水面情况(因水域的水华暴发最显著的变化是水面视觉特征的变化)并摄取水面的视频图像,达到实时监测水面图像的目的。 Monitoring the video image to the measured surface waters (algal bloom due to the most significant change is a change in surface waters visual features) and the water uptake, the purpose of real-time monitoring of the image surface.

[0037] ②反向投影处理 [0037] ② backprojection process

[0038] 第(2)_①步完成后,对第(2)_①步输入的每一帧视频图像进行反向投影处理(BackProjection),即对视频图像处理区域中的每一个像素,通过计算机查询该像素与第(1)步建立的颜色先验模型(即H分量色调概率模型)的匹配程度,就得到该像素为目标像素的概率(此区域之外的其他区域的概率为0,图像中每一个像素的值就变成了目标颜色信息出现在此处的可能性的一种离散化度量,此处出现的可能性越大,像素的值就越大,反之则越小)。 [0038] of (2) _① After completion of steps, the first (2) _① each frame of video image input step of reverse projection process (BackProjection), i.e., the video image processing area in each pixel, the query by the computer the pixel (1) color prior model created in step (i.e., hue H component probability model) degree of matching, the probability that the pixel is obtained as the target pixel (the probability of other regions outside this area is 0, the image value of each pixel becomes a target color metric discrete message here possibility, the greater the possibility of here, the greater the value of the pixel, whereas the smaller). 经过上述处理,就得到每帧图像的目标颜色反向投影图。 Through the above process, the target color is obtained in each image backprojection FIG.

[0039] ③检测分割水华色块并滤掉干扰色块 [0039] ③ detection and segmentation bloom filter out interference color patches

[0040] 第(2)-②步完成后,进行检测水华色块和滤掉干扰色块处理,即: [0040] section (2) -② After completion of the step, and detecting color bloom filter out interference color process, namely:

[0041] 先对第(2)_②步得到的反向投影图进行腐蚀、膨胀,检测目标色块的内外轮廓及找到目标色块中的空洞进行填充等预处理,这样就能去掉一部分干扰色块,在一定程度上避免噪声干扰。 [0041] First the first (2) _② FIG backprojection step was subjected to erosion, dilation, the target patch in the hole and outside contour detection target fill color and found pretreatment, so that interference colors can be partly removed block avoiding noise to some extent.

[0042] 再检测分割水华色块区域,即通过计算机按下式不断地对经过上述预处理之后的反向投影图计算均值平移向量: [0042] redetection bloom color region segmentation, i.e., the following equation is calculated continuously the mean shift vector through said reverse projection image after preprocessing by the computer:

[0044] 其中:h为带宽度,C为归一化常数,>Jx)为核K上带宽度为h的多变量密度估计: [0044] where: h is the width, C is a normalization constant,> Jx) width of the core K of h multivariate density estimation:

[0046] 集合{x' ^卜^'是」维欧氏空间炉的11个点,k(x)表示该像素点的核函数: [0046] The set {x '^ ^ Bu' is 11 points "dimensional Euclidean space of the furnace, k (x) denotes the pixel of the kernel function:

[0048] 其中:(^为d维单位球体体积。用以迭代更新目标色块的位置,直到收敛于的最优匹配点,该匹配点的区域就是当前帧图像中的目标色块(即水华色块区域)。 [0048] wherein: (^ for the d-dimensional unit sphere volume for the target location update patches iteration, until convergence to the optimum matching point, the area of ​​the patch is the goal of matching points in the current frame image (i.e., water Hua color area).

[0049] 进行判断:当没有检测到目标色块时,则认为水面未出现水华,返回第(2)_①步获取下一帧继续进行检测;当有检测到目标色块时,则以当前帧最优匹配点的质心和面积作为下一帧搜索窗口的初始值,继续进行迭代计算找到最优匹配点。 [0049] The judgment: when the target color is not detected, then the surface does not appear that water blooms, return to the first (2) _① step acquires the next frame is detected to continue; if there is a target patch is detected, places the current centroid and area of ​​optimal matching point of the frame as the initial value of the next frame search window, continued iterative calculation to find the optimal match point. 如此循环迭代至程序结束为止,就检测分割出每帧图像的目标色块。 Such loop iteration until end of the program, segment object is detected in each image patch.

[0050] 然后过滤干扰色块,即计算上步得到的当前帧和下一帧的目标色块的面积,根据水华区域面积的帧间变化规律,利用当前帧的目标色块面积来预测下一帧图像中的目标色块面积,即计算下一帧色块实际检测面积和下一帧预测面积之间的差值(即多余部分),根据下式进行判断: [0050] was then filtered interference color, i.e., calculation of the target color on the current frame obtained in step and the next frame in the area, according to the variation of the area of ​​inter-bloom, to predict the current frame using the target area of ​​the patch a target color image area, i.e., calculating the difference (i.e., the excess portion) between the detected actual color area of ​​the next frame and the next frame prediction area, is determined according to the following equation:

[0051] Sn+1-S' n+1| ^ I (5) [0051] Sn + 1-S 'n + 1 | ^ I (5)

8[0052] 其中:Sn+1为下一帧色块的实际测量面积,S' n+1为下一帧的预测面积,I为允许的误差范围。 8 [0052] where: Sn + 1 of the next frame the actual measurement patch area, S 'n + 1 for the next frame prediction area, I is the allowable error range.

[0053] 当|Sn+1_S' n+1| ( \时,则认为下一帧实际检测到的色块面积符合水华区域面积的帧间变化规律并将其作为检测结果;当|sn+1-s' n+1| > €,即差值超出允许的误差范围内时,则认为下一帧实际检测到的色块面积不符合帧间水华区域面积的变化规律,将多余部分定义为干扰色块并将其滤掉。 [0053] When | Sn + 1_S 'n + 1 | (\, then the next frame that is actually detected variation of the inter-color area in line with an area of ​​the blooms and as a detection result; when | sn + 1-s' n + 1 |> €, i.e., when the difference exceeds the allowable error range, then the next frame that is actually detected variation of color area does not meet the area of ​​the inter-bloom, the excess part of the definition and filter out interference color.

[0054] 经过上述的处理后,就检测分割出滤掉干扰色块后的水华色块,得到水华色块在图像中的像素面积。 [0054] After the above processing, it detects the divided color after bloom filter out the interference color, the pixel area to obtain blooms in the image patch. 然后对每帧图像中的水华色块区域进行标定并输出检测结果。 Then the patch to bloom in the region of each frame image calibration and outputs a detection result.

[0055] (3)水华预警 [0055] (3) blooms early warning

[0056] ①计算水华区域实际面积: [0056] ① calculating the actual area of ​​blooms region:

[0057] 第(2)步完成后,先计算第(2)步得到的水华色块在图像中的像素面积占视野像素总面积的百分比,乘以视野的实际总面积就能得到水华区域实际面积。 [0057] (2) After completion of steps, the first calculation (2) obtained in step bloom color vision percentage of the total area of ​​the pixel area in the pixels in the image, the actual total field of view can be obtained by multiplying the area of ​​bloom the actual area of ​​the region. 再通过摄像机标定,计算视野的实际总面积,即将已知面积的参照物置于摄像头下的水域位置,计算它在图像中的像素面积占视野像素总面积的比值,用该已知参照物的实际面积除以该比值就得到视野的实际总面积,并将得到的水华区域实际面积输出到计算机软件界面上,以便用户直观的了解水华区域的实际面积。 Through the camera calibration, the actual calculation of the total area of ​​the field of view, i.e. a reference of known area is placed under the camera position of the waters, which calculates the pixel area in the image field of view accounts for the ratio of the total area of ​​the pixel, the actual reference using the known this ratio is obtained by dividing the area of ​​the total area of ​​the actual field of view, and outputs the resulting actual area bloom area to the computer software interface for the user to intuitively know the actual area of ​​the bloom area.

[0058] ②建立预警模型: [0058] ② early warning model:

[0059] 第(3)_①步完成后,根据第(3)_①步得到的水华区域实际面积,先按6〜10小时为间隔,计算每个时间间隔内的水华区域实际面积的均值,再通过计算机分析该实际面积的均值变化规律建立一元线性回归预测模型,来预测下一个时间间隔的实际面积的均值, 并计算该实际面积的均值占水域总面积的百分比,然后以该百分比代表的水华灾害程度为依据,将水华暴发的预警等级划分为蓝色、黄色、橙色和红色四个等级,对应的预警区间分别是一般、较重、严重和特别严重四个等级,并进行判断:当代表的水华灾害程度的百分比小于预警值(预警值的大小根据被测水域的具体情况和应用需求确定,本发明取5% )时, 则返回到返回第(2)-①步获取下一帧继续进行检测;当代表的水华灾害程度的百分比大于预警值时,则根据划分的预警等级进行预警。 [0059] section (3) _① After completion of steps, in accordance with section (3) _① actual area obtained in step bloom area, press 6~10 hours intervals, calculate the mean actual area of ​​the bloom area within each time interval , then analyzed by computer mean variation of the actual area to establish a linear regression prediction model to predict the average of the actual area of ​​the next time interval, and calculate the mean of the actual area of ​​the percentage of the total area of ​​water, and then the percentages represent the bloom is based on the extent of the disaster, will bloom early warning level is divided into blue, yellow, orange and red four levels, corresponding warning range are generally heavier, serious and especially serious four grades, and Analyzing: Chinese when the percentage of water less than the degree of hazard represented by the warning value (warning value of magnitude depending on the circumstances and the application needs to determine the test waters, the present invention is taken 5%), then to the return section (2) -① step Get the next frame is detected to proceed; China when the percentage of water greater than the degree of hazard represented by the warning value, the early warning alarm according to the classification. 水华暴发预警等级的具体划分如下: They are divided bloom outbreak alert level is as follows:

[0060] 一般蓝色预警(即零星性水华):水华零星集聚,主要水域区藻类生物密度小于3000万个/L,水华面积大于等于水体总面积的5%。 [0060] Usually blue warning (i.e. sporadic water-Hua): gathering scattered bloom, the main areas of algae water density of less than 30 million / L, blooms area greater than or equal to 5% of the total water.

[0061] 较重黄色预警(即局部性水华):藻类在局部水域集聚,主要水域区藻类生物密度介于3000〜5000万个/L之间,水华面积大于等于水体总面积的10%。 [0061] The heavy yellow warning (i.e. blooms locality): water partial gathering of algae, algae density primary water zone between ten thousand 3000~5000 / L, blooms area greater than or equal to 10% of the total water .

[0062] 严重橙色预警(即区域性水华):当发生区域性水华,主要水域区藻类生物密度介于5000〜8000万个/L之间,水华面积大于等于水体总面积的40%。 [0062] severe orange alert (i.e. blooms regional): When the regional water bloom occurs, the main water algae density areas between 5000~8000 million / L, between bloom area greater than or equal to 40% of the total water .

[0063] 特别严重红色预警(即全面性水华):水华全面性暴发,主要水域区的藻类生物密度大于8000万个/L,水华面积水华面积大于等于水体总面积的60%。 [0063] particularly serious red alert (ie comprehensive blooms): bloom comprehensive outbreak, major water algae density region is greater than 80 million / L, the area blooms bloom area greater than or equal to 60% of the total area of ​​the water body.

[0064] 本发明采用上述技术方案后,主要有以下效果: [0064] The present invention adopts the above technical solution, the following main effects:

[0065] 1.具有良好的普及性。 [0065] 1. good popularity. 本发明提供了一种快速有效的水华监测预警方法,能自动识别水华区域并实时计算出该区域的实际面积,能按照预警等级进行预警以便通知相关部门实时了解水华灾害程度并采取相应的治理措施。 The present invention provides a fast and effective method for monitoring and warning of bloom, bloom can automatically identify the area in real time and calculates the actual area of ​​the region, can be performed according to the warning level warning to notify the relevant departments real-time understanding bloom disaster extent and take appropriate control measures. 在实际应用中,直接植入传统的水面视频监测系统中或开发成嵌入式设备都能实现对水华的监测预警,具有智能化、易操作等特点,便于在管理部门、企业和科研单位普及使用。 In practical applications, directly into the traditional video surveillance system in the water or to the development of embedded devices can achieve monitoring and early warning of blooms, with intelligent, easy to operate features such as ease of popularity in management departments, enterprises and research institutes use.

[0066] 2.成本低。 [0066] 2. Low Cost. 本发明采用的算法计算量小、速度快,对硬件要求较低。 Less calculation algorithm employed in the present invention, fast, low hardware requirements. 实际应用中, 无论直接在计算机上运行还是开发成嵌入式设备,能实时快速有效的实现水华监测预警, 很大程度上节约了监测成本,灵活方便。 Practical application, whether running or developed into an embedded device that can quickly and effectively to achieve real-time monitoring and early warning bloom, largely to save the cost of monitoring, flexible directly on the computer easy to use.

[0067] 3.对环境适应能力强。 [0067] 3. A strong ability to adapt to the environment. 本发明利用摄像头监测水面并很大程度上去除光照的影响,能有效避免了水面环境的复杂性和环境光线变化等带来的噪声干扰,从而能很好地适应环境变化,因此监测设备简单,成本低,便于维护,并能实时有效地进行监测。 The present invention utilizes camera monitoring water and remove the effects of lighting to a large extent, can effectively avoid the noise caused by ambient light changes and the complexity of the water environment, so as to well adapt to environmental changes, and therefore a simple monitoring device, low cost, easy to maintain, and can be monitored in real time effectively.

[0068] 本发明能直接植入传统的水面视频监测系统中或开发成嵌入式设备,广泛地应用于江、河上游水库、城镇饮用水源的水库、城镇取水区河段或常发生水华的江、河、湖泊区及景观水域等的小型水域的水华监测预警。 [0068] The present invention can be implanted directly into the conventional video surveillance system surface or embedded devices developed into widely used in rivers upstream of the reservoir, the reservoir of drinking water sources urban, urban area river water or water bloom often occurs bloom monitoring and early warning of small waters of rivers, lakes and other waters of the area and landscape.

附图说明 BRIEF DESCRIPTION

[0069] 图1为本发明方法的程序流程框图。 [0069] FIG 1 program flow diagram of the method of the present invention. 具体实施方式 detailed description

[0070] 下面结合具体实施方式,进一步说明本发明。 [0070] Next, with reference to specific embodiments, further illustrate the invention.

[0071] 实施例 [0071] Example

[0072] 如图1所示,一种基于图像处理的小型水域水华监测预警方法,对重庆市开县澎溪河(位于三峡库区腹心地带的长江一级支流)的水库进行水华监测预警,该段水域的总面积约6. 3万平方米,具体步骤如下: [0072] As shown, a water-bloom of small water warning method based on image processing, Reservoir Kaixian Pengxihe (located in the hinterland of Yangtze Three tributary Reservoir) is an early warning monitoring bloom the total area of ​​the section of water about 63,000 m2, the following steps:

[0073] (1)建立颜色先验模型 [0073] (1) establishing a color prior model

[0074] 首先根据被测水域的历史水华区域的样本图或视频数据,建立颜色先验模型, 即: [0074] First, based on a sample history of blooms FIG area measured waters or video data, a priori model of color, namely:

[0075] 设{x'」}」="...„*样本图或视频数据的n个像素,将每个象素从RGB(三原色) 空间转换到HSV(色、亮分离)空间,将H(色调)分量的离散化取值为i (i = 1,2. . .,360), 把H(色调)分量的范围[0,360]缩放到W,255],以便取值范围的值能用一个字节(byte) 来表示。 [0075] The set {x ' "}" = "..." * n pixels or samples of FIG video data, each pixel converted from RGB (three primary colors) to the HSV space (color light separation) space, the discrete values ​​of H (hue) component is i (i = 1,2..., 360), the range of H (hue) component [0,360] scaled to W, 255], so that the range of value can be a byte (byte) is represented. 通过计算机按下式计算H分量的色调概率: H tone probability calculated as computer components:

[0077] 其中:p是颜色模型,Pi是色调值为i的概率,5为5 (Kronecker delta)函数, 函数b(x'」)为空间R2— {1,2…m}的索弓丨,即位于位置x' j的像素向直方图量化特征空间的索引。 [0077] wherein: p is the color model, color tone Pi is the probability value of i, 5 5 (Kronecker delta) function, the function b (x '') is a space R2- {1,2 ... m} CABLE bow Shu , i.e., at a position x 'j the index space of the histogram of the pixel quantization characteristics.

[0078] (2)对水华区域进行检测 [0078] (2) the detection region to bloom

[0079] ①对小型水域进行监测 [0079] ① for small water monitoring

[0080] 第(1)步完成后,在重庆市开县澎溪河水库的小型水域岸边的水塔上架设一个摄像头,摄像头视野覆盖整个水域,并通过视频传输线与计算机连接。 [0080] (1) after completion of step, set up a camera in a compact tower shore waters Kaixian Pengxihe reservoir, the water covers the entire camera field of view, and is connected with the computer through the video transmission line. 摄像头架设的具体位置、数量及高度,根据被测水域的具体情况确定。 Camera set up a specific position, height and number, determined according to the specific circumstances of the test waters. 用以监测被测水域的水面情况(因水域的水华暴发最显著的变化是水面视觉特征的变化)并摄取水面的视频图像,达到实时监测水面图像的目的。 Monitoring the video image to the measured surface waters (algal bloom due to the most significant change is a change in surface waters visual features) and the water uptake, the purpose of real-time monitoring of the image surface.

[0081] ②反向投影处理 [0081] ② backprojection process

[0082] 第(2)_①步完成后,对第(2)_①步输入的每一帧视频图像进行反向投影处理(BackProjection),即对视频图像处理区域中的每一个像素,通过计算机查询该像素与第(1)步建立的颜色先验模型(即H分量色调概率模型)的匹配程度,就得到该像素为目标像素的概率(此区域之外的其他区域的概率为0,图像中每一个像素的值就变成了目标颜色信息出现在此处的可能性的一种离散化度量,此处出现的可能性越大,像素的值就越大,反之则越小)。 [0082] section (2) _① After completion of steps, the first (2) _① each frame of video image input step of reverse projection process (BackProjection), i.e., the video image processing area in each pixel, the query by the computer the pixel (1) color prior model created in step (i.e., hue H component probability model) degree of matching, the probability that the pixel is obtained as the target pixel (the probability of other regions outside this area is 0, the image value of each pixel becomes a target color metric discrete message here possibility, the greater the possibility of here, the greater the value of the pixel, whereas the smaller). 经过上述处理,就得到每帧图像的目标颜色反向投影图。 Through the above process, the target color is obtained in each image backprojection FIG.

[0083] ③检测分割水华色块并滤掉干扰色块 [0083] ③ detection and segmentation bloom filter out interference color patches

[0084] 第(2)-②步完成后,进行检测水华色块和滤掉干扰色块处理,即: [0084] section (2) -② After completion of the step, and detecting color bloom filter out interference color process, namely:

[0085] 先对第(2)_②步得到的反向投影图进行腐蚀、膨胀,检测目标色块的内外轮廓及找到目标色块中的空洞进行填充等预处理,这样就能去掉一部分干扰色块,在一定程度上避免噪声干扰。 [0085] First the first (2) _② FIG backprojection step was subjected to erosion, dilation, the target patch in the hole and outside contour detection target fill color and found pretreatment, so that interference colors can be partly removed block avoiding noise to some extent.

[0086] 再检测分割水华色块区域,即通过计算机按下式不断地对经过上述预处理之后的反向投影图计算均值平移向量: [0086] redetection bloom color region segmentation, i.e., the following equation is calculated continuously the mean shift vector through said reverse projection image after preprocessing by the computer:

[0087] [0087]

[0088] 其中:h为带宽度,C为归一化常数,>Jx)为核K上带宽度为h的多变量密度估计: [0088] where: h is the width, C is a normalization constant,> Jx) width of the core K of h multivariate density estimation:

[0089] [0089]

[0090] 集合{x'」}」=1,2.』是(1维欧氏空间炉的11个点浊00表示该像素点的核函数: [0090] The set {x ' "}" = 1 "is a (1-dimensional Euclidean space of the furnace a cloud point of 11 00 pixels of the kernel:

[0091] [0091]

[0092] 其中:(^为d维单位球体体积。用以迭代更新目标色块的位置,直到收敛于的最优匹配点,该匹配点的区域就是当前帧图像中的目标色块(即水华色块区域)。 [0092] wherein: (^ for the d-dimensional unit sphere volume for the target location update patches iteration, until convergence to the optimum matching point, the area of ​​the patch is the goal of matching points in the current frame image (i.e., water Hua color area).

[0093] 进行判断:当没有检测到目标色块时,则认为水面未出现水华,返回第(2)_①步获取下一帧继续进行检测;当有检测到目标色块时,则以当前帧最优匹配点的质心和面积作为下一帧搜索窗口的初始值,继续进行迭代计算找到最优匹配点。 [0093] judge: when the target color is not detected, then the surface does not appear that water blooms, return to the first (2) _① step acquires the next frame is detected to continue; if there is a target patch is detected, places the current centroid and area of ​​optimal matching point of the frame as the initial value of the next frame search window, continued iterative calculation to find the optimal match point. 如此循环迭代至程序结束为止,就检测分割出每帧图像的目标色块。 Such loop iteration until end of the program, segment object is detected in each image patch.

[0094] 然后过滤干扰色块,即计算上步得到的当前帧和下一帧的目标色块的面积,根据水华区域面积的帧间变化规律,利用当前帧的目标色块面积来预测下一帧图像中的目标色块面积,即计算下一帧色块实际检测面积和下一帧预测面积之间的差值(即多余部分),根据下式进行判断: [0094] was then filtered interference color, i.e., calculation of the target color on the current frame obtained in step and the next frame in the area, according to the variation of the area of ​​inter-bloom, to predict the current frame using the target area of ​​the patch a target color image area, i.e., calculating the difference (i.e., the excess portion) between the detected actual color area of ​​the next frame and the next frame prediction area, is determined according to the following equation:

[0095] Sn+1-S' n+1| ^ I (5) [0095] Sn + 1-S 'n + 1 | ^ I (5)

[0096] 其中:Sn+1为下一帧色块的实际测量面积,S' n+1为下一帧的预测面积,I为允许 [0096] where: Sn + 1 of the next frame the actual measurement patch area, S 'n + 1 for the next frame prediction area, I is allowed

11的误差范围。 11 error range.

[0097] 当|Sn+1_S' n+1| ( \时,则认为下一帧实际检测到的色块面积符合水华区域面积的帧间变化规律并将其作为检测结果;当|Sn+1_S' n+1| > €,即差值超出允许的误差范围内时,则认为下一帧实际检测到的色块面积不符合帧间水华区域面积的变化规律,将多余部分定义为干扰色块并将其滤掉。 [0097] When | Sn + 1_S 'n + 1 | (\, then the next frame that is actually detected variation of the inter-color area in line with an area of ​​the blooms and as a detection result; when | Sn + 1_S 'n + 1 |> €, i.e., when the difference exceeds the allowable error range, then the next frame that is actually detected variation of color area does not meet the area of ​​the inter-bloom, the excess portion is defined as an interference color and filtered out.

[0098] 经过上述的处理后,就检测分割出滤掉干扰色块后的水华色块,得到水华色块在图像中的像素面积。 [0098] After the above processing, it detects the divided color after bloom filter out the interference color, the pixel area to obtain blooms in the image patch. 然后对每帧图像中的水华色块区域进行标定并输出检测结果。 Then the patch to bloom in the region of each frame image calibration and outputs a detection result.

[0099] (3)水华预警 [0099] (3) blooms early warning

[0100] ①计算水华区域实际面积: [0100] ① calculating the actual area of ​​blooms region:

[0101] 第(2)步完成后,先计算第(2)步得到的水华色块在图像中的像素面积占视野像素总面积的百分比,乘以视野的实际总面积就能得到水华区域实际面积。 [0101] (2) After completion of steps, the first calculation (2) obtained in step bloom color vision percentage of the total area of ​​the pixel area in the pixels in the image, the actual total field of view can be obtained by multiplying the area of ​​bloom the actual area of ​​the region. 再通过摄像机标定,计算视野的实际总面积,即将已知面积的参照物置于摄像头下的水域位置,计算它在图像中的像素面积占视野像素总面积的比值,用该已知参照物的实际面积除以该比值就得到视野的实际总面积,并将得到的水华区域实际面积输出到计算机软件界面上,以便用户直观的了解水华区域的实际面积。 Through the camera calibration, the actual calculation of the total area of ​​the field of view, i.e. a reference of known area is placed under the camera position of the waters, which calculates the pixel area in the image field of view accounts for the ratio of the total area of ​​the pixel, the actual reference using the known this ratio is obtained by dividing the area of ​​the total area of ​​the actual field of view, and outputs the resulting actual area bloom area to the computer software interface for the user to intuitively know the actual area of ​​the bloom area.

[0102] ②建立预警模型: [0102] ② early warning model:

[0103] 第(3)_①步完成后,根据第(3)_①步得到的水华区域实际面积,先按8小时为间隔,计算每个时间间隔内的水华区域实际面积的均值,再通过计算机分析该实际面积的均值变化规律建立一元线性回归预测模型,来预测下一个时间间隔的实际面积的均值,并计算该实际面积的均值占水域总面积的百分比,然后以该百分比代表的水华灾害程度为依据,将水华暴发的预警等级划分为蓝色、黄色、橙色和红色四个等级,对应的预警区间分别是一般、较重、严重和特别严重四个等级,并进行判断:当代表的水华灾害程度的百分比小于预警值(预警值的大小根据被测水域的具体情况和应用需求确定,本发明取5%)时,则返回到返回第(2)-①步获取下一帧继续进行检测;当代表的水华灾害程度的百分比大于预警值时,则根据划分的预警等级进行预警。 [0103] section (3) _① After completion of steps, in accordance with section (3) _① actual area obtained in step bloom area, press 8 hours intervals, calculate the mean actual area of ​​the bloom area within each time interval, then by computer analysis of the mean variation of the actual area to establish a linear regression prediction model to predict the average of the actual area of ​​the next time interval, and calculating the actual area of ​​the mean percentage of the total area of ​​water, and then the percentages represent water China is based on the extent of the disaster, will bloom early warning level is divided into blue, yellow, orange and red four levels, corresponding warning range are generally heavier, serious and especially serious four grades, and to judge: when the percentage of water bloom degree of hazard represented less than the warning value (warning value of magnitude depending on the circumstances and the application needs to determine the test waters, the present invention is taken 5%), then to the return section (2) -① acquired at step detecting a continued; China when the percentage of water greater than the degree of hazard represented by the warning value, the early warning alarm according to the classification. 水华暴发预警等级的具体划分如下: They are divided bloom outbreak alert level is as follows:

[0104] 一般蓝色预警(即零星性水华):水华零星集聚,主要水域区藻类生物密度小于3000万个/L,水华面积大于等于水体总面积的5%。 [0104] Usually blue warning (i.e. sporadic water-Hua): gathering scattered bloom, the main areas of algae water density of less than 30 million / L, blooms area greater than or equal to 5% of the total water.

[0105] 较重黄色预警(即局部性水华):藻类在局部水域集聚,主要水域区藻类生物密度介于3000〜5000万个/L之间,水华面积大于等于水体总面积的10%。 [0105] heavy yellow warning (i.e. blooms locality): water partial gathering of algae, algae density primary water zone between ten thousand 3000~5000 / L, blooms area greater than or equal to 10% of the total water .

[0106] 严重橙色预警(即区域性水华):当发生区域性水华,主要水域区藻类生物密度介于5000〜8000万个/L之间,水华面积大于等于水体总面积的40%。 [0106] severe orange alert (i.e. blooms regional): When the regional water bloom occurs, the main water algae density areas between 5000~8000 million / L, between bloom area greater than or equal to 40% of the total water .

[0107] 特别严重红色预警(即全面性水华):水华全面性暴发,主要水域区的藻类生物密度大于8000万个/L,水华面积水华面积大于等于水体总面积的60%。 [0107] particularly serious red alert (ie comprehensive blooms): bloom comprehensive outbreak, major water algae density region is greater than 80 million / L, the area blooms bloom area greater than or equal to 60% of the total area of ​​the water body.

[0108] 对该基于图像处理的小型水域水华监测预警方法进行测试后,可以得到以下结论: [0108] After the test the bloom monitoring and early warning method based on image processing of small waters, the following conclusions:

[0109] ①在监测过程中,被测水域(重庆市开县澎溪河水库)水面发生水华,本发明方法能实时检测出水华色块区域并计算出该区域的实际面积,能准确按照预警等级进行预警, 在水华暴发前期,重庆市环保局能及时掌握该水域的水华灾情并通知相关部门采取相应措施进行治理(在零星性水华时进行打捞),有力的配合了防灾减灾工作。 [0109] ① in the monitoring process, the measured water (Kaixian Pengxihe reservoir) water bloom occurs, real-time detection method of the present invention, the water Chinese patch area and calculating the actual area of ​​the region, in accordance with accurately warning level early warning, early outbreaks in bloom, the Chongqing Municipal Environmental Protection Bureau to grasp the disaster blooms in the waters and notify the relevant authorities to take appropriate measures to control (in the salvage of sporadic blooms), effective with the disaster Prevention and mitigation jobs. [0110] ②本发明方法仅仅使用计算机和一个摄像头就能快速有效的实现水华监测预警, 很大程度上节约了监测成本,而且用户使用方便,维护成本低,监测效果能够满足重庆市环保局的实际需求,从而便于推广应用。 [0110] ② method of the invention using only a computer and a camera will be able to quickly and efficiently implement bloom monitoring and early warning, monitoring largely to save costs, and the user easy to use, low cost maintenance, monitoring results can satisfy the Chongqing Municipal Environmental Protection Bureau actual demand, thereby facilitating application.

[0111] ③本发明采用改进的连续自适应性均值漂移算法来检测分割水华色块并过滤干扰色块,能有效抑制水面背景复杂性和环境光线变化等带来的噪声干扰。 [0111] ③ of the present invention employs an improved noise caused by continuous adaptive mean shift segmentation algorithm to detect interference bloom color patch and filtered, the water can effectively suppress the complexity of the background light and other environmental changes. 该算法计算量小、 速度快,是保证水华监测预警实时性的基础。 The algorithm to calculate the amount of small, fast, is to ensure that bloom early warning monitoring real-time basis.

[0112] 上述结论说明利用本发明方法实现的基于图像处理的小型水域水华监测预警能够实现水华实时监测预警,监测结果符合水华区域的实际分布范围,监测效果能够满足使用部门的实际需求,监测成本低,对环境适应能力强,便于推广应用,因此本发明能应用于实际的项目中。 [0112] The conclusions described with reference to the actual needs of the method according to the present invention is realized based on small waters image processing blooms monitoring and early warning possible Bloom real-time monitoring and early warning monitoring results realistic distribution bloom area, monitor the effects of possible to meet the sector low cost of monitoring, environmental adaptability strong, easy application, and therefore the present invention can be applied to the actual project.

Claims (1)

  1. 一种基于图像处理的小型水域水华监测预警方法,其特征在于利用摄像头和计算机,通过程序进行计算,其具体步骤如下:(1)建立颜色先验模型首先根据被测水域的历史水华区域的样本图或视频数据,建立颜色先验模型,即:设{x′j}j=1,2n为样本图或视频数据的n个像素,将每个象素从三原色空间转换到色、亮分离空间,将色调分量的离散化取值为i,即i=1,2...,360,把色调分量的范围[0,360]缩放到[0,255];通过计算机按下式计算色调分量的色调概率:p={pi}i=1,2m; <mrow> <msub> <mi>p</mi> <mi>i</mi> </msub> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <msup> <mrow> <mo>|</mo> <mo>|</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mo>&prime;</mo> </msubsup> <mo>|</mo> <mo>|</mo> </mrow> <mn>2</mn> </msup> <mi>&delta;</mi> <mo>[</mo> <mi>b</mi> <mrow> <mo>(</mo> A water-bloom of small water warning method based on image processing, characterized by using a camera and a computer, calculated by the program, the specific steps are as follows: (1) establishing a color prior model is first measured in accordance with the history area waters bloom FIG sample or video data, a priori model of color, namely: set {x'j} j = 1,2n FIG into n pixel samples or video data, each pixel will be converted from the three primary colors to the color space, bright separating space, the hue component of the discrete values ​​of i, i.e., i = 1,2 ..., 360, the hue component of the range [0,360] scaled to [0, 255]; computer by the following formula hue hue component probability: p = {pi} i = 1,2m; <mrow> <msub> <mi> p </ mi> <mi> i </ mi> </ msub> <mo> = </ mo > <mfrac> <mrow> <munderover> <mi> & Sigma; </ mi> <mrow> <mi> j </ mi> <mo> = </ mo> <mn> 1 </ mn> </ mrow> <mi> n </ mi> </ munderover> <msup> <mrow> <mo> | </ mo> <mo> | </ mo> <msubsup> <mi> x </ mi> <mi> j < / mi> <mo> & prime; </ mo> </ msubsup> <mo> | </ mo> <mo> | </ mo> </ mrow> <mn> 2 </ mn> </ msup> <mi > & delta; </ mi> <mo> [</ mo> <mi> b </ mi> <mrow> <mo> (</ mo> <msubsup> <mi>x</mi> <mi>j</mi> <mo>&prime;</mo> </msubsup> <mo>)</mo> </mrow> <mo>-</mo> <mi>i</mi> <mo>]</mo> </mrow> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <mrow> <mo>(</mo> <msup> <mrow> <mo>|</mo> <mo>|</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mo>&prime;</mo> </msubsup> <mo>|</mo> <mo>|</mo> </mrow> <mn>2</mn> </msup> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>;</mo> </mrow> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>p</mi> <mi>i</mi> </msub> <mo>=</mo> <mn>1</mn> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow>其中:p是颜色模型,pi是色调值为i的概率,δ为δ函数,函数b(x′j)为空间R2→{1,2…m}的索引,即位于位置x′j的像素向直方图量化特征空间的索引;(2)对水华区域进行检测①对小型水域进行监测第(1)步完成后,在小型水域 <Msubsup> <mi> x </ mi> <mi> j </ mi> <mo> & prime; </ mo> </ msubsup> <mo>) </ mo> </ mrow> <mo> - </ mo> <mi> i </ mi> <mo>] </ mo> </ mrow> <mrow> <munderover> <mi> & Sigma; </ mi> <mrow> <mi> j </ mi> <mo > = </ mo> <mn> 1 </ mn> </ mrow> <mi> n </ mi> </ munderover> <mrow> <mo> (</ mo> <msup> <mrow> <mo> | </ mo> <mo> | </ mo> <msubsup> <mi> x </ mi> <mi> j </ mi> <mo> & prime; </ mo> </ msubsup> <mo> | < / mo> <mo> | </ mo> </ mrow> <mn> 2 </ mn> </ msup> <mo>) </ mo> </ mrow> </ mrow> </ mfrac> <mo> ; </ mo> </ mrow> <mrow> <munderover> <mi> & Sigma; </ mi> <mrow> <mi> i </ mi> <mo> = </ mo> <mn> 1 </ mn > </ mrow> <mi> m </ mi> </ munderover> <msub> <mi> p </ mi> <mi> i </ mi> </ msub> <mo> = </ mo> <mn > 1 </ mn> <mo> - </ mo> <mo> - </ mo> <mo> - </ mo> <mrow> <mo> (</ mo> <mn> 1 </ mn> < mo>) </ mo> </ mrow> </ mrow> wherein: p is a color model, pi is the probability of the tone value i, [delta] [delta] is a function, the function b (x'j) spatial R2 → {1, index 2 ... m}, i.e. the spatial location at index to the histogram of pixel x'j quantitative feature; (2) to bloom area ① of detecting small waters monitoring after completion of step (1), in a small water 岸边架设摄像头,并通过视频传输线与计算机连接;摄像头架设的具体位置、数量及高度,根据被测水域的具体情况确定;②反向投影处理第(2)‑①步完成后,对第(2)‑①步输入的每一帧视频图像进行反向投影处理,即对视频图像处理区域中的每一个像素,通过计算机查询该像素与第(1)步建立的颜色先验模型的匹配程度,就得到每帧图像的目标颜色反向投影图;③检测分割水华色块并滤掉干扰色块第(2)‑②步完成后,进行检测水华色块和滤掉干扰色块处理,即:先对第(2)‑②步得到的反向投影图进行腐蚀、膨胀,检测目标色块的内外轮廓及找到目标色块中的空洞进行填充预处理;再检测分割水华色块区域,即通过计算机按下式不断地对经过上述预处理之后的反向投影图计算均值平移向量Mh,G(x): <mrow> <msub> <mi>M</mi> <mrow> <mi>h</mi> <mo>,</mo> <mi>G</mi> <mr Shore set up the camera and connected with the computer through the video transmission line; camera erected particular location, number and height, determined according to the specific circumstances of the test waters; ② backprojection processing step after completion of -① (2), the first ( 2) -① each frame of video image input step of reverse projection process, i.e., the video image processing area in each pixel, pixel by querying the computer (1) the degree of color matching model created in step priori , the target color is obtained in each image reverse projection; ③ detection and segmentation bloom filter out the interference color of the patch (2) -② after completion of the step, and detecting color bloom filter out interference color processing , namely: the first first (2) -② FIG backprojection step was subjected to erosion, dilation, the target patch in the hole and outside contour detection target fill color and found pretreatment; bloom color segmentation redetection region, i.e., the following equation is calculated continuously mean shift vector Mh, G (x) after the above-described pretreatment of backprojection FIG computer: <mrow> <msub> <mi> M </ mi> <mrow> < mi> h </ mi> <mo>, </ mo> <mi> G </ mi> <mr ow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> </msub> <mo>=</mo> <mfrac> <mrow> <msup> <mi>h</mi> <mn>2</mn> </msup> <mo>&dtri;</mo> <msub> <mover> <mi>f</mi> <mo>^</mo> </mover> <mi>K</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mn>2</mn> <mo>/</mo> <mi>C</mi> <msub> <mover> <mi>f</mi> <mo>^</mo> </mover> <mi>G</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow>其中:h为带宽度,C为归一化常数,为核K上带宽度为h的多变量密度估计: <mrow> <msub> <mover> <mi>f</mi> <mo>^</mo> </mover> <mi>K</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <msup> <mi>nh</mi> <mi>d</mi> </msup> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <mi>k</mi> <mrow> <mo>(</mo> <msup> <mrow> <mo>|</mo> <mo>|</mo> <mfrac> <mrow> <mi ow> <mo> (</ mo> <mi> x </ mi> <mo>) </ mo> </ mrow> </ mrow> </ msub> <mo> = </ mo> <mfrac> < mrow> <msup> <mi> h </ mi> <mn> 2 </ mn> </ msup> <mo> & dtri; </ mo> <msub> <mover> <mi> f </ mi> <mo > ^ </ mo> </ mover> <mi> K </ mi> </ msub> <mrow> <mo> (</ mo> <mi> x </ mi> <mo>) </ mo> < / mrow> </ mrow> <mrow> <mn> 2 </ mn> <mo> / </ mo> <mi> C </ mi> <msub> <mover> <mi> f </ mi> <mo > ^ </ mo> </ mover> <mi> G </ mi> </ msub> <mrow> <mo> (</ mo> <mi> x </ mi> <mo>) </ mo> < / mrow> </ mrow> </ mfrac> <mo> - </ mo> <mo> - </ mo> <mo> - </ mo> <mrow> <mo> (</ mo> <mn> 2 </ mn> <mo>) </ mo> </ mrow> </ mrow> where: h is the width, C is a normalization constant, as the width of the core K of h multivariate density estimation: <mrow > <msub> <mover> <mi> f </ mi> <mo> ^ </ mo> </ mover> <mi> K </ mi> </ msub> <mrow> <mo> (</ mo> <mi> x </ mi> <mo>) </ mo> </ mrow> <mo> = </ mo> <mfrac> <mn> 1 </ mn> <msup> <mi> nh </ mi> <mi> d </ mi> </ msup> </ mfrac> <munderover> <mi> & Sigma; </ mi> <mrow> <mi> j </ mi> <mo> = </ mo> <mn> 1 </ mn> </ mrow> <mi> n </ mi> </ munderover> <mi> k </ mi> <mrow> <mo> (</ mo> <msup> <mrow> <mo> | </ mo> <mo> | </ mo> <mfrac> <mrow> <mi >x</mi> <mo>-</mo> <msub> <mi>x</mi> <mi>i</mi> </msub> </mrow> <mi>h</mi> </mfrac> <mo>|</mo> <mo>|</mo> </mrow> <mn>2</mn> </msup> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow>集合{x′j}i=1,2...n是d维欧氏空间Rd的n个点,k(x)表示该像素点的核函数:其中:cd为d维单位球体体积;用以迭代更新目标色块的位置,直到收敛于的最优匹配点,该匹配点的区域就是当前帧图像中的目标色块;进行判断:当没有检测到目标色块时,则返回第(2)‑①步获取下一帧继续进行检测;当有检测到目标色块时,则以当前帧最优匹配点的质心和面积作为下一帧搜索窗口的初始值,继续进行迭代计算找到最优匹配点,如此循环迭代至程序结束为止,就检测分割出每帧图像的目标色块;然后过滤干扰色块,即计算上步得到的当前帧和下一帧的目标色块的面积,根据水华 > X </ mi> <mo> - </ mo> <msub> <mi> x </ mi> <mi> i </ mi> </ msub> </ mrow> <mi> h </ mi> < / mfrac> <mo> | </ mo> <mo> | </ mo> </ mrow> <mn> 2 </ mn> </ msup> <mo>) </ mo> </ mrow> <mo> - </ mo> <mo> - </ mo> <mo> - </ mo> <mrow> <mo> (</ mo> <mn> 3 </ mn> <mo>) </ mo> </ mrow> </ mrow> set {x'j} i = 1,2 ... n d is a n-dimensional Euclidean space Rd of points, k (x) denotes the pixel of the kernel function: wherein: cd is d volume-dimensional unit sphere; location update for the target color of iterations, until convergence to the optimum matching point, the area of ​​the matching point of the current frame is a target color image; judge: when no object is detected patch , the return (2) -① step acquires the next frame is detected to continue; if there is a target patch is detected, and places the current centroid optimum matching point of the frame area as an initial value of the next frame search window, continued iterative calculation to find the optimal matching point, and so on until the end of the iteration procedure segment object is detected in each image patch; then filtered and the interference color, the target color that is calculated on the current frame obtained in step and the next frame block area, according to bloom 区域面积的帧间变化规律,利用当前帧的目标色块面积来预测下一帧图像中的目标色块面积,即计算下一帧色块实际检测面积和下一帧预测面积之间的差值,根据下式进行判断:|Sn+1‑S′n+1|≤ξ (5)其中:Sn+1为下一帧色块的实际测量面积,S′n+1为下一帧的预测面积,ξ为允许的误差范围;当|Sn+1‑S′n+1|≤ξ时,则认为下一帧实际检测到的色块面积符合水华区域面积的帧间变化规律并将其作为检测结果;当|Sn+1‑S′n+1|>ξ,即差值超出允许的误差范围内时,则认为下一帧实际检测到的色块面积不符合帧间水华区域面积的变化规律,将多余部分定义为干扰色块并将其滤掉;检测分割出滤掉干扰色块后的水华色块并进行标定并输出检测结果;(3)水华预警①计算水华区域实际面积:第(2)步完成后,先计算第(2)步得到的水华色块在图像中 Variation of the inter-regional area, the current frame using the target color patch area to predict the target area in the next frame, i.e., calculate the difference between the detected actual color area of ​​the next frame and the next frame prediction area , determined according to the following equation: | Sn + 1-S'n + 1 | ≤ξ (5) where: Sn + 1 is the actual color measurement area of ​​the next frame, S'n + 1 for the next frame prediction area, ξ is the allowable error range; if | Sn + 1-S'n + 1 | when ≤ξ, the next frame is considered the actual color detected area in line with changes of the area of ​​inter-bloom and as a result of the detection; if | Sn + 1-S'n + 1 |> ξ, i.e. when the difference exceeds the allowable error range, then the next frame that is actually detected color area does not meet the area of ​​inter-bloom the variation of the excess portion is defined as an interference color and was filtered; detecting divided bloom filter out interference color patch and the calibration and outputs a detection result; (3) warning ① calculation bloom bloom the actual area of ​​the region: (2) after completion of steps, the first calculation step (2) obtained in the image patch blooms 的像素面积占视野像素总面积的百分比,乘以视野的实际总面积就能得到水华区域实际面积;再通过摄像机标定,计算视野的实际总面积,即将已知面积的参照物置于摄像头下的水域位置,计算它在图像中的像素面积占视野像素总面积的比值,用该已知参照物的实际面积除以该比值就得到视野的实际总面积,并将得到的水华区域实际面积输出到计算机软件界面上;②建立预警模型:第(3)‑①步完成后,根据第(3)‑①步得到的水华区域实际面积,先按6~10小时为间隔,计算每个时间间隔内的水华区域实际面积的均值,再通过计算机分析该实际面积的均值变化规律建立一元线性回归预测模型,来预测下一个时间间隔的实际面积的均值,并计算该实际面积的均值占水域总面积的百分比,然后以该百分比代表的水华灾害程度为依据,将水华暴发的预警等级 Pixel area to total area of ​​the field of view of the pixel, the total area of ​​the actual field of view can be obtained by multiplying the actual area bloom region; through the camera calibration, to calculate the actual total area of ​​the field of view, i.e. reference of known area is placed under the camera water position, which is calculated in the pixel area of ​​the image at the ratio of the total area of ​​the field of view of the pixel, the actual area of ​​the known reference ratio is obtained by dividing the total area of ​​the actual field of view, and the actual area of ​​the output region blooms obtained to the computer software interface; ② model for early warning: a first (3) -① after completion of steps, in accordance with section (3) -① actual area obtained in step bloom area, press the 6 to 10 hour intervals, calculated for each time mean the actual area of ​​the bloom area within the interval, then create a linear regression prediction model to predict the average of the actual area of ​​the next interval by computer analysis of the mean variation of the actual area, and calculate the mean of the actual area accounted water the percentage of the total area, and then bloom to the extent of the disaster on behalf of the percentage is based, will bloom early warning level 分为蓝色、黄色、橙色和红色四个等级,对应的预警区间分别是一般、较重、严重和特别严重四个等级,并进行判断:当代表的水华灾害程度的百分比小于预警值时,则返回到返回第(2)‑①步获取下一帧继续进行检测;当代表的水华灾害程度的百分比大于预警值时,则根据划分的预警等级进行预警;水华暴发预警等级的具体划分如下:一般蓝色预警:水华零星集聚,主要水域区藻类生物密度小于3000万个/L,水华面积大于等于水体总面积的5%;较重黄色预警:藻类在局部水域集聚,主要水域区藻类生物密度介于3000~5000万个/L之间,水华面积大于等于水体总面积的10%;严重橙色预警:当发生区域性水华,主要水域区藻类生物密度介于5000~8000万个/L之间,水华面积大于等于水体总面积的40%;特别严重红色预警:水华全面性暴发,主要水域区的藻类 Classified as blue, yellow, orange and red four levels, corresponding to the warning section are generally heavier, four particularly serious and severe levels, and is determined: when the percentage of water bloom degree of hazard represented by the warning value is less than , then to the return section (2) -① step acquires the next frame is detected to proceed; China when the percentage of water greater than the degree of hazard represented by the warning value, the early warning alarm according to the classification; bloom specific warning level divided as follows: general blue warning: gathering scattered blooms, the main areas of algae water density of less than 30 million / L, blooms area greater than or equal to 5% of the total area of ​​the body of water; heavy yellow warning: algae concentrate in local waters, mainly water algae density zone is between 3,000 to 50,000,000 / L, blooms area greater than or equal to 10% of the total water; severe orange warning: as regional water bloom occurs, the main water area density of from 5,000 to algae 80 million / L, between bloom area greater than or equal to 40% of the total area of ​​the water body; particularly serious red warning: bloom comprehensive outbreak, algae major area of ​​waters 物密度大于8000万个/L,水华面积水华面积大于等于水体总面积的60%。 The density is greater than 80 million / L, the area of ​​water bloom bloom area greater than or equal to 60% of the total water. FSA00000171985200014.tif,FSA00000171985200021.tif FSA00000171985200014.tif, FSA00000171985200021.tif
CN 201010218511 2010-07-06 2010-07-06 Method for monitoring and early warning water bloom in small water area based on image processing CN101900687B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010218511 CN101900687B (en) 2010-07-06 2010-07-06 Method for monitoring and early warning water bloom in small water area based on image processing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010218511 CN101900687B (en) 2010-07-06 2010-07-06 Method for monitoring and early warning water bloom in small water area based on image processing

Publications (2)

Publication Number Publication Date
CN101900687A true true CN101900687A (en) 2010-12-01
CN101900687B CN101900687B (en) 2012-01-04

Family

ID=43226406

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010218511 CN101900687B (en) 2010-07-06 2010-07-06 Method for monitoring and early warning water bloom in small water area based on image processing

Country Status (1)

Country Link
CN (1) CN101900687B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102135531A (en) * 2010-12-24 2011-07-27 中国科学院南京地理与湖泊研究所 Method for forecasting blue-green algae water bloom in large-scale shallow lake within 72 hours
CN104459071A (en) * 2014-12-05 2015-03-25 河海大学 Simulation system for algal bloom emergency regulation and control under changing environment and operation method of simulation system
CN104699979A (en) * 2015-03-24 2015-06-10 北京工商大学 Urban lake and reservoir algal bloom chaos time sequence predication method based on complicated network

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102998664B (en) * 2012-10-29 2014-10-22 中国科学院对地观测与数字地球科学中心 A water-based Chinese recognition method and apparatus of the synthetic aperture radar

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH08194876A (en) * 1995-01-17 1996-07-30 Hitachi Ltd Water surface monitoring device
US6119630A (en) * 1997-05-26 2000-09-19 3042015 Nova Scotia Limited Installation for in situ monitoring the quality of habitat of aquatic organisms
CN101706494A (en) * 2009-11-11 2010-05-12 清华大学;北京城市排水集团有限责任公司科技研发中心 Method for judging and early warning outbreak potential of water bloom of water body

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH08194876A (en) * 1995-01-17 1996-07-30 Hitachi Ltd Water surface monitoring device
US6119630A (en) * 1997-05-26 2000-09-19 3042015 Nova Scotia Limited Installation for in situ monitoring the quality of habitat of aquatic organisms
CN101706494A (en) * 2009-11-11 2010-05-12 清华大学;北京城市排水集团有限责任公司科技研发中心 Method for judging and early warning outbreak potential of water bloom of water body

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
《Estuarine, Coastal and Shelf Science》 20060110 Tiit Kutser etc. Monitoring cyanobacterial blooms by satellite remote sensing 303-312 1 第67卷, 2 *
《仪器仪表学报》 20101231 石为人 基于图像处理的三峡库区水域水华检测算法及预警研究 2641-2647 1 第31卷, 第12期 2 *
《地球信息科学》 20080430 魏清宇 太湖蓝藻水华遥感动态监测预警模型的建立 156-160 1 第10卷, 第2期 2 *
《计算机工程与应用》 20091231 吴慧敏 改进的高效Camshift 跟踪算法 178-180 1 第45卷, 第27期 2 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102135531A (en) * 2010-12-24 2011-07-27 中国科学院南京地理与湖泊研究所 Method for forecasting blue-green algae water bloom in large-scale shallow lake within 72 hours
CN104459071A (en) * 2014-12-05 2015-03-25 河海大学 Simulation system for algal bloom emergency regulation and control under changing environment and operation method of simulation system
CN104459071B (en) * 2014-12-05 2016-04-27 河海大学 Changes in environmental regulation under emergency bloom simulation system and method of operation
CN104699979A (en) * 2015-03-24 2015-06-10 北京工商大学 Urban lake and reservoir algal bloom chaos time sequence predication method based on complicated network

Also Published As

Publication number Publication date Type
CN101900687B (en) 2012-01-04 grant

Similar Documents

Publication Publication Date Title
CN101214851A (en) Intelligent all-weather actively safety early warning system and early warning method thereof for ship running
CN101510358A (en) Method and apparatus for processing real time statistical vehicle flowrate using video image
CN101105690A (en) Multiple sensor integrated method for city long tunnel monitoring system
CN102494733A (en) Water level monitoring system based on image processing and method
CN101430313A (en) Water quality monitoring system based on global positioning system and universal grouping wireless service
CN101430195A (en) Method for computing electric power line ice-covering thickness by using video image processing technology
CN101807352A (en) Method for detecting parking stalls on basis of fuzzy pattern recognition
CN102346015A (en) Method for measuring icing thickness of insulator of power transmission line based on video difference analysis
CN101793501A (en) Transmission line ice coating status detection method based on image
CN103067734A (en) Video image color cast detecting method of video quality diagnostic system
Zhao et al. Estimation of water clarity in Taihu Lake and surrounding rivers using Landsat imagery
Underwood et al. Subtidal assemblages on rocky reefs at a cliff-face sewage outfall (North Head, Sydney, Australia): What happened when the outfall was turned off?
CN201718000U (en) Intelligent dynamic field video monitoring system
Gallop et al. Storm-driven changes in rip channel patterns on an embayed beach
Nelson et al. Land-cover change in upper Barataria Basin estuary, Louisiana, 1972–1992: increases in wetland area
Yu et al. Detecting land use-water quality relationships from the viewpoint of ecological restoration in an urban area
CN1996190A (en) Industrial process fault diagnosis system and method based on wavelet analysis
CN102855638A (en) Detection method for abnormal behavior of vehicle based on spectrum clustering
Liu et al. Wavelet analysis of SAR images for coastal monitoring
CN102565294A (en) Water source area monitoring and evaluation method
CN101860664A (en) Moving target detection method robust to illumination variation
CN103729842A (en) Fabric defect detection method based on local statistical characteristics and overall significance analysis
CN102354457A (en) General Hough transformation-based method for detecting position of traffic signal lamp
CN102879038A (en) Multi-parameter online monitoring system for water environments
CN102110366A (en) Block-based accumulated expressway vehicle parking event detecting method

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model
EXPY Termination of patent right or utility model