公开(公告)号：KR1020120136568A

公开(公告)日：2012-12-20

申请号：KR1020110055574

申请日：2011-06-09

Applicant: 목포대학교산학협력단

Inventor： 박선 ,  이성호 ,  최명수 ,  정민아 ,  이성로

IPC: G06F19/00 ,  G06T1/00

CPC classification number: C40B40/02 ,  G06F17/10 ,  G06T7/62

Abstract: PURPOSE: A red tide image recognition method using meaning characteristic is provided to construct a recognition candidate image group by creating a learning image as an meaning feature image based on NMF(non-negative matrix factorization). CONSTITUTION: A learning red tide image is normalized and is constructed as an image vector group. A meaning characteristic is extracted through an NMF for a processed red tide image(210). A candidate red tide image group is constructed by referring to the meaning characteristic by inputting a red tide image which is recognized by the candidate red tide image group(220). The red tide image is recognized in the candidate red tide image group. The roundness of the inputted red tide image and a learning original image is calculated(310). Proximity images are selected in the roundness. An image which is similar with entropy is recognized as the red tide image by calculating the entropy of the image(320). [Reference numerals] (100) Pre-processing; (200) Searching candidate red tide image; (210) Non-negative matrix factorization; (220) Group of recognition of candidate red tide image; (300) Recognition of red tide image; (310) Calculating roundness; (320) Calculating entropy; (AA,BB) Learning image; (CC) Meaning characteristics

Abstract translation: 目的：提供一种使用意义特征的红潮图像识别方法，通过创建基于NMF（非负矩阵因式分解）的学习图像作为意义特征图像来构建识别候选图像组。 构成：学习红潮图像被归一化，构成为图像矢量组。 通过用于处理的红潮图像（210）的NMF提取意义特征。 通过输入由候选红潮图像组（220）识别的红潮图像，参考含义特征构建候选红潮图像组。 红潮图像在候选红潮图像组中被识别。 计算输入的红潮图像的圆度和学习原始图像（310）。 在圆度中选择接近图像。 通过计算图像的熵，将与熵相似的图像识别为红潮图像（320）。 （附图标记）（100）预处理; （200）搜索候选红潮图像; （210）非负矩阵分解; （220）候选红潮图像识别组; （300）识别红潮图像; （310）计算圆度; （320）计算熵; （AA，BB）学习形象; （CC）含义特征

公开(公告)号：KR1020120136565A

公开(公告)日：2012-12-20

申请号：KR1020110055570

申请日：2011-06-09

Applicant: 목포대학교산학협력단

Inventor： 박선 ,  임정수 ,  이성호 ,  김영주 ,  이성로 ,  양후열

IPC: G06N7/02 ,  G06F19/00

CPC classification number: G06N5/048 ,  G06N7/023

Abstract: PURPOSE: A red tidal occurrence prediction method is provided to predict red tidal by using a fuzzy estimation method and water temperature, temperature, and precipitation information. CONSTITUTION: A group including similar characteristics is estimated from a data group(210). A fuzzy membership is calculated by using a fuzzy membership function for the group(220). A fuzzy rule is created for input data by referring to the fuzzy membership(230). Clustering uses a method of Chiu for estimating a cluster and is used for estimating a potential center. The fuzzy rule is created by using a fuzzy estimation model. [Reference numerals] (100) Pre-processing; (200) Fuzzy estimation; (210) Grouping; (220) Calculating fuzzy membership; (230) Creating fuzzy rule; (AA) Learning material; (BB) Input material; (CC) Predicting occurrence of red tide

Abstract translation: 目的：提出一种红潮预报方法，通过模糊估计法和水温，温度，降水信息预测红潮。 构成：从数据组（210）估计包含相似特征的组。 通过使用组的模糊隶属函数来计算模糊隶属度（220）。 通过参照模糊隶属度（230）为输入数据创建模糊规则。 聚类使用Chiu的方法来估计一个聚类，并用于估计一个潜在的中心。 模糊规则是通过使用模糊估计模型创建的。 （附图标记）（100）预处理; （200）模糊估计; （210）分组; （220）计算模糊隶属度; （230）创建模糊规则; （AA）学习资料; （BB）输入材料; （CC）预测红潮的发生