CN106845532A - A kind of screening sample method - Google Patents
A kind of screening sample method Download PDFInfo
- Publication number
- CN106845532A CN106845532A CN201611270693.1A CN201611270693A CN106845532A CN 106845532 A CN106845532 A CN 106845532A CN 201611270693 A CN201611270693 A CN 201611270693A CN 106845532 A CN106845532 A CN 106845532A
- Authority
- CN
- China
- Prior art keywords
- probability
- picture
- pictures
- average value
- files
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
本发明适用于图像抽取对比技术改进领域，提供了一种样本筛选方法，所述样本筛选方法包括以下步骤：A、计算两个待比较文件夹中图片的相似值的平均值；B、根据得到的平均值求出该平均值下的概率；C、根据概率判断两个文件夹中图片是否为同一人物；当概率越大时，则为同一个人物几率越大，当概率较小时，则为同一个人物几率较小。通过计算图片文件夹之间为同一人物的概率，将图片按照概率从大到小的顺序展示在筛选工具界面上供用户筛选，能够让用户很快的锁定与当前图片有极大可能为同一人物的所有图片。该方法能极大的加快筛选的速度和效率，同时能够尽可能的找到训练样本中所有的重复数据。
The present invention is applicable to the improvement field of image extraction and comparison technology, and provides a sample screening method, which includes the following steps: A, calculating the average value of the similarity values of the pictures in two folders to be compared; B, according to the obtained Calculate the probability under the average value; C. According to the probability, judge whether the pictures in the two folders are the same person; when the probability is higher, the probability of the same person is higher; when the probability is smaller, it is The same person is less likely. By calculating the probability of the same person among the picture folders, the pictures are displayed on the screening tool interface in order of probability from large to small for users to filter, allowing users to quickly lock in the current picture that is most likely to be the same person All images of . This method can greatly speed up the screening speed and efficiency, and at the same time, it can find all the repeated data in the training samples as much as possible.
Description
技术领域technical field
本发明属于图像抽取对比技术改进领域，尤其涉及一种样本筛选方法及系统。The invention belongs to the technical improvement field of image extraction and comparison, and in particular relates to a sample screening method and system.
背景技术Background technique
人脸是人的重要信息，是区分不同的人的重要依据，因此人脸比对是较指纹、虹膜等技术更自然、更直接的比对方式。Face is important information of people and an important basis for distinguishing different people. Therefore, face comparison is a more natural and direct comparison method than fingerprints, irises and other technologies.
人脸比对是将图像或视频输入的人脸通过提取特定的人脸特征信息，与数据库中已注册的人脸特征信息相比较，获得匹配的人脸极其相似度，确认是否与数据库中人脸为同一。Face comparison is to compare the face feature information registered in the database by extracting specific face feature information from the image or video input to obtain the extremely similarity of the matching face and confirm whether it is consistent with the face feature information in the database. The face is the same.
人脸比对在很多场合下都具有非常重要的作用，例如手机彩信中的视频彩信、人机界面、权限控制、智能监视系统等。比对的准确性、精度和鲁棒性问题一直是业界关心的主要问题。Face comparison plays a very important role in many occasions, such as video MMS in mobile MMS, man-machine interface, authority control, intelligent monitoring system, etc. The accuracy, precision and robustness of comparison have always been the main concerns of the industry.
现有技术方法是通过人工的方式依次去判断不同文件夹之间是否为同一人物。该方法效率低，耗时长，且不够准确。The method in the prior art is to sequentially judge whether different folders are the same person in a manual manner. This method is inefficient, time-consuming, and not accurate enough.
发明内容Contents of the invention
本发明的目的在于提供一种样本筛选方法，旨在解决现有技术效率低、耗时长、不够准确的技术问题。The purpose of the present invention is to provide a sample screening method, aiming to solve the technical problems of low efficiency, long time consumption and inaccuracy in the prior art.
本发明是这样实现的，一种样本筛选方法，所述样本筛选方法包括以下步骤：The present invention is achieved in this way, a sample screening method, the sample screening method comprises the following steps:
A、计算两个待比较文件夹中图片的相似值的平均值；A. Calculate the average value of the similarity values of the pictures in the two folders to be compared;
B、根据得到的平均值求出该平均值下的概率；B. Calculate the probability under the average value based on the obtained average value;
C、根据概率判断两个文件夹中图片是否为同一人物；当概率越大时，则为同一个人物几率越大，当概率较小时，则为同一个人物几率较小。C. Judging whether the pictures in the two folders are the same person according to the probability; when the probability is higher, the probability of being the same person is higher; when the probability is smaller, the probability of being the same person is smaller.
本发明的进一步技术方案是：所述步骤A中还包括以下步骤：A further technical solution of the present invention is: the step A also includes the following steps:
A1、依次计算两个文件夹中所有图片的相似值；A1. Calculate the similarity values of all pictures in the two folders in turn;
A2、根据所得到的所有相似值求和并进而求出平均值。A2. According to the sum of all obtained similar values, and then calculate the average value.
本发明的进一步技术方案是：所述步骤A2中两文件夹中图片的平均值为savg，其公式为：m，n为文件夹A和B中分别有图片张数，为A中图片ai和B中图片bj的相似值。A further technical solution of the present invention is: the average value of the pictures in the two folders in the step A2 is s avg , and its formula is: m, n are the number of pictures in folders A and B respectively, is the similarity value of picture a i in A and picture b j in B.
本发明的进一步技术方案是：所述步骤B中两文件夹中图片为同一人的概率为PAB，其公式为其中，savg为两文件夹中图片的平均值，Psc为两张图片为同一人物的概率，1-Psc为两张图片为不同人物的概率，Pf(s)为两张图片是不同人物且相似度为s的概率，Pt(s)为两张图片为相同人物且相似度为s的概率。A further technical solution of the present invention is: the probability that the pictures in the two folders in the step B are of the same person is P AB , and its formula is Among them, s avg is the average value of the pictures in the two folders, P sc is the probability that the two pictures are the same person, 1-P sc is the probability that the two pictures are different people, P f (s) is the probability that the two pictures are The probability of different characters with a similarity of s, P t (s) is the probability that two pictures are of the same character with a similarity of s.
本发明的另一目的在于提供样本筛选方法，所述样本筛选方法包括以下步骤：Another object of the present invention is to provide a sample screening method, which includes the following steps:
a、依次计算两个文件夹中所有图片的相似值；a. Calculate the similarity value of all pictures in the two folders in turn;
b、根据所得到的所有相似值求和；b. Summing all similar values obtained;
c、根据相似值求得的和计算出平均值；c. Calculate the average value based on the sum obtained from similar values;
d、根据得到的平均值判断两个文件夹中图片是否为同一人，当平均值越高时，则两个文件夹中的图片为同一人物的几率越大，当平均值越低时，则两个文件夹中的图片为同一人物的几率越小。d. Judging whether the pictures in the two folders are the same person according to the obtained average value. When the average value is higher, the probability that the pictures in the two folders are the same person is greater. When the average value is lower, The less likely the pictures in the two folders are of the same person.
本发明的进一步技术方案是：所述步骤a中的相似值为sAB，其公式为其中，m、n为两个文件夹中图片的张数，为A中图片ai和B中图片bj的相似值。A further technical solution of the present invention is: the similarity value in the step a is s AB , and its formula is Among them, m and n are the number of pictures in the two folders, is the similarity value of picture a i in A and picture b j in B.
本发明的另一目的在于提供一种样本筛选方法，所述样本筛选方法包括以下步骤：Another object of the present invention is to provide a sample screening method, which includes the following steps:
(1)、依次计算两个待比较文件夹中图片为同一人物的概率；(1), sequentially calculate the probability that the pictures in the two folders to be compared are the same person;
(2)、将所求得的所有概率值相加求出平均值；(2), adding all the obtained probability values to obtain the average value;
(3)、根据求得的平均概率值判断两文件夹中图片是否为同一人物，当平均概率值越大，则两个文件夹中图片为同一个人的几率越大，当平均概率值越小，则两个文件夹中图片为同一个人的几率越小。(3) According to the obtained average probability value, judge whether the pictures in the two folders are the same person. When the average probability value is larger, the probability that the pictures in the two folders are the same person is greater. When the average probability value is smaller , the probability that the pictures in the two folders are the same person is smaller.
本发明的进一步技术方案是：所述步骤(1)中根据局部人工抽样筛选之后的数据统计与计算两张图片为同一人物的概率为Psc，其公式为：N为图片集中的图片张数，S为相同人物图片对数。A further technical solution of the present invention is: in the step (1), according to the data statistics and calculation after local manual sampling and screening, the probability that the two pictures are the same person is P sc , and its formula is: N is the number of pictures in the picture set, and S is the number of pairs of pictures of the same person.
本发明的进一步技术方案是：所述步骤(1)中两张图片为相同人物且相似度为s的概率为Pt(s)，其公式为两张图片是不同人物且相似度为s的概率为Pf(s)，其公式为N为图片集中的图片张数，S为相同人物图片对数，TS为相同人物且相似度为s的图片对数，FS为不是同一人物且相似度为s的图片对数。A further technical solution of the present invention is: the probability that the two pictures in the step (1) are the same person and the similarity is s is P t (s), and its formula is The probability that two pictures are of different characters and the similarity is s is P f (s), and its formula is N is the number of pictures in the picture set, S is the logarithm of pictures of the same person, TS is the logarithm of pictures of the same person with a similarity of s, and FS is the logarithm of pictures of not the same person with a similarity of s.
本发明的有益效果是：通过计算图片文件夹之间为同一人物的概率，将图片按照概率从大到小的顺序展示在筛选工具界面上供用户筛选，能够让用户很快的锁定与当前图片有极大可能为同一人物的所有图片。该方法能极大的加快筛选的速度和效率，同时能够尽可能的找到训练样本中所有的重复数据。The beneficial effects of the present invention are: by calculating the probability of the same person among the picture folders, the pictures are displayed on the screening tool interface in order of probability from large to small for users to filter, so that users can quickly lock in the current picture All pictures with a high probability of being the same person. This method can greatly speed up the screening speed and efficiency, and at the same time, it can find all the repeated data in the training samples as much as possible.
附图说明Description of drawings
图1是本发明实施例提供的样本筛选方法的流程图一。Fig. 1 is a flow chart 1 of the sample screening method provided by the embodiment of the present invention.
图2是本发明实施例提供的样本筛选方法的流程图二。Fig. 2 is the second flowchart of the sample screening method provided by the embodiment of the present invention.
图3是本发明实施例提供的样本筛选方法的流程图三。Fig. 3 is the third flowchart of the sample screening method provided by the embodiment of the present invention.
具体实施方式detailed description
图1示出了本发明提供的样本筛选方法的流程图，其详述如下：Fig. 1 shows the flowchart of the sample screening method provided by the present invention, which is described in detail as follows:
步骤S11，计算两个带比较文件夹中图片的相似值的平均值；通过相似值求和的平均值法和概率法相结合，可以先算出待比较的两个文件夹中图片相似值的平均值；其中求平均值，首先，依次计算两个文件夹中所有图片的相似值；对于任意两个待比较的文件夹，可以依次计算一个文件夹中图片与另一文件夹中所有图片的相似值；其次，根据所得到的所有相似值求和并进而求出平均值；然后将计算所得的所有相似值求和，进而求出平均值；其中，两文件夹中图片的平均值为savg，其公式为：m，n为文件夹A和B中分别有图片张数，为A中图片ai和B中图片bj的相似值。Step S11, calculate the average value of the similarity values of the pictures in the two folders with comparison; through the combination of the mean value method and the probability method of the similarity value summation, the average value of the similarity values of the pictures in the two folders to be compared can be calculated first ;Average, first, calculate the similarity value of all pictures in the two folders in turn; for any two folders to be compared, you can calculate the similarity value of the pictures in one folder and all the pictures in the other folder in turn ;Secondly, according to the summation of all the obtained similarity values and then calculate the average value; then sum up all the calculated similarity values and then calculate the average value; wherein, the average value of the pictures in the two folders is s avg , Its formula is: m, n are the number of pictures in folders A and B respectively, It is the similarity value of picture a i in A and picture b j in B.
步骤S12，根据得到的平均值求出该平均值下的概率；根据该平均值求出该平均值下的概率为PAB，其中，两文件夹中图片为同一人的概率为PAB，其公式为其中，savg为两文件夹中图片的平均值，Psc为两张图片为同一人物的概率，1-Psc为两张图片为不同人物的概率，Pf(s)为两张图片是不同人物且相似度为s的概率，Pt(s)为两张图片为相同人物且相似度为s的概率；Pf(s)，Pt(s)是根据前期大量的实验数据统计出来的两个函数，s是某个相似度。Step S12, calculate the probability under the average value according to the obtained average value; calculate the probability under the average value according to the average value as P AB , where the probability that the pictures in the two folders are the same person is P AB , where The formula is Among them, s avg is the average value of the pictures in the two folders, P sc is the probability that the two pictures are the same person, 1-P sc is the probability that the two pictures are different people, P f (s) is the probability that the two pictures are The probability of different characters with a similarity of s, P t (s) is the probability that two pictures are the same person with a similarity of s; P f (s), P t (s) is calculated based on a large number of previous experimental data Two functions of , s is a certain similarity.
步骤S13，根据概率判断两个文件夹中图片是否为同一人物；当概率越大时，则为同一个人物几率越大，当概率较小时，则为同一个人物几率较小。可以根据前面求得的概率来表示这两个文件夹之间图片为同一人物的可能性。当该概率越大，则表明这两个文件夹之间图片为同一人物的可能性越大。反之则当该概率越小，则表明这两个文件夹之间图片为同一人物的可能性越小。Step S13, judge whether the pictures in the two folders are the same person according to the probability; when the probability is higher, the probability of being the same person is higher, and when the probability is smaller, the probability of being the same person is small. The probability that the pictures between the two folders are the same person can be expressed according to the probability obtained above. The greater the probability, the greater the possibility that the pictures between the two folders are the same person. On the contrary, when the probability is smaller, it indicates that the possibility that the pictures between the two folders are the same person is smaller.
如图2所示，本发明的另一目的在于提供一种样本筛选方法的流程图，其详述如下：As shown in Figure 2, another object of the present invention is to provide a flow chart of a sample screening method, which is described in detail as follows:
步骤S21，依次计算两个文件夹中所有图片的相似值；对于任意两个待比较的文件夹，可以依次计算一个文件夹中图片与另一文件夹中所有图片的相似值；其中相似值为sAB，其公式为其中，m、n为两个文件夹中图片的张数，为A中图片ai和B中图片bj的相似值。假设待比较文件夹A和B中分别有m，n张图片，为A中图片ai和B中图片bj的相似值，AB文件夹的相似值为： Step S21, calculate the similarity value of all the pictures in the two folders in turn; for any two folders to be compared, you can calculate the similarity value of the pictures in one folder and all the pictures in the other folder in turn; where the similarity value is s AB , whose formula is Among them, m and n are the number of pictures in the two folders, is the similarity value of picture a i in A and picture b j in B. Assuming that there are m and n pictures in the folders A and B to be compared respectively, is the similarity value of the picture a i in A and the picture b j in B, and the similarity value of the AB folder is:
步骤S22，根据所得到的所有相似值求和；将计算所得的所有相似值进行求和，其中，在该式中分子为相似值求和，该式的值就是平均值，也就是AB文件夹的相似值。Step S22, summing according to all obtained similarity values; summing up all calculated similarity values, wherein, in the In the formula, the numerator is the sum of similar values, and the value of this formula is the average value, which is the similar value of the AB folder.
步骤S23，根据相似值求得的和计算出平均值；利用相似值所求取到的和，进而求出平均值。Step S23 , calculating the average value based on the obtained sum of the similar values; using the obtained sum of the similar values to further obtain the average value.
步骤S24，根据得到的平均值判断两个文件夹中图片是否为同一人，当平均值越高时，则两个文件夹中的图片为同一人物的几率越大，当平均值越低时，则两个文件夹中的图片为同一人物的几率越小。利用求得到的平均值可以用于衡量两个文件夹之间图片为同一人物的可能性大小，当平均值越高时，则表明这两个文件夹之间图片为同一人物的可能性越大，当平均值越低时，则两个文件夹中的图片为同一人物的几率越小。Step S24, judging whether the pictures in the two folders are the same person according to the obtained average value, when the average value is higher, the probability that the pictures in the two folders are the same person is greater; when the average value is lower, Then the probability that the pictures in the two folders are the same person is smaller. The average value obtained can be used to measure the possibility of the same person in the picture between the two folders. The higher the average value, the greater the possibility of the same person in the picture between the two folders. , when the average value is lower, the probability that the pictures in the two folders are the same person is smaller.
如图3所示，本发明的另一目的在于提供一种样本筛选方法的流程图，其详述如下：As shown in Figure 3, another object of the present invention is to provide a flow chart of a sample screening method, which is described in detail as follows:
步骤S31，依次计算两个带比较文件夹中图片为同一人物的概率；对于任意两个待比较文件夹，可以用两个文件夹之间图片为同一人物的概率来衡量两个文件夹之间图片为同一人物的可能性。根据局部人工抽样统计得到在一个很大的图片集中两张图片为同一人物的概率。根据大量实验统计可以得到，任意两张图片为相同人物且相似度为s时的概率以及任意两张图片为不同人物且相似度为s的概率。根据以上两个概率可以求出任意两张图片为同一人物的概率，可以依次计算待比较文件夹中一个文件夹中图片与另一文件夹中所有图片的为同一人物的概率，根据局部人工抽样筛选之后的数据统计与计算两张图片为同一人物的概率为Psc，其公式为：N为图片集中的图片张数，S为相同任务图片对数。在两张图片为相同人物且相似度为s的概率为Pt(s)，其公式为两张图片是不同人物且相似度为s的概率为Pf(s)，其公式为N为图片集中的图片张数，S为相同任务图片对数，TS为相同人物且相似度为s的图片对数，FS为不是同一人物且相似度为s的图片对数。Step S31, sequentially calculate the probability that the picture in the two folders with comparison is the same person; for any two folders to be compared, the probability of the picture between the two folders being the same person can be used to measure the difference between the two folders. Possibility that the pictures are of the same person. According to local artificial sampling statistics, the probability that two pictures in a large picture set are the same person is obtained. According to a large number of experimental statistics, the probability that any two pictures are the same person and the similarity is s, and the probability that any two pictures are different people and the similarity is s. According to the above two probabilities, the probability that any two pictures are the same person can be calculated, and the probability that the pictures in one folder in the folder to be compared and all the pictures in the other folder are the same person can be calculated in turn. According to local manual sampling After screening, the statistics and calculation of the probability that the two pictures are the same person is P sc , and its formula is: N is the number of pictures in the picture set, and S is the number of pairs of pictures of the same task. The probability that two pictures are the same person and the similarity is s is P t (s), and its formula is The probability that two pictures are of different characters and the similarity is s is P f (s), and its formula is N is the number of pictures in the picture set, S is the logarithm of pictures of the same task, TS is the logarithm of pictures of the same person with a similarity of s, and FS is the logarithm of pictures of different characters with a similarity of s.
根据局部人工抽样筛选之后的数据统计与计算，可以得到，在一个很大的图片集中，两张图片为同一人物的概率为Psc，两张图片为不同人物的概率是为1-Psc。According to the data statistics and calculation after local manual sampling and screening, it can be obtained that in a large picture set, the probability of two pictures being the same person is P sc , and the probability of two pictures being different people is 1-P sc .
假设实验图片集中共有N张图片，其中相同人物图片对数为S，可以得知：Assuming that there are N pictures in the experimental picture set, and the logarithm of the pictures of the same person is S, it can be known that:
(2)根据大量实验可以得出，两张图片为相同人物且相似度为s的概率为Pt(s)，两张图片是不同人物且相似度为s的概率为Pf(s)。(2) According to a large number of experiments, it can be concluded that the probability that two pictures are the same person and the similarity is s is P t (s), and the probability that two pictures are different people and the similarity is s is P f (s).
假设实验图片集中共有N张图片，统计得到相同人物图片对数S，是相同人物且相似度为s的图片对数为TS，不是同一人物且相似度为s的图片对数为FS，则可以计算：Assuming that there are N pictures in the experimental picture set, the logarithm of the pictures of the same person is statistically S, the logarithm of the pictures of the same person with a similarity of s is TS, and the logarithm of pictures of not the same person with a similarity of s is FS, then you can calculate:
计算：calculate:
(1)计算A文件夹中图片ai和B文件夹中图片bj的相似度，记为。根据ai和bj的相似度，计算出ai和bj为同一人物的概率：(1) Calculate the similarity between the picture a i in folder A and the picture b j in folder B, denoted as . According to the similarity of a i and b j , calculate the probability that a i and b j are the same person:
(2)假设A和B文件夹中分别有m，n张图片，A和B文件夹为同一人物的概率：(2) Assuming that there are m and n pictures in the A and B folders respectively, the probability that the A and B folders are the same person:
即 which is
步骤S32，将所求得的所有概率值相加求出平均值；将步骤S31中得到的所有的概率值相加之后求出其平均值。Step S32, adding all the obtained probability values to obtain the average value; adding all the probability values obtained in the step S31 to obtain the average value.
步骤S33，根据求得的平均概率值判断两文件夹中图片是否为同一人物，当平均概率值越大，则两个文件夹中图片为同一个人的几率越大，当平均概率值越小，则两个文件夹中图片为同一个人的几率越小。利用步骤S32中求出的平均概率，来衡量这两个文件夹中图片为同一人物的可能性，其中求出的平均概率越大，则表明这两个文件夹之间图片为同一人物的可能性越大。反之，其中求出的平均概率越小，则表明这两个文件夹之间图片为同一人物的可能性越小。Step S33, judging whether the pictures in the two folders are the same person according to the obtained average probability value, when the average probability value is greater, the probability that the pictures in the two folders are the same person is greater, and when the average probability value is smaller, Then the probability that the pictures in the two folders are the same person is smaller. Utilize the average probability obtained in step S32 to measure the possibility that the pictures in the two folders are the same person, and the greater the average probability obtained, it shows that the pictures between the two folders are the possibility of the same person The greater the sex. Conversely, the smaller the calculated average probability, the smaller the possibility that the pictures between the two folders are the same person.
当两张图片的比对分数s与P(s)之间满足线性关系时，易知方案一公式(9)和方案三公式(7)是等价的，但是实验结果表明s与P(s)之间并不满足线性关系，在这种情况下，公式(9)的计算结果并不准确，公式(7)更能准确的计算出两个文件夹为同一人物的概率。所以方案三比方案一的计算结果更准确可靠。When the comparison scores s and P(s) of the two pictures satisfy a linear relationship, it is easy to know that formula (9) of scheme 1 and formula (7) of scheme 3 are equivalent, but the experimental results show that s and P(s ) does not satisfy the linear relationship. In this case, the calculation result of formula (9) is not accurate, and formula (7) can more accurately calculate the probability that two folders are the same person. Therefore, the calculation result of scheme three is more accurate and reliable than scheme one.
方案二用两个文件夹图片相似度的平均值savg表示两文件夹图片为同一人物的可能性，方案一用savg对应的概率表示两文件夹图片为同一人物的可能性。假设A文件夹和B，C文件夹相似度的平均值分别为sAB，sAC，对应的概率分别为很明显可函数P(s)为递增函数，即若sAB≥sAC则所以可以知道方案二和方案一的衡量方式是等价，其衡量结果也是一致的。Scheme 2 uses the average s avg of the similarity of two folder pictures to indicate the possibility that the two folder pictures are the same person, and scheme 1 uses the probability corresponding to s avg Indicates the possibility that the pictures in the two folders are the same person. Assuming that the average similarity between folder A and folder B and C is s AB , s AC , the corresponding probabilities are It is obvious that the function P(s) is an increasing function, that is, if s AB ≥ s AC then Therefore, it can be known that the measurement methods of Scheme 2 and Scheme 1 are equivalent, and the measurement results are also consistent.
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. within range.
Claims (9)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611270693.1A CN106845532B (en) | 2016-12-30 | 2016-12-30 | A kind of screening sample method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611270693.1A CN106845532B (en) | 2016-12-30 | 2016-12-30 | A kind of screening sample method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106845532A true CN106845532A (en) | 2017-06-13 |
| CN106845532B CN106845532B (en) | 2018-07-20 |
Family
ID=59116986
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611270693.1A Active CN106845532B (en) | 2016-12-30 | 2016-12-30 | A kind of screening sample method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106845532B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109241311A (en) * | 2018-08-09 | 2019-01-18 | 上海常仁信息科技有限公司 | A kind of view field's automatic screening system and method based on robot |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6272249B1 (en) * | 1997-03-21 | 2001-08-07 | Sharp Kabushiki Kaisha | Image processing device |
| CN101546424A (en) * | 2008-03-24 | 2009-09-30 | 富士通株式会社 | Method and device for processing image and watermark detection system |
| CN101888469A (en) * | 2009-05-13 | 2010-11-17 | 富士通株式会社 | Image processing method and image processing device |
| CN102509303A (en) * | 2011-11-22 | 2012-06-20 | 鲁东大学 | Binarization image registration method based on improved structural similarity |
-
2016
- 2016-12-30 CN CN201611270693.1A patent/CN106845532B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6272249B1 (en) * | 1997-03-21 | 2001-08-07 | Sharp Kabushiki Kaisha | Image processing device |
| CN101546424A (en) * | 2008-03-24 | 2009-09-30 | 富士通株式会社 | Method and device for processing image and watermark detection system |
| CN101888469A (en) * | 2009-05-13 | 2010-11-17 | 富士通株式会社 | Image processing method and image processing device |
| CN102509303A (en) * | 2011-11-22 | 2012-06-20 | 鲁东大学 | Binarization image registration method based on improved structural similarity |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109241311A (en) * | 2018-08-09 | 2019-01-18 | 上海常仁信息科技有限公司 | A kind of view field's automatic screening system and method based on robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106845532B (en) | 2018-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210374386A1 (en) | Entity recognition from an image | |
| CN106295482B (en) | A kind of update method and device of face database | |
| CN110457680B (en) | Entity disambiguation method, device, computer equipment and storage medium | |
| WO2020108075A1 (en) | Two-stage pedestrian search method combining face and appearance | |
| CN110222560B (en) | Text person searching method embedded with similarity loss function | |
| CN109960727B (en) | Personal privacy information automatic detection method and system for unstructured text | |
| WO2015090215A1 (en) | Voice data recognition method and device for distinguishing regional accent, and server | |
| CN113505824B (en) | Judgment updating method and device and face card punching system | |
| CN105006230A (en) | Voice sensitive information detecting and filtering method based on unspecified people | |
| CN109829071B (en) | Face image searching method, server, computer device and storage medium | |
| CN111046879A (en) | Certificate image classification method and device, computer equipment and readable storage medium | |
| CN104317784A (en) | Cross-platform user identification method and cross-platform user identification system | |
| CN104615715A (en) | Social network event analyzing method and system based on geographic positions | |
| CN110458091A (en) | Recognition of face 1 based on position screening is than N algorithm optimization method | |
| CN108595422A (en) | A method of the bad multimedia message of filtering | |
| CN106776543B (en) | New word discovery method, apparatus, terminal and server | |
| CN106570367B (en) | Method for identifying ID and device based on keyboard operation | |
| CN109344233B (en) | A method of Chinese name recognition | |
| CN110381218A (en) | A kind of method and device identifying telephone fraud clique | |
| CN108170716B (en) | Text duplicate checking method based on human vision | |
| CN115062186B (en) | Video content retrieval method, device, equipment and storage medium | |
| RU2612608C2 (en) | Social circle formation system and method and computer data carrier | |
| CN115730073A (en) | Text processing method, device and storage medium | |
| WO2023035362A1 (en) | Polluted sample data detecting method and apparatus for model training | |
| CN106845532A (en) | A kind of screening sample method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |