-1

使用下表我需要找到重复的图像记录。该查询用于查找记录但不准确的结果。在获取重复的图像记录之后，我们只想获取那些 FM_NAME_EN、LASTNAME_EN、RLN_TYPE、RLN_FM_NM_EN、RLN_L_NM_EN、GENDER、AGE 和 PHOTO 相同的记录。使用的表具有如下所示的列

 CREATE TABLE [dbo].[ERoll](
    [AC_NO] [int] NULL,
    [PART_NO] [int] NULL,
    [SLNOINPART] [int] NULL,
    [FM_NAME_EN] [nvarchar](250) NULL,
    [LASTNAME_EN] [nvarchar](250) NULL,
    [RLN_TYPE] [nvarchar](1) NULL,
    [RLN_FM_NM_EN] [nvarchar](250) NULL,
    [RLN_L_NM_EN] [nvarchar](250) NULL,
    [EPIC_NO] [nvarchar](25) NULL,
    [GENDER] [nvarchar](1) NULL,
    [AGE] [int] NULL,
    [PHOTO] [image] NULL
);

在上表中，一些具有重复图像的记录。那么我们如何才能找到重复的图像记录呢？

只有 50% 的记录通过以下查询正确出现：

SELECT e.AC_NO, e.PART_NO, e.SLNOINPART, e.FM_NAME_EN, e.LASTNAME_EN, 
    e.RLN_FM_NM_EN, e.RLN_L_NM_EN, e.RLN_TYPE, e.GENDER, e.AGE
FROM S21_EROLL_MANAGEMENT_PROD.dbo.ERoll e
JOIN (
    select FM_NAME_EN, LASTNAME_EN, RLN_FM_NM_EN, RLN_L_NM_EN, RLN_TYPE, GENDER, AGE, hashbytes('md5', cast([PHOTO] as varbinary)) PHOTO ,count(*) PHOTO_COUNT
    from S21_EROLL_MANAGEMENT_PROD.dbo.ERoll
    group by FM_NAME_EN, LASTNAME_EN, RLN_FM_NM_EN, RLN_L_NM_EN, RLN_TYPE, GENDER, AGE, hashbytes('md5', cast([PHOTO] as varbinary)) 
    having count(*) > 1
) d ON e.FM_NAME_EN = d.FM_NAME_EN and e.LASTNAME_EN = d.LASTNAME_EN and e.RLN_FM_NM_EN=d.RLN_FM_NM_EN 
    and e.RLN_L_NM_EN = d.RLN_L_NM_EN and e.RLN_TYPE = d.RLN_TYPE and e.GENDER = d.GENDER and e.AGE = d.AGE
order by AC_NO, PART_NO, SLNOINPART
4

3 回答 3

2

使用 GROUP BY、HAVING、COUNT() ：

select hashbytes('md5', cast([imagecolumn] as varbinary(max))),count(*)
from yourtable
group by hashbytes('md5', cast([imagecolumn] as varbinary(max)))
having count(*) > 1
于 2021-02-11T09:27:59.987 回答
1

我想你可以尝试找到这样的重复照片

SELECT HASHBYTES('SHA2_256', cast(photo as varbinary(max))) from dbo.ERoll
于 2021-02-11T09:59:33.030 回答
0

使用窗口函数：

select e.*
from (select e.*
             count(*) over (partition by FM_NAME_EN, LASTNAME_EN, RLN_FM_NM_EN, RLN_L_NM_EN, RLN_TYPE, GENDER, AGE, hashbytes('md5', cast([PHOTO] as varbinary))) as cnt
      from S21_EROLL_MANAGEMENT_PROD.dbo.ERoll e
     ) e
where cnt > 1;
于 2021-02-11T12:23:41.360 回答