您可以使用NOT EXISTS

var query = dbo.Table
    .Where(t => !dbo.Table.Any(t2 => t.Image = t.Image && t.Perimeter != t2.Perimeter));

您可以很容易地将其修改为只选择图像部分。（）”组计算，那么您将希望查看.SelectMany（）LINQ方法.这允许您“将划分到组中的数据重新组合在一起.”虽然您的需要是只返回“每个组中的一个，”可以在SSDT 2015的“C#交互式窗口”中运行该程序：

struct imagePerimeter { //this might be whatever object type it is for you...
    public string Image { get; set; } //a,b,b,d,e
    public int Perimeter { get; set; } //1,1,2,3,1
}
Func<string, int, imagePerimeter> newIP = (i, p) => new imagePerimeter() { Image = i, Perimeter = p };
List<imagePerimeter> results = new List<imagePerimeter>() { {newIP("a",1) }
    ,{newIP("b",1) }
    ,{newIP("b",2) }
    ,{newIP("d",3) }
    ,{newIP("e",1) } };

Func<imagePerimeter, string> ipImage = (ip) => ip.Image; //the Func's "ipImage" and "newIP" could just be inlined into LINQ, but it helps to see and debug at times IMO. 
var imagesWithOnePerimeter = results.GroupBy<imagePerimeter, string>(ipImage) //even in SQL, the "GROUP BY" conceptually comes first, in LINQ, it comes first in code too!
    .Select(grp => new { Image = grp.Key, PerimeterCount = grp.Count(), Details = grp }) //there's probably a more technical term, but notice how we "carry forward" the original reference to [grp]
    .Where(subTotals => subTotals.PerimeterCount == 1)
    .SelectMany(filtered => filtered.Details.AsEnumerable())
    .ToList();