How to get only one record for each duplicate rows of the id in oracle?

SELECT  group_id, image, image_id
FROM    a_table
WHERE   (group_id, image_id) IN
        (
            SELECT  group_id, MIN(image_id)
            FROM    a_table
            GROUP   BY
                    group_id
        )
;

There are no standard aggregate functions in Oracle that would work with BLOBs, so GROUP BY solutions won't work.

Try this one based on ROW_NUMBER() in a sub-query.

SELECT inn.group_id, inn.image, inn.image_id
FROM
(
    SELECT t.group_id, t.image, t.image_id, 
        ROW_NUMBER() OVER (PARTITION BY t.group_id ORDER BY t.image_id) num
    FROM theTable t
) inn
WHERE inn.num = 1;

The above should return the first (based on image_id) row for each group.

SQL Fiddle