PostgreSQL 在視訊、圖片去重，影象搜尋業務中的應用圖片搜尋PostgreSQL的影象搜尋外掛使用了非常主流的Haar wavelet技術對影象進行變換後儲存gist 索引方法(支援pattern和signature型別), 以及KNN操作符，可以用於搜尋相似度1,建立圖片表create table image (id serial, data bytea);2,匯入圖片到資料庫insert into image(data) select pg_read_binary_file('檔案全路徑');錯誤：ERROR: absolute path not allowedPGDATA 為/var/lib/pgsql/9.6/data/，在vi /usr/lib/systemd/system/postgresql-9.5.service裡面有拷貝檔案到該目錄後insert into image(data) select pg_read_binary_file('./2/1K7A9699.JPG');3,將圖片轉換成 patten 和 signature,很耗時CREATE TABLE pat AS (SELECT id, shuffle_pattern(pattern) AS pattern, pattern2signature(pattern) AS signature FROM ( SELECT id, jpeg2pattern(data) AS pattern FROM image) x );4,建立索引ALTER TABLE pat ADD PRIMARY KEY (id);CREATE INDEX pat_signature_idx ON pat USING gist (signature);5,相似度排行SELECT id, smlr FROM ( SELECT id, pattern <-> (SELECT pattern FROM pat WHERE id = 74) AS smlr FROM pat WHERE id <> 74 ORDER BY signature <-> (SELECT signature FROM pat WHERE id = 74) LIMIT 100) x ORDER BY x.smlr ASC LIMIT 10;視訊去重抽取視訊中的關鍵幀1,建立圖片表2,匯入圖片到資料庫3,將圖片轉換成 patten 和 signatureCREATE TABLE pat AS ( SELECT id, movie_id, shuffle_pattern(pattern) AS pattern, pattern2signature(pattern) AS signature FROM ( SELECT id, movie_id, jpeg2pattern(data) AS pattern FROM image ) x );create table targer_table as select * from source_table是會複製表結構+表資料，SELECT name FROM (SELECT name FROM agentinformation) 這種是不對的You need to alias the subquery.SELECT name FROM (SELECT name FROM agentinformation) a or to be more explicitSELECT a.name FROM (SELECT name FROM agentinformation) a 即select x.id, x.movie_id...4.計算不同視訊的相似度select t1.movie_id, t1.id, t1.signature<->t2.signature from pat t1 join pat t2 on (t1.movie_id<>t2.movie_id) order by t1.signature<->t2.signature desc or select t1.movie_id, t1.id, t1.signature<->t2.signature from pat t1 join pat t2 on (t1.movie_id<>t2.movie_id) where t1.signature<->t2.signature > 0.9 order by t1.signature<->t2.signature desc 多上面紅色那一行，其它一樣報錯：ERROR: function jpeg2pattern(bytea) does not exist這個是因為擴充套件imgsmlr安裝不正確\dx命令檢視所有已安裝的擴充套件模組imgsmlr$ cd imgsmlr$ export PGHOME=/usr/pgsql-9.6$ export PATH=$PGHOME/bin:$PATH:.$ make USE_PGXS=1$ make USE_PGXS=1 installpsql -U postgres -d exampledb -h 127.0.0.1 -p 5432then exeute the sql.create extension imgsmlr;檢視下\dx命令檢視所有已安裝的擴充套件模組然後執行CREATE TABLE pat AS (SELECT id, shuffle_pattern(pattern) AS pattern, pattern2signature(pattern) AS signature FROM ( SELECT id, jpeg2pattern(data) AS pattern FROM image) x );記得結束加分號The idea is to find top N similar images by signature using GiST index. Then find top n (n < N) similar images by pattern from top N similar images by signature.參考下一步需要測試增量新增圖片