ABSTRAKSI: Pada tugas akhir ini image retrieval khususnya Content-Based Image Retrieval (CBIR) dikembangkan menggunakan metode Stochastic Paintbrush Transformation (SPT), yaitu suatu algoritma baru untuk transformasi citra kedalam representasi lukisan (paintbrush). Algoritma SPT dipilih karena metode ini bersifat otomatis, mampu menginterpretasikan citra, dan mampu menangkap konten visual dari suatu citra. SPT melakukan ekstraksi citra didasarkan pada representasi lukisan dari citra asli yaitu melalui fitur paintbrush stroke parameter yang terdiri dari bentuk dan ukuran brush, warna, orientasi dan lokasi. TA ini dibagi menjadi dua subsistem. Pertama subsistem penyimpanan data, yaitu melakukan ekstraksi fitur paintbrush citra dan menyimpannya kedalam database. Kedua subsistem pencarian citra query, hasil ekstraksi fitur paintbrush citra query dibandingkan dengan fitur paintbrush semua citra database, hasil akhir berupa sejumlah N citra database yang memiliki beberapa tingkat kemiripan dengan citra query. Pengukuran kemiripan citra didapat dari proses similarity berdasarkan parameter warna, orientasi dan lokasi brush. Hasil analisis yang didapat adalah penggunaan metode SPT dalam sistem CBIR menghasilkan performansi sistem yang baik tetapi membutuhkan waktu transformasi citra yang cukup lama.Kata Kunci : image retrieval, Content-Based Image Retrieval (CBIR), Stochastic Paintbrush Transformation (SPT), paintbrush stroke parameter, brush, orientasi, similarity.ABSTRACT: This final project describes a new method to develop image retrieval, especially in Content-Based Image Retrieval (CBIR), called Stochastic Paintbrush Transformation (SPT). It is an algorithm to transform the image into a painting representation (paintbrush). SPT is chosen because it is completely automatic and it also can provide an interpretation of an image and capture its visual content. The image extraction on SPT is based on painting representation of the original image by using paintbrush stroke parameter features which includes shape, size, color, orientation, and location of the brush. This final project is divided into two subsystems. The first is data storage subsystem, which extracts the image paintbrush stroke features and store them into the database. Second, the query image retrieval subsystem matches the resulting paintbrush features of the query image and the resulting paintbrush features of all images from the database. This process results in N number of the database images which have some similarity levels to the query image. The similarity measurement of an image is obtained from similarity process based on color, orientation, and location parameter of the brush. Experimental analysis results show that the SPT method in CBIR system has a good performance system but not so good in image time transformation.Keyword: image retrieval, Content-Based Image Retrieval (CBIR), Stochastic Paintbrush Transformation (SPT), paintbrush stroke parameter, brush, orientation, similarity.