Рангли тасвирларни қайта ишлашнинг ўзига хос хусусиятлари
- № 4 (10) 2019
Страницы:
103
–
106
Язык: узбекский
Аннотация
Ушбу мақолада тасвирларни табиий рангларда ёки оддий рангли тасвирларда қайта ишлаш вақтида қўлланиладиган усуллар кўриб чиқилди. Кўриб чиқилаётган усуллар, тасвирларни қайта ишлашнинг турли масалаларини ечишда қандай муносабатда бўлишни кўрсатиб беришда хизмат қилади. Рангли моделнинг сегментациясида янада яхши натижа RGB ранг фазосида ишлаганда эришилади. Таклиф этилаётган усул, етарли даража тушунарли. Тасаввур қилайлик, бизнинг вазифамиз ранги қайсидир аниқланган оралиқда ётган RGB-тасвирда объектларни сегментацияси ҳисобланади. Бизни қизиқтирган рангларга нисбатан репрезентатив бўлган ранг фазосида векторларнинг қайсидир танлови учун ажратиб кўрсатиш мақсадида рангнинг “ўртача” баҳосини оламиз. RGB-фазодаги а вектор, ушбу ўртача рангни белгилайди. Сегментация масаласи берилган тасвирнинг ҳар бир пикселини, унинг ранги берилган оралиқда ёки йўқлигига мос равишда синфлаштиришдан иборат. Бундай таққослашни амалга ошириш учун ранг фазосида ўхшашликнинг қайсидир ўлчовига эга бўлиш керак. Ўлчовнинг энг оддийси бўлиб, Евклид ўсимлиги ҳисобланади. Бундай солиштиришни амалга ошириш учун ранг фазосида ўшашликнинг қандайдир ўлчовига эга бўлиш керак. Ушбу ўлчовнинг энг оддийси бўлиб, Евклид масофа ҳисобланади.
This article explores the methods used to process images in natural colors or in simple color images. The methods under consideration will serve to illustrate how we should address different problems of image processing. The best results in the color model segmentation are achieved when working in the RGB color space. The proposed method is suf-ficiently clear. Let us assume that our task is to segment objects in an RGB image with a certain color range. We get the «aver-age» color rating to distinguish some vectors in the color space that are representative of the colors we are interested in. The vector a in the RGB space defines this average color. The issue of segmentation is to classify each pixel of a given image ac-cording to whether or not its color is within a given range. To make such a comparison, it is necessary to have some measure of similarity in the color space. The simplest measure is the Eu-clidean plant. To make such a comparison, it is necessary to have some measure of similarity in the color space. The sim-plest of these measurements is Euclidean distance.
This article explores the methods used to process images in natural colors or in simple color images. The methods under consideration will serve to illustrate how we should address different problems of image processing. The best results in the color model segmentation are achieved when working in the RGB color space. The proposed method is suf-ficiently clear. Let us assume that our task is to segment objects in an RGB image with a certain color range. We get the «aver-age» color rating to distinguish some vectors in the color space that are representative of the colors we are interested in. The vector a in the RGB space defines this average color. The issue of segmentation is to classify each pixel of a given image ac-cording to whether or not its color is within a given range. To make such a comparison, it is necessary to have some measure of similarity in the color space. The simplest measure is the Eu-clidean plant. To make such a comparison, it is necessary to have some measure of similarity in the color space. The sim-plest of these measurements is Euclidean distance.