Ƙididdigar Launi don Kayan Kawa, Fasaha, da Zane - Bincike da Nazari

Teburin Abubuwan Ciki

1. Gabatarwa

Cirewar launi ta atomatik ta sami kulawa sosai a cikin aikin fasaha na dijital da aikace-aikacen zane, musamman a cikin kayan kawa, kayan ado, da tsarin ba da shawara. Hotunan dijital suna aiki azaman farkon hanyar wakiltar abubuwa na ainihi, amma kalubale irin su lalata launi da kuma babban bakan launi sun sa ƙididdigar launi ta atomatik ta zama matsala mai sarƙaƙiya.

Muhimmin mataki na farko a cikin cirewar launi daidai shine tantance adadin launukan da ke cikin wani wuri ko abu. Duk da yake wannan na iya zama a bayyane, yana gabatar da manyan kalubale ko da ga fahimtar ɗan adam. Bincike ya nuna cewa ƙididdigar launi na buƙatar hanyoyin fahimi guda biyu: gane launi yayin da ake watsar da bayanan sararin samaniya, da kuma hankalin ƙidaya.

2. Hanyoyi

2.1 Hanyar Tarihin Tarawa da Aka Tsara

Sabuwar hanyar tarihin launi ta tarawa tana nazarin tsarin rarraba launi don tantance mafi kyawun adadin launuka. Hanyar ta ƙunshi:

• Canza hotunan RGB zuwa wuraren launi masu dacewa
• Ƙididdige tarihin tarawa na kowane tashoshi
• Gano wuraren jujjuyawar da ke wakiltar launuka daban-daban
• Aiwatar da dabarun ƙetare don raba launi

2.2 Tsarin Gaurayawan Gaussian (GMM)

GMM yana ƙirƙira rarraba launi ta amfani da aikin yuwuwar yawa:

$p(x) = \sum_{i=1}^{K} \phi_i \mathcal{N}(x|\mu_i,\Sigma_i)$

inda $\mathcal{N}(x|\mu_i,\Sigma_i) = \frac{1}{\sqrt{(2\pi)^K|\Sigma_i|}} \exp\left(-(x-\mu_i)^T\Sigma_i^{-1}(x-\mu_i)\right)$

kuma $K$ yana nuna adadin launuka, $\phi_i$ yana wakiltar ma'aunin gaurayawa, $\mu_i$ yana nufin, da $\Sigma_i$ matrices na haɗin kai.

2.3 Rarrabuwa ta K-Means

Rarrabuwar K-means ta gargajiya tare da cikakken bincike don mafi kyawun ƙimar K ta amfani da hanyar gwiwar hannu da binciken silhouette.

2.4 Hanyoyin Zurfin Koyo

Cibiyoyin sadarwa na Convolutional da aka horar don ƙididdigar launi, gami da ResNet da gine-ginen al'ada waɗanda aka ƙera musamman don ayyukan nazarin launi.

3. Nazarin Rarraba Launi

Hotunan launi suna fama da murdiya daban-daban da suka haɗa da ingancin bugawa, haɗakar launi, lissafin hoto, yanayin haske, matsi na hoto, da halayen na'ura. Waɗannan abubuwan suna yin tasiri sosai ga bayyanar launi kuma suna shigar da amo a cikin hanyoyin nazarin launi.

Binciken ya ginu akan aikin da Al-Rawi da Joeran suka yi a baya wanda ya nuna cewa za a iya yin amfani da hotunan RGB masu yawa ta amfani da Tsarin Gaurayawan Gaussian azaman rarrabuwa na farko, yana ba da tushen ƙididdiga don nazarin launi a cikin wurare masu amo.

4. Sakamakon Gwaji

5. Aiwatar da Fasaha

import numpy as np
import cv2
from scipy.signal import find_peaks

def count_colors_cumulative_histogram(image_path, threshold=0.05):
    # Loda da shirya hoto
    image = cv2.imread(image_path)
    image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    
    # Canza zuwa sararin launi na HSV
    image_hsv = cv2.cvtColor(image, cv2.COLOR_BGR2HSV)
    
    # Ƙididdige tarihin tarawa na tashar hue
    hue_hist = cv2.calcHist([image_hsv], [0], None, [180], [0, 180])
    cumulative_hist = np.cumsum(hue_hist) / np.sum(hue_hist)
    
    # Nemo wuraren jujjuyawa
    derivatives = np.diff(cumulative_hist.flatten())
    peaks, _ = find_peaks(derivatives, height=threshold)
    
    # Adadin launuka daidai yake da manyan kololuwa + 1
    num_colors = len(peaks) + 1
    
    return num_colors

# Misalin amfani
color_count = count_colors_cumulative_histogram('fashion_image.jpg')
print(f"An gano {color_count} launuka daban-daban")

6. Aikace-aikace da Hanyoyin Gaba

Aikace-aikace na Yanzu

• Tsarin Ba da Shawara na Kaya: Ingantaccen shawarwarin samfura bisa launi
• Zanen Ciki: Cirewar fentin launi ta atomatik daga hotunan ƙwaƙwalwa
• Fasaha ta Dijital: Nazarin launi don abun da ke ciki na fasaha da canja wurin salo
• Kasuwancin E: Ingantaccen binciken samfura da tacewa ta halayen launi

Hanyoyin Bincike na Gaba

• Haɗa kai tare da gine-ginen transformer don ingantaccen fahimtar launi
• Ƙididdigar launi na ainihi don aikace-aikacen wayar hannu
• Daidaitawa ta yanki daban-daban don yanayin hoto daban-daban
• Hanyoyin nau'i-nau'i masu haɗa launi tare da nazarin laushi da tsari

7. Bayanan Kafa