Pillow 사용법 - Basic 03 - Image Enhance

ImageEnhance 모듈을 통해 Image Enhancement 기능을 지원.

 

참고로, torchvision.transforms 의
image adjustment와 알고리즘이 유사함

 

api document: https://pillow.readthedocs.io/en/stable/reference/ImageEnhance.html

 

관련 gist

https://gist.github.com/dsaint31x/2973d9478c5cd9bb188d66095539f02d

 

---

Image Enhancement란?

• 영상에서 원하는 정보를  사람이 더 쉽게 해석할 수 있도록,
• Brightness, Contrast, Spacial Characteristics 등을 개선해 
• 주관적 화질을 향상 시키는 과정.

---

1. 사용방법

1. ImageEnhnace.XXX(img)로 대상 img에 대한 Enhance 객체를 생성.
2. XXX.enhance(factor)를 호출하여 Image Enhancement 수행 결과인 새로운 Image 객체 반환.

factor는 1.0을 기준으로 하고, 1.0보다 크면 강화, 작으면 약화

---

2. 지원되는 Enhance 종류

• ImageEnhance.Brightness : 밝기
• ImageEnhance.Contrast : 대비
• ImageEnhance.Color : 색의 채도(Saturation)
• ImageEnhance.Sharpness : 선명도

---

2-1. 밝기 (Brightness)

$$I_{\text{out}} = I_{\text{in}} \times \text{factor}$$

• factor=1.0: 원본 brightenss
• factor<1.0: 어둡게
• factor>1.0: 밝게

enhancer_bright = ImageEnhance.Brightness(img)
img_bright = enhancer_bright.enhance(1.5)  # 1.5배 밝게

https://ds31x.tistory.com/451#12.-adjust_brightnessinpt,-brightness_factor

[PyTorch] Color-Torchvision.transforms.v2.functional

---

2-2. 대비 (Contrast)

$$I_{\text{out}} = (I_{\text{in}} - \mu) \times \text{factor} + \mu$$

• $\mu$는 대상 img의 평균 brightness
• factor=1.0: 원본 contrast
• factor<1.0: contrast 낮추기
• factor>1.0: contrast 증가

enhancer_contrast = ImageEnhance.Contrast(img)
img_contrast = enhancer_contrast.enhance(0.8)  # 대비 줄이기

 

https://ds31x.tistory.com/451#11.-adjust_contrastinpt,-contrast_factor

[PyTorch] Color-Torchvision.transforms.v2.functional

 

---

2-3. Saturation

$$I_{\text{out}} = \mu_{\text{gray}} + (I_{\text{in}} - \mu_{\text{gray}}) \times \text{factor}$$

• $\mu_{\text{gray}}$는 픽셀의 그레이스케일 값 (ex: 0.299R + 0.587G + 0.114B)
• factor=0: 흑백으로 변환
• factor=1: 원본 색감
• factor>1: 채도 증가

enhancer_color = ImageEnhance.Color(img)
img_color = enhancer_color.enhance(2.0)  # saturation 두 배로 진하게

https://ds31x.tistory.com/451#13.-adjust_saturationinpt,-saturation_factor

[PyTorch] Color-Torchvision.transforms.v2.functional

---

2-4. Sharpness

$$I_{\text{out}} = I_{\text{in}} \times \text{factor} + I_{\text{blur}} \times (1 - \text{factor})$$

• $I_{\text{blur}}$ : blurred image
• factor=0: blurring
• factor=1: 원본
• factor>1: sharpening

enhancer_sharpness = ImageEnhance.Sharpness(img)
img_sharp = enhancer_sharpness.enhance(5.)  # sharpening

https://ds31x.tistory.com/451#10.-adjust_sharpnessinpt,-sharpness_factor

[PyTorch] Color-Torchvision.transforms.v2.functional

https://ds31x.tistory.com/453#randomadjustsharpness

[PyTorch] Photometric-torchvision.transforms.v2

---

같이 보면 좋은 자료들

2025.06.30 - [Python] - Pillow 사용법 - Basic 01

Pillow 사용법 - Basic 01

 

2024.06.03 - [Python] - [Python] PIL, Pillow, OpenCV, and Scikit-image

[Python] PIL, Pillow, OpenCV, and Scikit-image

 

https://ds31x.tistory.com/451

[PyTorch] Color-Torchvision.transforms.v2.functional