Computer Vision Process (CBSE Class 12 Artificial Intelligence)
Class 12 · Artificial Intelligence
Computer Vision Process
Computer Vision is a systematic process that enables machines to understand visual information. Instead of recognizing an image instantly like humans, a computer follows a sequence of well-defined steps to convert raw image data into meaningful information.
Every Computer Vision application, whether it is Face Recognition, Medical Diagnosis, Traffic Monitoring, or Self-driving Cars, follows a similar workflow. Each stage transforms the image into a form that is easier for the Machine Learning model to understand and analyze.
The Computer Vision process combines image processing techniques with Artificial Intelligence algorithms to recognize patterns, classify objects, detect multiple objects, and support intelligent decision-making.
Learning Objectives
After studying this topic, you will be able to:
- Understand the complete Computer Vision process.
- Identify each stage of image processing.
- Explain the role of every stage.
- Understand how Machine Learning integrates with Computer Vision.
- Apply the Computer Vision process to real-world applications.
Complete Computer Vision Process
A Computer Vision system generally follows the steps shown below.
Capture Image
│
▼
Preprocess Image
│
▼
Extract Features
│
▼
Train / Apply Machine Learning Model
│
▼
Recognize Objects
│
▼
Make Decision
Step 1: Capture the Image
The first step is capturing an image or video using an imaging device.
The quality of the captured image directly affects the performance of the Computer Vision system.
Image Sources
- Digital Cameras
- Smartphones
- CCTV Cameras
- Medical Imaging Devices
- Satellites
- Drones
- Web Cameras
Step 2: Image Preprocessing
The captured image may contain unwanted information such as noise, shadows, poor lighting, or unnecessary background objects. Image preprocessing improves the image before analysis.
Common Image Preprocessing Operations
| Operation | Purpose |
|---|---|
| Noise Removal | Removes unwanted disturbances. |
| Image Resizing | Standardizes image dimensions. |
| Cropping | Removes unnecessary regions. |
| Brightness Adjustment | Improves visibility. |
| Contrast Enhancement | Highlights important details. |
| Grayscale Conversion | Reduces computational complexity. |
Step 3: Feature Extraction
After preprocessing, important visual characteristics are extracted from the image.
Instead of processing every pixel directly, the Computer Vision system identifies meaningful features that help distinguish one object from another.
Examples of Features
- Edges
- Corners
- Shapes
- Colours
- Texture
- Patterns
- Facial Landmarks
Step 4: Machine Learning Model
The extracted features are supplied to a trained Machine Learning model.
During training, the model learns patterns from thousands of labelled images. During prediction, it compares the extracted features with previously learned patterns and predicts the most likely object.
A Machine Learning model is a trained algorithm that recognizes patterns from image features and predicts the class of a new image.
Training Phase vs Prediction Phase
| Training Phase | Prediction Phase |
|---|---|
| Uses labelled images. | Uses unseen images. |
| Learns visual patterns. | Recognizes learned patterns. |
| Builds the model. | Uses the trained model. |
| Performed once or periodically. | Performed whenever a new image is received. |
Step 5: Object Recognition
The trained Machine Learning model predicts the object present in the image.
Depending on the application, the output may be:
- Image Classification
- Object Detection
- Face Recognition
- Text Recognition (OCR)
- Pose Estimation
- Image Segmentation
Example
Suppose the input image contains a dog.
Input Image
│
▼
Feature Extraction
│
▼
Machine Learning Model
│
▼
Prediction
Dog
Step 6: Decision Making
Once the object has been recognized, the Computer Vision system performs the required action.
Examples
| Recognized Object | Decision |
|---|---|
| Authorized Face | Unlock Phone |
| Pedestrian | Apply Vehicle Brakes |
| Cancer Cell | Alert Doctor |
| Fire | Trigger Alarm |
| Defective Product | Reject Product |
Workflow Diagram
Camera
│
▼
Capture Image
│
▼
Image Processing
│
▼
Feature Extraction
│
▼
Machine Learning
│
▼
Recognition
│
▼
Decision
Real-Life Example
Google Lens captures an image of a flower using the smartphone camera. The image is preprocessed to improve quality, important visual features such as petal shape and colour are extracted, and a trained Machine Learning model identifies the flower species. Finally, Google Lens displays the flower's name and related information to the user.
Think Like an AI Engineer
A smart dustbin automatically opens its lid when it detects a person's hand near the opening.
Explain how the Computer Vision process helps the dustbin perform this task.
Click to View Answer
The camera captures the image of the approaching hand. The image is preprocessed, important features are extracted, and a trained Machine Learning model recognizes the hand. Based on this prediction, the controller sends a signal to automatically open the dustbin lid.
Complete Example: Face Recognition System
Face Recognition is one of the most common applications of Computer Vision. It follows every stage of the Computer Vision process before identifying a person.
| Stage | Process |
|---|---|
| Image Acquisition | The camera captures the person's face. |
| Image Preprocessing | The image is resized, brightness is adjusted, and noise is removed. |
| Feature Extraction | Important facial features such as eyes, nose, mouth, and jawline are identified. |
| Machine Learning Model | The extracted features are compared with the trained facial database. |
| Recognition | The person's identity is predicted. |
| Decision Making | The phone unlocks if the face matches the stored user. |
Case Study
An airport installs an Artificial Intelligence-based security system to verify passengers before boarding.
As passengers approach the boarding gate, cameras capture their facial images. The images are preprocessed to improve quality, and important facial features are extracted. A trained Machine Learning model compares these features with the passport database.
If the passenger's identity matches the stored record, the boarding gate opens automatically. Otherwise, airport security personnel are alerted for manual verification.
This Computer Vision system improves security, reduces waiting time, and minimizes manual verification.
Advantages of the Computer Vision Process
- Processes images automatically without human intervention.
- Provides fast and accurate recognition.
- Reduces human errors.
- Supports real-time decision-making.
- Can process thousands of images in a short time.
- Improves automation across multiple industries.
- Works continuously without fatigue.
Applications Following the Computer Vision Process
| Application | Purpose |
|---|---|
| Face Recognition | Identity verification and authentication. |
| Medical Diagnosis | Detection of diseases from medical images. |
| Autonomous Vehicles | Detection of roads, vehicles, and pedestrians. |
| Traffic Monitoring | Vehicle counting and traffic analysis. |
| Retail Stores | Automatic product recognition and inventory management. |
| Agriculture | Crop disease detection and yield monitoring. |
| Manufacturing | Automatic defect detection and quality inspection. |
| Security Surveillance | Detection of suspicious activities and unauthorized access. |
Computer Vision Process at a Glance
| Step | Purpose |
|---|---|
| Image Acquisition | Capture an image. |
| Image Preprocessing | Improve image quality. |
| Feature Extraction | Identify important characteristics. |
| Machine Learning | Recognize learned patterns. |
| Object Recognition | Identify the object. |
| Decision Making | Perform the required action. |
Activity
Choose any one Computer Vision application such as Face Unlock, Google Lens, Self-driving Cars, or Medical Image Analysis. Draw the complete Computer Vision process and explain the role of each stage in that application.
Competency-Based Question
A smart classroom uses cameras to automatically record student attendance using face recognition technology.
Explain how each stage of the Computer Vision process helps the system identify students and mark attendance automatically.
Think Like an AI Engineer
A self-driving car captures an image of a traffic signal during heavy rain. Explain how image preprocessing and feature extraction help the vehicle recognize the traffic signal correctly.
Click to View Answer
Image preprocessing improves the image by reducing noise, adjusting brightness, and enhancing contrast. Feature extraction then identifies important characteristics such as the shape and colour of the traffic signal. These features are analyzed by the Machine Learning model, which correctly recognizes the signal despite poor weather conditions.
Common Beginner Mistakes
- Believing that Computer Vision directly recognizes images without processing.
- Ignoring the importance of image preprocessing.
- Confusing Feature Extraction with Machine Learning.
- Assuming that Machine Learning models work without training data.
- Thinking that all Computer Vision applications follow different workflows.
Quick Revision
- Computer Vision follows a systematic sequence of processing stages.
- Images are captured using cameras or sensors.
- Preprocessing improves image quality.
- Feature Extraction identifies important visual information.
- Machine Learning recognizes patterns learned during training.
- The final stage performs intelligent decision-making.
Memory Trick
C → P → F → M → R → D
Remember:
- C → Capture Image
- P → Preprocess Image
- F → Feature Extraction
- M → Machine Learning Model
- R → Recognition
- D → Decision Making
Exam Tips
- Memorize the stages of the Computer Vision process in the correct order.
- Understand the purpose of each stage.
- Practice explaining the workflow using real-life examples.
- Differentiate between Feature Extraction and Machine Learning.
- Use examples such as Face Recognition or Self-driving Cars in descriptive answers.
Frequently Asked Questions (FAQs)
1. Why is Image Preprocessing necessary?
Image Preprocessing improves image quality by removing noise, adjusting brightness, and preparing the image for further analysis.
2. What is the role of Feature Extraction?
Feature Extraction identifies important visual characteristics such as edges, shapes, colours, and textures that help recognize objects.
3. What does the Machine Learning model do?
The Machine Learning model compares extracted features with learned patterns from training data and predicts the object's class.
4. Why is Decision Making the final stage?
After recognizing the object, the Computer Vision system performs the required action, such as unlocking a phone, applying brakes, or triggering an alarm.
5. Do all Computer Vision applications follow the same process?
Yes. Although the implementation may differ, most Computer Vision applications follow the same basic workflow of image acquisition, preprocessing, feature extraction, recognition, and decision-making.
Summary
- The Computer Vision process transforms raw images into meaningful information.
- It consists of Image Acquisition, Image Preprocessing, Feature Extraction, Machine Learning, Object Recognition, and Decision Making.
- Each stage contributes to improving recognition accuracy.
- The process is widely used in healthcare, transportation, agriculture, manufacturing, security, and retail.
- Understanding this workflow helps explain how intelligent vision-based systems operate in the real world.
Next Topic: Applications of Computer Vision