HSV Camera Robotic

ICT 361: Introduction to Robotics

Mr. Seng Theara

Learning Objective

By the end of the lesson, student will be able to:

Understand how a joystick works electrically
Explain pull-up button logic
Read analog and digital signals
Map controller inputs to robot movements
Design basic control logic for robots

Learning Objective

By the end of the lesson, student will be able to:

Explain the HSV color model and its advantage over RGB
Describe how a camera can replace manual control
Convert camera image to HSV using OpenCV
Apply thresholding to detect a line
Generate a binary mask from the image
Calculate line position (cx) from the mask

4 Buttons

Joystick

Big Picture

Joystick

Big Picture

Joystick

ROI (Region of Interest)

Joystick

ROI (Region of Interest)

ROI helps the robot focus only on what matters.

In this image, the ROI is divided into three regions:
Left | Center | Right

If the black line is in the center → the robot moves straight
If the black line is in the left → the robot turns left
If the black line is in the right → the robot turns right

Goal: Keep the line in the center region

4 Buttons

Joystick

What is a HSV?

4 Buttons

What is a HSV?

HSV (Hue, Saturation, Value) is like a color wheel + brightness control

1. Hue (H) represent the type of color and measure in degree

Example:

0° → Red
60° → Yellow
120° → Green
240° → Blue

2. Saturation (S) represent controls color intensity

Value	Meaning
0	Gray
255	Full Color

4 Buttons

3. Value (V) represent the brightness

Value	Meaning
0	Black
255	Bright

HSV separates color from brightness. That means even if lighting changes, we can still detect the object
We are not detecting black by color. we are detecting it by LOW brightness.

Black line Detection

A black line has

No color → Saturation (S) ≈ 0
No light → Value (V) ≈ 0

HSV Threshold

Lower = (0, 0, 0)
Upper = (180, 255, 80)

1. Hue (H)

We set H value range between 0 to 180 because we don't care about the color. Black has no color

2. Saturation (S)

We set S in the range between 0 to 255. All saturation value allowed because black already has low S but lighting noise may change S slightly

3. Value (V)

V is the most important which decide only the dark pixels are selected. We set V in the range of value between 0 to 80.

We set V in the range of value between 0 to 80 because if we do 50, the camera can miss the black line and if we do 150 it can detect other color as the black line as well.

Compare Different Objects

Object	H	S	V	Why
black	Any	Low	Low	No light
white	Any	Low	High	Bright
Gray	Any	Low	Medium	Medium Brightness
Red	Low	High	High	Strong Color

You must tune the HSV value because it is related to the condition of where the robot is used

For example

Bright room set V to 60
Normal light set V to 80
Dark room you might set it to 100

Camera Image (RGB) to HSV

when you do

frame = cap.read()

The image from the camera is BGR Format because of the openCV library.

We cannot use BGR to detect the black line directly because of some reason

when the light changes the BGR values change a lot
When there is shadow, the color become confusing
The BGR is hard to isolate the object

That is why the solution is convert it to HSV through the code below

hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)

Code for the black line detection with ROI

from flask import Flask, Response
import cv2
import numpy as np

app = Flask(__name__)


cap = cv2.VideoCapture(0)
cap.set(3, 640)
cap.set(4, 480)


def get_mask(roi):
    hsv = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
    lower = (0, 0, 0)
    upper = (180, 255, 80)

    mask = cv2.inRange(hsv, lower, upper)

    return mask

def generate_frames():
    while True:
        ret, frame = cap.read()
        if not ret:
            break

        h, w = frame.shape[:2]

        top = int(0.6 * h)
        roi = frame[top:h, :]

        mask = get_mask(roi)

        M = cv2.moments(mask)

        decision = "NO LINE"

        if M["m00"] > 0:
            cx = int(M["m10"] / M["m00"])

            cv2.circle(roi, (cx, 50), 8, (0, 0, 255), -1)

            # =========================
            # 4. Decision
            # =========================
            if cx < w // 3:
                decision = "LEFT"
            elif cx > 2 * w // 3:
                decision = "RIGHT"
            else:
                decision = "FORWARD"


        cv2.rectangle(frame, (0, top), (w, h), (0, 255, 0), 2)

        cv2.line(frame, (w//3, top), (w//3, h), (0, 0, 255), 2)
        cv2.line(frame, (2*w//3, top), (2*w//3, h), (255, 0, 0), 2)
        cv2.putText(frame, decision, (10, 40),
                    cv2.FONT_HERSHEY_SIMPLEX, 1, (0,255,0), 2)

        # =========================
        # 6. Send to web
        # =========================
        _, buffer = cv2.imencode('.jpg', frame)
        frame_bytes = buffer.tobytes()

        yield (b'--frame\r\n'
               b'Content-Type: image/jpeg\r\n\r\n' + frame_bytes + b'\r\n')

@app.route('/')
def video():
    return Response(generate_frames(),
                    mimetype='multipart/x-mixed-replace; boundary=frame')

if __name__ == "__main__":
    app.run(host='0.0.0.0', port=5000)