Camera Hardware Initialization and Configuration

Proper setup of the image sensor is required before capturing frames. The following routiness manage hardware state and image acquisition parameters:

• sensor.reset(): Reinitializes the camera module to its default state.
• sensor.set_pixformat(fmt): Defines the color depth. Use sensor.GRAYSCALE for 8-bit monochrome data or sensor.RGB565 for 16-bit color output.
• sensor.set_framesize(size): Adjusts the output resolution (e.g., sensor.QVGA).
• sensor.skip_frames(count=15): Discards a specified number of frames immediately after configuration changes to allow the sensor's auto-exposure and white balance algorithms to stabilize.
• sensor.snapshot(): Captures a single frame and returns an image object for processing.
• sensor.set_auto_gain(enable): Toggles automatic gain control.
• sensor.set_auto_whitebal(enable): Toggles automatic white balance adjustment.
• sensor.set_auto_exposure(enable, exposure_us): Controls automatic exposure. When disabled, exposure_us manually sets the shutter duration in microseconds.
• sensor.set_windowing(roi): Crops the sensor output to a specific region of interest. The roi tuple follows the (x, y, width, height) structure.
• sensor.set_hmirror(state) / sensor.set_vflip(state): Applies horizontal mirroring or vertical flipping to the raw sensor data.

Frame Annotation and Drawing Primitives

The image class provides built-in methods for overlaying graphics directly onto captured frames. These operations modify the pixel buffer in place.

• img.draw_line(coords, color): Renders a line segment. coords expects (x1, y1, x2, y2).
• img.draw_rectangle(bounds, color): Outlines a rectangular area using (x, y, w, h).
• img.draw_circle(cx, cy, r, color): Plots a circle centered at (cx, cy) with radius r.
• img.draw_cross(cx, cy, size, color): Draws a crosshair marker.
• img.draw_string(cx, cy, message, color): Overlays ASCII text at the specified coordinates.

Drawing Implementation Example:

import sensor
import image

sensor.reset()
sensor.set_pixformat(sensor.RGB565)
sensor.set_framesize(sensor.QVGA)
sensor.skip_frames(15)

frame = sensor.snapshot()
primary_color = (255, 255, 0)  # Yellow
secondary_color = (0, 255, 255) # Cyan

# Render geometric overlays
frame.draw_line((20, 20, 150, 150), color=primary_color)
frame.draw_rectangle((40, 60, 80, 50), color=secondary_color)
frame.draw_circle(160, 120, 25, color=primary_color)
frame.draw_cross(80, 40, size=8, color=secondary_color)
frame.draw_string(10, 200, "Vision Overlay Test", color=primary_color)

Color Blob Detection and Filtering

Machine vision tasks frequently rely on isolating regions based on color thresholds. The find_blobs() method scans the frame and returns a list of matching regions.

Key Configuration Parameters:

• thresholds: A tuple or list of tuples defining the acceptable color range (typically LAB format: L_MIN, L_MAX, A_MIN, A_MAX, B_MIN, B_MAX).
• x_stride / y_stride: Minimum pixel step size during the scan. Increasing these values improves performence but may miss narrow features.
• invert: When set to True, the algorithm searches for colors outside the provided threshold range.
• area_threshold / pixels_threshold: Filters out detected regions that fall below the specified bounding box area or total pixel count.
• merge: Enables automatic combination of overlapping or adjacent detected regions into a single bounding box.

Blob Tracking Implementation:

import sensor
import image

# Define LAB color range for detection
target_thresholds = [(30, 100, -64, -8, -32, 32)]

sensor.reset()
sensor.set_pixformat(sensor.RGB565)
sensor.set_framesize(sensor.QVGA)
sensor.skip_frames(15)

current_frame = sensor.snapshot()

# Execute color segmentation with filtering
detected_regions = current_frame.find_blobs(
    target_thresholds,
    pixels_threshold=50,
    area_threshold=50,
    merge=True,
    margin=5
)

highlight_color = (0, 0, 255)  # Blue

for region in detected_regions:
    if region.code() == 1:  # Verify threshold match index
        bounds = region.rect()
        centroid = region.cx(), region.cy()

        # Analyze and annotate the detected blob
        current_frame.draw_rectangle(bounds, color=highlight_color, thickness=2)
        current_frame.draw_cross(centroid[0], centroid[1], size=6, color=highlight_color)
        current_frame.draw_string(bounds[0], bounds[1] - 10, f"Area:{region.pixels()}", color=highlight_color)