image-manipulation's Introduction

Image Manipulation

Repository for documenting about some of the image manipulation techniques using Python and other open-source libraries.

1. Instance Segmentation and Object Detection

Make use of pixellib library
Use of a Mask R-CNN model
- trained on coco dataset
Python script here

2. Basic Image Manipulation using Pillow

Some basic PIL functions for image processing

Merge two images
Add/subtract two images
Convert to grayscale
Convert to Black-and-White
Colour Inversion
Rotation
Gaussian Blur
Edge Detection

Python notebook with the above implemented operations can be accessed here.

3. OpenCV based Basic Stuff

3.1 Basic image loading

3.2 Extracting snippet of an image

3.3 Webcam based video capture

Capturing webcam feed
Multiply, manipulate and fit to grid

3.4 Drawing and Writing on Image

Drawing overlay shapes and text on a webcam feed
Can be generalized to any image or video

3.5 Colour Detection and Masking

Creating a masking image to detect a certain range of colour
Applying mask to the video frame/image using bitwise_and

3.6 Corner Detection

Shi-Tomasi Corner Detector

3.7 Template Matching

Input:

Base input image:
Template image(s), aka images to detect:

We can make use of one or several methods for performing the template matching. Available methods are:

methods = [cv2.TM_CCOEFF,
            cv2.TM_CCOEFF_NORMED,
            cv2.TM_CCORR,
            cv2.TM_CCORR_NORMED,
            cv2.TM_SQDIFF,
            cv2.TM_SQDIFF_NORMED]

Output:

Blue car
Red car

3.8 Face and Eye Detection

4. Image Stitching

A simple Python script to stitch all the images in a specific directory to create a final collage.
Pillow library utilized for image reading/writing
Images are stitched in alphabetical order, from left to right, top to bottom
Preserves the quality, as it stitches the images as it is, without compressing
Python script here
Output snapshot: