Image is a two dimensional signal can be represented by function f(x,y). Binary image, Gray-scale images and color images are the three types of images.
- Binary image is a simplest type consists of two values 0 and 1 i.e, black and white. These images are produced by threshold operation. If pixel reaches above the threshold value, then it will become white. When pixel is below the threshold then it will be in black. The below figure shows binary image representation.
- Gray-scale images have only one color. Normally this type of image holds 8bits/pixel data that have 256 different grey levels. The below figure shows gray-scale image representation.
- Color images are three band monochrome images. Every band consists of a different color and the original picture is saved in the digital representation. Denoted by Red, Green and Blue (RGB images).
Image Processing is an approach to process an image for retrieving applicable information or to acquire an enhanced image. Analog and digital image processing are two method types. Analog image processing executes only two dimensional signals and it is applied over the analog signals. These signals are either periodic or non-periodic. Examples are photographs, paintings, television images, etc. Digital image processing is applied to digital images. Some of the examples are image detection, processing the video frames. Image processing can be seen in various domains like robotics, medical diagnosis, automated inspection of products, surveillance, biometric recognitions, etc.
In order to become suitable for digital image processing f(x,y) should be digitized both in spatially and in amplitude. For creating a digital image, continuous data should be converted into a digital form. These conversions can be done by
The sampling rate discovers the spatial resolution of digitized picture whereas the quantization level finds out the count of grey levels found in the digitized picture. The change of continuous value of image function to its digital value is called as quantization. The typical areas of digital image processing are image acquisition, enhancement, restoration, compression, segmentation, representation and object recognition.
Image acquisition: The object of interest is given here in the digital format. Pre-processing techniques are applied over here. Cropping is a basic type of preprocessing method which eliminates the unwanted areas. Cropping is possible through manual or by the special spatial coordinates [a b c d]. a refers to the pixels from left, b refers to the pixels from bottom, c refers to the selected area width, d refers to the selected area height.
Image enhancement: It is an area in which image will be modified so as to improve its quality for specific application. It is subjective in nature. Spatial domain and frequency domain are two broad categories of image enhancement methods. Spatial domain method refers to cluster of pixels forming an image. It operates directly over the image. In frequency domain method, fourier transform of image is calculated for image enhancement.
Image restoration: A method concentrates on recovering or reconstructing a degraded image by using a prior knowledge about the degradation phenomena. It improves the image appearance. Some of the restoration algorithm are median filter, adaptive filter, linear filters, on negative and support constraints recursive inverse filtering, super-resolution restoration algorithm, deconvolution using a sparse prior, block matching, wiener filter, deconvolution using regularized filter and lucy-richardson algorithm.
Image compression: Storage requirement to save an image or the bandwidth requirement for transmission will be reduced in image compression approach. Original data can be extracted via decompression method. Encode and decoders are involved in compression. Lossy and lossless compression are the two types of compression techniques. Satellite images, geographical survey, weather maps, teleconferencing are the examples of image compression.
Image segmentation: It is a process of isolating or grouping homogeneous areas in an image. Partitioning a digital picture into various segments can be made possible by this approach. Semantic and instance are the two kinds of segmentation. A simple method of segmentation is a threshold technique. It depends on the value of threshold to convert the gray scale into binary image. Traffic control systems, self-driving cars, locating objects in satellite images are the examples of image segmentation.
Image representation: Process of extracting the features from the given image and storing it in memory. Representation of region involves external (boundary) characteristics and internal characteristics. External is preferred for shape property and internal is preferred for region characteristics like texture and color.
Image recognition: It concentrates on identifying an object from the given image. These methods are used in lot of applications are toll booth monitoring, factory automation and surveillance. Optical character recognition, pattern matching, gradient matching, face recognition, license plate matching, deep learning, machine learning is some of the algorithm used in image recognition.
APPLICATIONS OF DIGITAL IMAGE PROCESSING:
- Biometric Recognition is one of the real time examples for image processing.
In olden days, protection is done by token or password based methods. Limitation of this system is security. To increase the level of security, biometric recognition method provides access to the one who has right to do that. This system is very secure, efficient and increased user convenience due to biometric identifiers cannot be easily misplaced. It is implemented in ATM, employee tracking, home applications and so on. Fingerprints recognition procedures involves various algorithms like image processing, feature extraction and feature matching. Fingerprints are represented in gray scale images. Image may contain unwanted noises. It can be removed by process of filtering. After the image features will be extracted and it is saved for matching procedure.
- Image sharpening and restoration:
Image sharpening and restoration is a method in which we can change the image size and appearance. It consists of sharpening, conversion, blurring, edge detection, image recognition and retrieval.
- Medical Area:
Enormous amount of applications is available in medical field. They are X-ray imaging, CT scan, UV imaging, Gamma-ray imaging and so on.
- Robot vision:
Robot vision is one of the applications of image processing like detecting the hurdle path and line follower robot.
- Video processing:
Frame rate conversions, detecting the motion, noise reduction, color space conversion are the applications of image processing.
Image representation and how to process the images using image acquisition, enhancement, restoration, compression, segmentation, representation and object recognition are described here. Nowadays Image processing is available everywhere in this digital environment. Most of our real world application like online scenic wallpapers, camera based artificial intelligent systems, etc. uses image and video processing techniques due to its faster processing capability and it is cost-effective. In future we can see more applications will be based on image processing because of its efficient behavior.