In-depth Computer Vision

Computer vision is an automated process that integrates a large number of processes for visual perception, such as image acquisition, image processing, recognition, and decision-making. Computer vision tries to imitate the workings of the human visual system (human vision), which is actually very complex. For this reason, computer vision is expected to have the same high-level skills as human vision. These capabilities include:

• Object detection
• Recognition
• Description
• 3D Inference
• Interpreting motion

Computer vision is often defined as a branch of science that studies how computers can recognize observed objects. This branch of science, together with artificial intelligence (artificial intelligence), will be able to produce a visual intelligence system (visual intelligence system). Computer vision is a combination of image processing and pattern recognition, the relationship between the three of which can be seen in picture below. Image processing is the initial process of computer vision, while pattern recognition is the process of interpreting images.

Image processing is a field related to the image transformation process. This process aims to improve image quality. Meanwhile, pattern recognition relates to the process of identifying objects in an image or interpreting an image. This process aims to extract the information or message conveyed by the image.

There are three processes that occur in computer vision, namely:

• Obtaining or acquiring digital images
• Image processing operations
• Analyze and interpret images and use the processing results for specific purposes, e.g., guiding robots, controlling equipment, etc.

Hierarchy in computer vision there are 3 stages, namely:

• Low Level Processing (Image to image) → Remove noise, and image enhancement (image enhancement).
• Intermediate Level Processing (Image to symbolic) → Collection of lines / vectors that represent the boundaries of an object in the image.
• High-Level Processing (Symbolic to symbolic) → Symbolic representation of object boundaries generates the name of the object.

To support the tasks of computer vision, there will be several supporting functions added to this system, namely:

1. Image acquisition

 The image capture process is the process of capturing visual information and converting analog signals into digital data that are ready to be processed by a computer.

• Image acquisition in humans begins with the eyes, and then visual information is translated into a format that can be manipulated by the brain.
• In line with the process above, computer vision requires an eye to perceive a visual signal.
• Generally, the eye in computer vision is a VI camera.
• The camera translates a scene or image.
• Then this electrical signal is converted into binary numbers, which will be used by the computer for processing.
• The output from the camera is an analog signal, where the frequency and amplitude (the former is related to the number of signals in one second, while the latter is related to the height of the electrical signal produced) represent the details of the sharpness (brightness) of the scene.
• The camera observes an incident on one line at a time, scans it, and divides it into hundreds of equal horizontal lines.
• Each line creates an analog signal whose amplitude describes the change in brightness along the signal line.
• Because computers do not work with analog signals, an analog-to-digital converter (ADC) is required for the computer to process all of these signals.
• This ADC will change the analog signal that is represented in the form of single signal information into a stream of a number of binary numbers.
• This binary number is then stored in memory and will become raw data to be processed.

2. Image Processing

The image recognition process is the processing of image information that has been digitized by an analog-to-digital converter.

• Computer vision will involve a number of initial manipulations of the binary data.
• Image processing helps increase and improve image quality so that it can be further analyzed and processed more efficiently.
• Image processing will improve the signal-to-noise ratio (s/n).
• These signals are information that will represent the objects in the image.
• Noise, on the other hand, is any form of interference (lack of blurring) that occurs on an object. 

3. Image Analysis

Image data analysis is the process of analyzing visual images that have been previously processed.

• Image analysis will explore the scene in terms of the main characteristics of the object through an investigative process.
• A computer program will begin to look through the binary numbers that represent visual information to identify specific features and characteristics.
• More specifically, an image analysis program is used to find the edges and boundaries of objects in an image.
• An edge is formed between an object and its background or between two specific objects.
• This edge will be detected as a result of different brightness levels on different sides of one of the boundaries.

4. Image Understanding

In this process, computer vision applies the concepts of artificial intelligence (AI) to understand the visual data it captures.

• This is the final step in the computer vision process, in which specific objects and their relationships are identified.
• This section will involve a study of artificial intelligence techniques.
• Comprehension of template matching in a scene
• This method uses a search program and pattern-matching techniques.