NobleBlocks

Research indexBook chapter

Computational Methods

Dobay, Golomingi, Ebert

CRC Press eBooks12 Aug 2026

NobleIDark:/48914/w1/20261105/79F49091DOI10.4324/9781003413998-28

Paywalled

The full text is at the publisher

You can read the abstract here, or open it at the publisher.

Read at the publisher

Abstract

Image segmentation can be traced back to the emergence of computer vision. The field of computer vision was born in the mid 1960s, and it is traditionally credited to Lawrence Gilman Roberts (1937–2018), an American engineer whose MIT dissertation was entitled “Machine Perception of Three-Dimensional Solids” [ 1 ]. Computer vision was developed to gain a higher level of understanding from digital images. Essentially, the question was how to equip a computer with an electronic eye able to see and distinguish what a human does. In the 1970s, several computer vision algorithms were developed that are today part of image segmentation. These are, for instance, edge detections or polyhedral modeling. More recent ones include machine learning and deep learning techniques. Image segmentation is one of the first steps in quantifying information in a digital image. A digital image displayed on a computer screen exists only as a series of ordered pixels. A very simple way to analyze an image would be to count the number of pixels of a certain kind, for instance, pixel intensity or pixel color. In some cases, this information can be enough to derive knowledge about the actual content of an image. Beyond counting and classifying pixel values, it is necessary to delineate borders of the objects represented on the image. This operation can be done manually by labelling each pixel and assigning each pixel to a class of objects. If we consider an image of https://www.w3.org/1998/Math/MathML" display="inline"> 2400 × 1600 , the total number of pixels is 3,840,000. Manual labelling is therefore not possible, with constant technological improvements bringing larger and larger image resolutions. This operation can be done only with an algorithm and taking advantage of the speed of computers. As comparison, a processor in the 70s clocked at 1 MHz would have taken approximately 4 seconds to process an image of https://www.w3.org/1998/Math/MathML" display="inline"> 2400 × 1600 pixels in size (1 million instructions per second). Today, the fastest processors are clocked around 5 GHz. The same operation takes only a millisecond.

Discussion

More

Suggest a correction or report