When it comes to image recognition, this ability of humans is particularly outstanding. Visual stimuli act on sensory organs, allowing people to recognize them as a familiar shape, and even perceive changes in distance or shape of the image. This process is called image recognition. In image recognition, both the information entering the senses at that moment and the information stored in memory are required. Only through the processing of comparing stored information with current information can the recognition of the image be achieved.
Computer image recognition technology uses computers to process, analyze, and understand images to identify various different patterns and objects. The issue studied in image recognition is how to use computers to automatically process large amounts of physical information, solving problems that humans cannot recognize or are too resource-intensive to recognize, thus greatly liberating human labor.
Image recognition technology is an important field of artificial intelligence. It is based on the main features of images. Each image has its characteristics, and studies on eye movement during image recognition show that the gaze is always focused on the main features of the image, specifically where the curvature of the image contour is greatest or where the contour direction suddenly changes; these areas contain the most information. Moreover, the scanning path of the eyes always sequentially moves from one feature to another. Thus, in the image recognition process, the perceptual mechanism must exclude redundant information from the input and extract key information. Simultaneously, there must be a mechanism in the brain responsible for integrating information, which can organize the phased information into a complete perceptual image.
To create computer programs that simulate human image recognition activities, various image recognition models have been proposed, such as the template matching model. This model posits that to recognize a certain image, there must be a memory template of that image from past experiences. If the current stimulus can match a template in the brain, the image is recognized. However, this model emphasizes that the image must completely match the template in the brain to be recognized, which has certain limitations. Gestalt psychologists subsequently proposed a prototype matching model. This model suggests that what is stored in long-term memory is not countless templates to be recognized, but certain “similarities” of images. The “similarities” abstracted from images can serve as prototypes to test the image to be recognized. If a similar prototype can be found, the image is recognized. However, this model does not explain how humans distinguish and process similar stimuli, making it difficult to implement in computer programs. Therefore, a more complex model, the “pandemonium” recognition model, was proposed.
