Human eye visual characteristics (HVS) [from Asuka's blog]

The human eye resembles an optical system, but it is not an optical system in the ordinary sense, but is also regulated by the nervous system. The human eye observation image can be used in the following aspects of the reaction and characteristics:

(1) From the spatial frequency domain, the human eye is a low-pass linear system, the ability to distinguish the scenery is limited. Because the pupil has certain geometrical size and certain optical aberration, the visual cell has certain size, so the resolution of the human eye cannot be infinite, the HVS is not sensitive to too high frequency.

(2) The response of the human eye to brightness has a logarithmic nonlinear property to achieve the dynamic range of its luminance. Because of the nonlinearity of the human eye on the luminance response, the human eye is insensitive to the gray-scale error in areas with large average luminance.

(3) The spatial resolution of the luminance signal of the human eye is greater than the spatial resolution of the chroma signal.

(4) Because the human eye is regulated by the nervous system, from the angle of space frequency, the human eye has the characteristic of the linear system with the continuity. The theory of signal analysis shows that the human eye vision system is weighted and summed to the signal, which is equivalent to making the signal pass through a bandpass filter, which results in a one-side inhibitory effect of the edge-enhancing sensation on the human eye.

(5) The edge information of the image is important to the vision, especially the position information of the edge. The human eye can easily feel the change of position on the edge, but the human eye is not sensitive to the gray-scale error of the edge.

(6) The visual masking effect of human eye is a local effect, which is influenced by background illumination, texture complexity and signal frequency. Regions with different local characteristics allow different signal intensities to be changed without being perceived by the human eye.

The visual characteristic of human eye is a multi-channel (multichannel) model. Alternatively, it has a multi-frequency channel decomposition feature (Mutifrequency channel Decompositon). For example, given a prolonged period of light stimulation on the human eye, the stimulation sensitivity decreases for the same amount of stimulation, but the stimulation of other different frequency segments is not affected (this experiment allows the human eye to observe the sinusoidal grating of different spatial frequencies to confirm). There are many kinds of visual models, such as neuron model, black-and-white model and color visual model, respectively, which reflect the different characteristics of human eye vision. Campbell and Robosn thus assume that there are many independent linear bandpass filters on the retina of the human eye, which decompose the images into different frequency bands, and the bandwidth of different frequency bands is very narrow. Further study of visual physiology also found that the band width of these filters is multiplied, in other words, the images in the retina are decomposed into certain frequency segments, which are equal widths on the logarithmic scale. These characteristics of visual physiology are also confirmed by our observation of things. A low-resolution landscape, we may only be able to distinguish its general outline, the result of improving the resolution, so that we may be able to distinguish the house, trees, lakes and other content, and further improve the resolution, so that we can distinguish the shape of the leaves. Different resolutions can depict different structures of image details.

Human eye visual characteristics (HVS) [from Asuka's blog]