Image quality test method based on parallel convolutional neural network

Abstract

The invention discloses an image quality test method based on a parallel convolutional neural network. The method comprises the following steps that (1) the parallel convolutional neural network is used to establish an image quality test model which comprises five convolution layers and three full-connection layers; (2) input data is preprocessed, and a database is made balanced; (3) the model is pre-trained by using a pre-training data set to carry out pre-training learning on the image quality test model and obtaining a network weight; (4) parallel model training is carried out by initializing the image quality test model, carrying out parallel model training on the basis the pre-trained and initialized image quality test model, and obtaining a trained image quality evaluation model; and (5) a target image is tested by using the trained image quality evaluation model. According to the method of the invention, an obtained test result satisfies an aesthetic standard of human, manual operation is not needed in the whole determination process, and the image quality can be evaluated by a machine in a full automatic manner.

Description

technical field [0001] The invention relates to the field of image signal processing, in particular to an image quality testing method based on a parallel convolutional neural network. Background technique [0002] Although human beings' aesthetic feelings and judgments are influenced by cultural background, personal experience, time background, etc., they generally have a lot of commonality. Numerous paintings, photographs and works of art are generally appreciated and loved by people as the common aesthetic wealth of mankind. Aesthetic quality assessment is to use computers to simulate the high-level perception of human beings to judge the beauty of images, to classify images as high-quality or low-quality, or to score the quality of images. [0003] Most of the traditional image quality assessment methods use hand-selected recognition features, and the effective extraction of image features plays a vital role in the classification results. For example, try to learn from...

AI Technical Summary

Problems solved by technology

[0005] However, the images of different scene categories are very different, which leads to poor adaptability of different image features to images of different scene categories.

In addition, some complex composition rules and quality evaluation rules of images are difficult to be modeled and quantified in engineering, which has become a bottleneck in image feature extraction

Images