Printed image quality analysis and evaluation
We are a big printing company in Shenzhen China . We offer all book publications, hardcover book printing, papercover book printing, hardcover notebook, sprial book printing, saddle stiching book printing, booklet printing,packaging box, calendars, all kinds of PVC, product brochures, notes, Children's book, stickers, all kinds of special paper color printing products, game cardand so on.
For more information please visit
http://www.joyful-printing.com. ENG only
http://www.joyful-printing.net
http://www.joyful-printing.org
email: info@joyful-printing.net
First. Introduction
The printed matter is the result of the copying of the graphic, and the combined effect of the original, material, equipment, imaging and transfer process results in the formation of the printed image. The vector attributes of graphics and text are essential for page description, typesetting, and RIP interpretation. In particular, you must distinguish between objects of different natures, such as graphics, text, and images, in order to get the most reasonable results. However, the vector properties of graphics and text have become insignificant to the printed results, as these objects lose their vector characteristics once transferred to the surface of the paper. Therefore, when evaluating the print quality, it is not necessary to distinguish the dot matrix description and the vector description feature of the object, and it is considered that the print quality evaluation is equivalent to the print image quality evaluation (hereinafter referred to as image quality evaluation).
The research and application of human image quality evaluation technology has never stopped, and many progress has been made due to the continuous development of image quality improvement activities in the field of imaging technology. For example, image quality evaluation techniques in the fields of traditional printing, silver halide photography, color television, and electrophotographic digital printing, many of which have a good correlation with physical properties and subjective evaluation results, and have also been developed for Image quality evaluation system in different fields. But it is difficult to be universally accepted only by subjective evaluation of a scale.
Color offset printing is the main means of modern graphic printing. The evaluation of the quality of printed images has long been judged by the customer's visual and subjective impressions. This is a very poor and difficult job. Therefore, how to objectively evaluate and fully apply technical processes to achieve color printing image quality control is very important. Print image quality assessment refers to the use of data to represent the correlation between a product and an original. With certain data and indicators, the allowable range and quality degradation of product quality errors are controlled to ensure image repeatability and quality stability.
There are three key factors in the quality of printed images: 1 the sharpness of the image, that is, whether the change in edge density of the image is consistent with the degree of human visual sensitivity, mainly depending on the screen printing and the dot enlargement in printing. 2 The level of the image, that is, the visually distinguishable density level in the image, depends mainly on the contrast expansion and reduction in the prepress and printing. 3 The color of the image, that is, the reproduction accuracy of various colors of the image and the neutral gray balance, mainly depend on the quality of the color separation mechanism, photosensitive material, paper, ink, and the like.
Second, the main method of traditional printed image quality evaluation
The method of evaluating the quality of printed images, no matter which method is adopted, must ultimately be attributed to the three aspects of image clarity, hierarchy and color. The evaluation methods are mainly divided into subjective evaluation and objective evaluation and a comprehensive method based on visual evaluation.
Subjective evaluation
The subjective evaluation of printed images is a method of evaluating the quality of images based on experience. Subjective evaluation methods commonly used are visual evaluation methods and qualitative evaluation methods. The visual evaluation method means that under the same evaluation environment conditions (such as light source and illuminance), the experienced manuscript, the technician and the user observe the manuscript and the printed matter, and then the experience, emotions and hobbies of each person are According to the basis, the quality of each print is excellent, good, medium, and differential, and the frequency of each grade is counted. Those who get the praise are excellent and good, and vice versa. The qualitative index evaluation method refers to a certain qualitative index, and lists the important factors affecting the quality of each index. It is scored by many experienced assessors. The quality of the total score is excellent, and the lower one is poor.
1 multidimensional scaling method:
Multidimensional scaling is a scale technique based on mathematical statistics. When comparing differences between pairs of samples or determining the degree of satisfaction with the sample, the multi-dimensional scale method can be used to analyze and identify the main parameters used in people's assessment. When judging a printed sample by this method, the relative importance of the main parameters of the print quality can be determined; the value obtained by the judgment can make an intrinsic correlation between the subjective evaluation and the objective evaluation or the paper property; and the quality of each print can be obtained. The reliability of the evaluation, the consistency of each evaluator (such as printing, papermaking experts, readers, advertising staff, etc.) and the evaluation of the evaluation team.
The multidimensional scaling technique was proposed by Togosen. The content is: If there is a difference in the feeling between the two elements, then the difference can be expressed by a geometric distance. If you record this difference on a linear scale, the scale value on the scale shows the distance, and then you can use the distance to create more than one-dimensional geometric model that reflects the relationship between the samples.
An important feature of multidimensional scaling techniques is that multidimensional scaling can be performed for the subjective psychological factors of the judges. The role of each parameter in an evaluation can be represented by a desired vector.
2 pairs of comparison methods:
People have subjective characteristics when judging the quality of printed matter. Different people will make completely different conclusions. This objective existence inconsistency cannot be ignored as deviation or randomness. Even if there is a reference that can be used as a benchmark in the evaluation, the evaluation results will be inconsistent, and the inconsistency in subjective evaluation can not be ignored as deviation and randomness.
There are no references at all that can be used as a baseline for comparison. At this time, internal comparison can only be made between the judged objects. There are two common methods: one is to arrange the judged samples in a certain order; the other is to compare each sample in a group of judged samples. The other judged samples are compared one by one, scored on the basis of comparison, and judged according to the points. This is the pairwise comparison method. It belongs to a subjective evaluation method.
Care should be taken when implementing the pairwise comparison method: Ensure that when the judges compare two samples at the same time, do not have unrelated factors to interfere with this comparison. The printed materials should be in the same standard lighting state, and the background should be neutral. The room used for comparison should ensure that the judges concentrate and can not interrupt the judgement or interfere with the judges. The samples to be compared should be provided to the judges in random order so that the comparison process is free of any bias. The judge must be clearly and clearly stated on what basis to judge and not to be ambiguous.
Such subjective contrast experiments are only a comparison between the internal samples, and the comparison results are only related to the samples being compared. Whether the results of the pairwise comparison are reliable can be known by examining the "reliability of the judgement" by the judges, that is, by calculating the reliability coefficient.
2. Objective evaluation method
Evaluation of 1st-order (hierarchical) reproduction:
The contrast range of the transmitted originals is very different, while the color prints are reproduced as a picture with a substantially uniform density range. This density range is much lower than the density range of the color originals. In the color separation plate making, compression adjustment is necessary. What kind of redistribution adjustment is adopted for each tone level depends on the hierarchical distribution of the manuscript and the visual perception of the person. Among them, there are both the Mengsel brightness value factor of visual response and the subjective factor of people's visual psychological requirements. The adjustment of the manuscript gradation level needs to integrate the visual psychological needs of people, add the physical quantity of the visual response, and then combine the state of the original layer distribution to consider the density gradation level reproduction curve of the printed picture, and then incorporate the copy. With the evolution data in the reproduction process, the tone level replication curve of the specific manuscript is designed.
The evaluation of the tone reproduction of the printed image, from the point of view of the quality inspection of the printed matter, is: measuring the printed solid density of each color ink layer; measuring and calculating the overprinting rate of each color ink layer; measuring the dot tone of the printing dot is increased or The measurement prints the K value; the transfer quality of the ink dot is checked; and the density gradation reproduction curve of the print to the original is measured and depicted. Through the measurement of these objective technical data, and then compared with the quality standard standards formulated by the department, the quality level of the specific color printing products can be determined.
2 evaluation of color reproduction:
There are three different concepts for copying and reproducing colors. One is the color reproduction in the physical sense, which requires that the reproduction color be exactly the same as the original color at each color point. The printed matter is for visual viewing. It is difficult to achieve the same-grain color reproduction in the physical sense, and it is not necessary. The second is the reproduction in the sense of colorimetry, so that the printed reproduction image is consistent with or close to the chromaticity of the original color point, that is, the heterogeneous color effect, which is a measure standard for realistic objective evaluation of color reproduction. The third is the color reproduction in the psychological sense, that is, the color of the print reproduction may be somewhat different from the original color in the chromaticity, but the color effect may reach the satisfaction of the visual psychology, and the subjective evaluation factor is added here.
Due to the lack of chromaticity performance characteristics of printing materials (paper, ink, etc.), the incompleteness of color separation and equipment performance, and the lack of color in the printing and reproduction method itself, the actual printing technology cannot faithfully restore the original or All the colors of the original scene, even if the reproducible parts are not up to the level of faithful restoration, can only be relatively close. This brings certain difficulties to the objective evaluation of the quality of print reproduction.
The results of the chromaticity measurement can only be compared from the closeness of the printed matter to the original or the original color. Between similarity and dissimilarity, the psychological resilience of color vision, that is, the degree of psychological reappearance, can be incorporated into the comprehensive and comprehensive evaluation of the color reproduction of printed matter.
If the objective technical measurement scale standard is set in terms of the proximity of the printed color to the reproduction of the original color, it should include: measurement inspection of the color reproduction range of the printing ink, measurement inspection of the printed gray balance reproduction, and absolute reproduction accuracy of the printed color on the original color. Measurement test, and measurement calculation of relative reproduction degree.
3 evaluation of definition reproduction:
The sharpness of color prints is an important quality indicator for image reproduction. Except for the special artistic conception of expressing images, each picture should always have a certain level (subject or background) clear. There are three aspects related to the evaluation of the sharpness of the printed image: the realness of the outline of the image; the sharpness of the contrast between the two adjacent layers of the image is the slight contrast; the resolution of the original or printed image hierarchy, that is, its subtle The level of detail is the essence of the material composition of the objective scene, the so-called texture.
3. Vision-based evaluation method
After years of research on the human visual system (HVS), people have made significant progress in understanding the functions of the visual system from the aspects of neurophysiology and psychophysics. The main characteristics related to image quality are: Weber-Fechner's law, that is, the perceptible threshold of the stimulus signal is proportional to the background intensity; the modulation transfer function (MTF) is the band-pass characteristic, and the maximum response is between 2-5 c/d; The area masking effect, that is, the visibility of the signal appearing on the background of the threshold is reduced; the decomposition of the multi-channel of the area, frequency and direction of the visual signal.
A visually relevant objective distortion measure is usually derived by using the early information processing stage of HVS, from the retina through the optical nerve, the optic chiasm and the lateral geniculate nucleus to the cerebral cortex, to convert the error image into a "perceptual domain". So that the error of the same amplitude has the same visual perception at this time, and finally the sum of the errors is a distortion scale with the appropriate distance measure. The complexity of HVS and the type of quality damage has brought many difficulties to the study of the problem. In the past, some scholars have used visual characteristics obtained from relatively simple psychovisual experiments, such as frequency selection and masking characteristics, to establish visual models for quality prediction. With the continuous development and requirements of coding technology and applications, research on vision-based quality measurement has also been deepened and made significant progress.
S. Karunasekera & N. Kinsbery proposed a distortion measure for damage. This type of impairment is common in vector quantization and DCT coding, consisting of vertical and horizontal edge distortion. Subjective experiments were carried out on synthetic test images. The visual sensitivity of edge damage was measured by adjusting four parameters: edge amplitude, edge length, background brightness and background activity, and a visual model was established. It is worth mentioning that visual sensitivity is measured by the response time of the stimulus, rather than the usual visual threshold method, and the model parameters are determined by the matching of the Constant Sensitivity Point. The model for predicting vertical edge error is (similar in horizontal direction), and the edge information is first extracted from the error image by a low-pass filter with appropriate bandwidth, and then the activity and region masking effects are considered, where the activity masks N× around the edge. The N-square windowed weighted energy and decision of the Fourier transform coefficients; then nonlinear transformation is performed to adapt to the nonlinear characteristics of the HVS, and finally the total error (absolute average) is calculated. In the test of the actual natural image, the author designed some images with the same MSE but different block structure distribution. The model prediction is in good agreement with the subjective test, and it also shows that MSE is not a good distortion measure.
1 Image quality evaluation method based on visual perception:
The effect of image degradation on human visual vision is determined by the sensitivity of the human visual system, which is determined by the visual cells of the human eye. In addition, the sensitivity of the human visual system is also affected by the local spatial frequency of the image. A large number of experimental results prove that the factor affecting the visibility of the pixel error is the local regional environment around the error, not the background environment of the entire image.
According to the above visual characteristics, various HVS models have been established to evaluate the quality of images. The most typical HVS model is shown in Figure 1. It simulates three significant characteristics of visual perception, namely: visual nonlinear characteristics (Weber's law). , visual sensitivity bandpass and visual multichannel and masking effects.
2 Image quality evaluation method based on visual interest:
The coding technology based on image content has revealed the research on image quality evaluation methods based on visual interest. From the perspective of visual psychology, vision is a positive feeling behavior, not only related to physiological factors, but also depends to a considerable extent on psychological factors. When people observe and understand images, they often unconsciously interested in some areas. These areas are called “ROI, Region of Interest”. The visual quality of the entire image is often determined by the quality of the ROI, and the degradation of the region of interest is sometimes difficult to detect. In real life, people's evaluation of the same image will have a large deviation due to the influence of cultural background, surrounding environment and emotion. However, they have commonality to the areas of interest in the image, and they convey the majority of the whole image. Objective information.
The measurement method based on visual interest has opened up a new way for image quality evaluation. However, this method is still in the initial research stage. There are still many problems to be further studied, for example, how to determine the region of interest in the image; When there are multiple regions of interest, how to determine the interest weights of these regions, and so on.
Third, the conclusion
The correct evaluation of image quality is of great significance for the development of overall image information engineering. It is believed that with the rapid development of multimedia information technology, the research on image quality evaluation will receive more and more attention.
In the traditional fusion image quality evaluation methods, there are some defects such as subjective evaluation methods that are too cumbersome and non-repeatable, and the objective evaluation results are inconsistent with or even contradictory to the actual image quality, and the existing evaluation methods are more suitable for correlation. The image fusion evaluation is not suitable for the fusion image evaluation in the field of entertainment photography. Therefore, the introduction of HVS characteristics in the fusion image measurement method and the organic combination of the traditional objective evaluation method and the subjective evaluation method are effective ways to solve this problem, and also the development direction of the fusion image quality evaluation.