Generative models have become a foundational technology for medical imaging, supporting image generation, enhancement, segmentation, registration, image-to-image translation, and anomaly detection. Despite the rapid growth of methodological work, the literature still lacks a unified comparative framework that helps practitioners decide between Generative Adversarial Networks, Variational Autoencoders, and diffusion models for specific medical-imaging tasks; existing surveys also provide limited treatment of validation strategies, dataset constraints, and ethical or clinical considerations. This survey reviews representative methods from each model family across six core tasks, covering anomaly detection, image-to-image translation, image generation, segmentation, denoising, and registration, consolidating them into a unified comparison spanning accuracy, computational cost, data requirements, robustness, uncertainty quantification, and practical deployment. We further compile commonly used benchmark datasets together with their scale, diversity, and known limitations, and we critically discuss validation strategies, model complexity and regularization, and ethical, clinical, and practical considerations. Our key findings are threefold. First, no single model family dominates across all tasks, as Generative Adversarial Networks remain leading for image-to-image translation and data augmentation, Variational Autoencoders are well-suited to reconstruction-based anomaly detection where their posterior provides native uncertainty estimates, and diffusion models currently offer the strongest results in image generation, segmentation, and denoising. Second, practical deployment metrics and external validation remain under-reported in the literature, which limits clinical translation. Third, domain shift, demographic bias, regulatory uncertainty, and reproducibility are open challenges shared across model families. We close with a conceptual taxonomy organized along the model-family by imaging-task axis and outline future research directions.

The optimized sampling is very important for obtaining representative observations over heterogeneous surfaces. However, spatial heterogeneity (SH) is influenced by both randomness and structure factors and varies with scale. A comprehensive understanding of how the contribution of these factors to SH varies with scale is crucial for optimizing sampling. This study quantified the scale dependence of SH caused by structure and randomness factors based on the geostatistical attributes of semivariogram and explored the relationship between the optimal deployment of ground samples and SH dominated by randomness or structure factors. The results showed that as the plot size increased, the proportion of SH caused by spatial structure factors (${P} _{\text {SSF}}$) increased. When the plot size was larger than 20 m, the${P} _{\text {SSF}}$gradually approached 80%–100%. As the plot size increased, the optimal sample plots were distributed on the typical structural features in the area. However, when the plot size was small, the optimal samples were not necessarily located on the predominant surface types. Optimizing sampling can characterize the SH, and the optimal deployment of ground samples should comprehensively consider the plot size and the number of sample plots.

Satellite products have served as the foundation for subsequent analysis, modeling, and decision-making. However, the errors or inconsistencies of satellite products may bias or even mislead the conclusions and decisions based on them. Using ground-based observation data to directly correct the errors in satellite products provides a more relaxed and direct method for constraining the errors of satellite products. However, it is challenged by the sparsity of ground station distribution and the spatial scale mismatch between ground observations and satellite pixels. To address this issue, this study pioneers an integrated and comprehensive methodological framework for pixel-by-pixel error correction based on sparsein situsite observation data across the globe. This methodological framework comprises several core components: the error correction models over the regions within situsites based on the pixel scale ground "truth", the spatial extension model to allocate optimal error correction model for regions withoutin situsites, and finally the pixel-by-pixel error correction of satellite products. MCD43A3 v061 was taken as an example to illustrate the methodology as well as its effectiveness. The RMSE of error-corrected MCD43A3 based on the optimal correction model was reduced from 0.05 to approximately 0.02. To conclude, the results and comparative analysis shown in this study suggested that the proposed framework for pixel-by-pixel error correction of satellite products based on ground observations from sparse networks has the potential to further improve the quality of satellite products across the globe.

Validation of satellite albedo products relies on reference value on the coarse pixel scale which is acquired by independent means. In previous researches, reference value was generally obtained within the nominal spatial extent of the validation pixel fromin situobservations or high spatial resolution airborne/spaceborne albedo references. Nevertheless, the signal of the validation pixel may correspond to different areas due to geometric errors. This geolocation mismatch will introduce large uncertainty into validation results, particularly for pixels covering heterogeneous areas. Therefore, this study first proposed a geometric location matching method on the coarse pixel level to establish the actual position of the validation pixel. The results show that geolocation error of the validation pixels of the high-order satellite products occurs widely. And they are not systematically shifted. The errors of reference values resulting from geolocation errors range from -10% to 25%, which are very likely to be greater than the accuracy requirement of satellite albedo products. Such errors caused by geometric shifts of validation pixels significantly amplify the errors in satellite albedo products. With this geometric location matching method, the reported relative RMSE of MCD43A3 V061 reduced from 9.8% to 3.2%, and the reported correlation coefficient increased from 0.204 to 0.611. This method is very helpful to reduce the uncertainty of validation results and identify the real accuracy of satellite albedo products. Moreover, it has the generalization ability for other numerical variables over other types of land surfaces.

The spatial scale mismatch between satellite andin-situ-based measurements can be reduced by deploying multiplein-situsites within the coarse pixel. However, upscalingin-situmeasurements from the ground-support scale to the coarse pixel scale is still necessary due to their “point” measurement characteristics. The previous upscaling methods were generally developed merely for thein-situmeasurements. Nevertheless, the uncertainty ofin-situmeasurements such as measurement errors and spatial representativeness errors was not dealt with. Consequently, the upscaling results inevitably suffer from errors, which will finally propagate into the pixel scale ground “truth”. For the first time, this study presents an improved upscaling method with the consideration of the uncertainty ofin-situmeasurements based on the error theory and measurement adjustment theory. The effectiveness of the corrected upscaling coefficients was evaluated by comparing the accuracy of the corrected upscaling results with those based on the upscaling coefficients without considering the uncertainty ofin-situmeasurements. The results indicate that the accuracy of the upscaling results can be enhanced by 11.06% in the condition in whichin-situmeasurements suffer from large uncertainty. However, if the uncertainty ofin-situmeasurements is negligible, the corrected upscaling model is not necessary because it does not bring many benefits. Although the effectiveness of this method was only tested on a limited study area, it makes an important first step toward a higher precision pixel-scale ground “truth”, especially when the uncertainty ofin-situmeasurements is non-negligible.

Uncertainty quantification is an important part of validation, because the pixel scale reference generally suffers from uncertainty caused by different factors, lowering the accuracy of validation results. In order to take a step forward to characterize the uncertainty of validation results, this study proposed a simulated shift-based pixel matching (SSPM) method with the aim of quantifying the uncertainty caused by geometric mismatch in the multiscale validation. Furthermore, its relationships with spatial heterogeneity and subpixel size were also explored. It was found that the uncertainty caused by the geometric mismatch is nonnegligible in multiscale validation, which would obscure the true accuracy of satellite products. Spatial heterogeneity makes a positive contribution to the uncertainty caused by geometric mismatch, but the magnitude depends on subpixel size, being weaker with small subpixel size and stronger with larger subpixel size. Subpixel size is generally positively related to geometric uncertainty. But in the case of very large spatial heterogeneity, their correlation is very weak. This study is an important step toward quantitatively characterizing the uncertainties of pixel scale reference in order to increase the confidence of validation results.

Validation of satellite albedo products is an essential step because their quantitative application lie in their ability to record the real state of the earth surface. Upscalingin situmeasurements to the corresponding pixel scale is necessary due to the spatial scale mismatch betweenin situand satellite measurements. Machine learning-based models have been increasingly used for upscaling because they can yield more reliable results than traditional methods. Nevertheless, the main controlling factors on upscaled results have rarely been discussed. This article explores the control factors that bring uncertainties to the upscaled results based on machine learning models. Three machine learning models, including random forest (RF),$k$-nearest neighbor (KNN), and Cubist models, were selected to upscale single sitein situ-based albedo to the coarse pixel scale. The upscaled results were carefully assessed through comparison with pixel scale albedo reference. The results indicate that the accuracy of upscaled results depends on the machine learning models, the inclusion of key variables related to albedo, the dataset selection of these variables, the amount of training data, and the sensitivity of machine learning models to these factors. Despite the dependence on control factors, the machine learning-based upscaling methods generally have excellent applicability across different spatial scales and over other untrained areas. Therefore, they open the door to generating a time series of globally, spatially continuous distributed reference datasets with sufficient length, consistency, and continuity to adequately fulfill the requirement of a comprehensive validation.

A comprehensive assessment of satellite-derived albedo products is undeniably essential for better use consideration and the further refinement of the retrieval algorithm. Although satellite albedo products have been extensively validated over spatially homogeneous areas, it remains a challenge to validate them over rugged terrain. Consequently, the accuracy of satellite albedo products over rugged terrain is still unknown. This study for the first time systematically evaluated two widely used satellite albedo products (i.e., MCD43A3 V006 and GLASS V04 albedo) over mountainous areas with a Mountain Radiation Transfer (MRT) coupled multi-scale validation strategy. Fine-scale albedo was first generated with a root mean square error (RMSE) smaller than 0.0317. Then they were upscaled to the coarse pixel and as the reference data for validation. The validation results indicated that the accuracy of the two products tends to decrease with the increase of means slopes. The RMSE and relative RMSE (RMSER) of full retrieval MCD43A3 C6 black-sky albedo (BSA) and white-sky albedo (WSA) over abrupt slopes (mean slope >10°) increase to 0.0432 and 31.87% and to 0.0436 and 32.21%, respectively. The RMSE and RMSERof high quality GLASS V04 were 0.0452 and 33.71% of BSA and 0.0458 and 33.92 % of WSA respectively over abrupt slopes. Particularly, if the backup retrievals were included over the abrupt slopes, the RMSE and RMSER of MCD43A3 C6 can reach to 0.0600 and 36.92% for BSA and 0.0613 and 37.67% for WSA, and those of GLASS V04 can reach to 0.0567 and 36.28% for BSA and 0.0540 and 35.72% respectively.

A comprehensive assessment of satellite-derived albedo products is undeniably essential for better use consideration and the further refinement of the retrieval algorithm. Although satellite albedo products have been extensively validated over spatially homogeneous areas, it remains a challenge to validate them over rugged terrain. Consequently, the accuracy of satellite albedo products over rugged terrain is still unknown. This study for the first time systematically evaluated two widely used satellite albedo products (i.e., MCD43A3 C6 and Global Land Surface Satellite (GLASS) V04 albedo) over mountainous areas with a mountain radiation transfer (MRT) coupled multiscale validation strategy. Fine-scale albedo was first generated with a root-mean-square error (RMSE) smaller than 0.0317. Then, they were upscaled to the coarse pixel and as the reference data for validation. The validation results indicated that the accuracy of the two products tends to decrease with the increase of means slopes. The RMSE and relative RMSE (RMSER) of full retrieval MCD43A3 C6 black-sky albedo (BSA) and white-sky albedo (WSA) over abrupt slopes (mean slope >10°) increase to 0.0432 and 31.87% and to 0.0436 and 32.21%, respectively. The RMSE and RMSERof high-quality GLASS V04 were 0.0452 and 33.71% of BSA and 0.0458 and 33.92% of WSA, respectively, over abrupt slopes. In particular, if the backup retrievals were included over the abrupt slopes, the RMSE and RMSERof MCD43A3 C6 can reach 0.0600 and 36.92% for BSA and 0.0613 and 37.67% for WSA, respectively, and those of GLASS V04 can reach 0.0567 and 36.28% for BSA and 0.0540 and 35.72%, respectively.

Characterizing the subpixel heterogeneity within satellite pixels is a key issue in validation. Nevertheless, it is challenging due to multi-scale problems in the geological description based on remote sensing. Based on an airborne platform, the multi-scale variation laws of several key indicators in validation including spatial heterogeneity (SH), representativeness errors, and representative area with subpixel size were analyzed and discussed. Furthermore, this article discussed the optimal subpixel size to assess SH within a coarse pixel and the optimal footprint ofin situmeasurements for building dense and sparse validation networks. SH decreases with the increase of subpixel size. And a reduction of about 10% can be obtained from 5 m$\times 5$m to 150 m$\times150$m subpixel size, depending on the degree of SH within the typical satellite pixels. And the sensitiveness of SH to subpixel size decreases gradually with the increasing of subpixel size. Ideally, SH should be assessed using maps with pixel sizes corresponding to the footprint ofin situmeasurements. Regarding the deployment of future validation networks, the footprint ofin situsites should be designed at least larger than 25 m for dense networks. And much larger footprints (e.g., 100 m) are preferred in designing sparse networks. The representativeness error is not fully related to subpixel sizes because it is affected by many factors. The findings are also transferable to model evaluation when comparing model grid values to local observations.

The in situ measurements from globally distributed sparse networks provide a valuable data source for the validation of satellite products. However, the representativeness errors resulting from the spatial scale mismatch between in situ-satellite measurements and surface heterogeneity have generally limited past validation to only very few spatially representative sites, which cannot meet the requirement of a comprehensive validation. In response to this challenge, this article offers a strategy for upscaling sparse in situ measurements and removing the impact of representativeness errors on the evaluation of coarse-pixel albedo products. The main idea of the upscaling method is to establish the correspondence relationship between each subpixel albedo time series within a coarse pixel and in situ albedo time series by using high-resolution albedo maps as the prior knowledge. Furthermore, the performance of the upscaling method is carefully evaluated over the sites featured by different degrees of spatial representativeness. The results indicate that the upscaling method improves the representativeness of single-site measurements with respect to a coarse pixel, and the improvement is most significant over the sites with relatively low representativeness. Therefore, the upscaling method is particularly useful for the validation at heterogeneous sites in strengthening the reliability of validation results. It is expected to open the door to maximizing the use of existing sparse networks and generating a time series of globally distributed reference data sets with sufficient length, consistency, and continuity.

Validation of remote sensing albedo products involves comparisons between point-scale in situ observations and footprint-scale satellite retrievals. However, the observed differences between product and in situ observations are not only attributable to intrinsic errors of satellite products but also to inadequate spatial representativeness of in situ observations. Here, representativeness errors of in situ observations and their effects on validation results were quantitatively explored. Furthermore, the contributors and their influences on representativeness errors were quantified. In the case of large representativeness errors, validation result errors are mainly controlled by representativeness errors. When representativeness errors are small, validation result errors are likely affected by other factors and can be so large that cannot be ignored. Surface heterogeneity is most positively related to representativeness errors, followed by the deviation distance of in situ site from the pixel center. The representative area surrounding in situ sites only shows a weak negative correlation with representativeness errors. The range seems to be not a good indicator of spatial representativeness of in situ sites since there is almost no relationship between them. When these factors are combined, surface heterogeneity contributes more to representativeness errors on the 500-m pixel scale, while quantitative impacts of the representative area and location deviation of in situ sites are not fully understood because magnitudes of these effects are dependent on the choice of high-resolution data set. These findings enhance our understanding about spatial representativeness of in situ observations and improve the quality of validation results based on single in situ observations.

The systemic validation works were carried out at a watershed scale based on the ground-based observation data of the Heihe Watershed Allied Telemetry Experimental Research (HiWATER). Three validation strategies, scaling-up, spatial representation analysis, footprint analysis were used based on different data acquirement techniques. Some studies were performed and four types of remote sensing products were validated. This paper makes a general introduction on the validation results based on these systematic validation activities, which aims to support the integrated study of the water-ecosystem-economy in the Heihe River Basin.

This study describes the use of ground-based albedometer measurement based on the automatic weather stations (AWS) for validating MCD43A3 and GLASS albedo products over heterogeneous landscapes in Heihe river Basin, China. Because the footprint of ground observed albedo was far less than the spatial resolution of albedo products, high-resolution albedo imageries were used as an upscaling bridge to reduce the scale discrepancy. Based on this scheme, we present the results from an accuracy assessment of MODIS and GLASS. The validation results show that MODIS and GLASS have RMSEs less than 0.05 over large areas and over a full year of measurements.

Presents a hierarchical genetic algorithm (GA) to solve the cell formation and layout decisions of cellular manufacturing. The intrinsic features of our proposed GA include using a hierarchical chromosome structure to encode concurrent cell design and layout decisions, developing a new selection scheme to dynamically consider two highly correlated fitness functions, and proposing a group mutation operator to increase the probability of mutation. Our tests show that these modifications are fairly effective in improving solution quality as well as shortening the speed of convergence.