In order to evaluate the qualities of fruits, a new method based on HSI color model was proposed in this paper. An image of fruits was taken and transferred from the RGB color model to the HSI color model. Its simplified histogram of hue H was calculated, which was used as the input of a designed BP network. The output of the BP network was the description of evaluated quality of the fruits. After training, the quality of fruits was identified by the BP network according to the simplified histogram of H of their colored image. Experiments were conducted with satisfied results to show the feasibility and usefulness of the proposed method of fruit quality evaluation.

Study on HSI color model-based fruit quailty evaluation

The present work aims to analyze di erent existing image descriptors for contentbased
image retrieval (CBIR) and combine in such a way as to obtain a hybrid
descriptor that allows a better accuracy in feature extraction and automatic image
grouping, than their predecessors.
To achieve this it is studied, analyzed and evaluated the di erent color descriptors,
texture and form to verify its usefulness in this work. Subsequently focused on
the color descriptor CLD (Color Layout Descriptor) and the Tamura texture descriptor
for feature extraction; for clustering was used classi er K-means partition and
SOM (Self-Organizing Map) classi er, with a database of 1000 images grouped en
10 classes.
The results showed a better performance against the metrics of precision and recall
for Hybrid descriptor (CLD-Tamura) with K-means in the extraction, clustering
and image retrieval by similarity. Finally, we can conclude that hybrids are more
e ective descriptors in the feature extraction performance by increasing the recovery
process by increasing the percentage of its precision and recall in the ruling class
groups

Recuperación de imágenes basadas en contenido mediante descriptores híbridos

Sobel is one of the most popular edge detection operators used in image processing. To date, most users utilize the two-directional 3x3 Sobel operator as detectors because of its low computational cost and reasonable performance. Simultaneously, many studies have been conducted on using large multi-directional Sobel operators to satisfy their needs considering the high stability, but at an expense of speed. This paper proposes a fast graphics processing unit (GPU) kernel for the four-directional 5x5 Sobel operator. To improve kernel performance, we implement the kernel based on warp-level primitives, which can significantly reduce the number of memory accesses. In addition, we introduce the prefetching mechanism and operator transformation into the kernel to significantly reduce the computational complexity and data transmission latency. Compared with the OpenCV-GPU library, our kernel shows high performances of 6.7x speedup on a Jetson AGX Xavier GPU and 13x on a GTX 1650Ti GPU.

Multi-directional Sobel operator kernel on GPUs

Deep convolutional neural network (DCNN for short) models are vulnerable to examples with small perturbations. Adversarial training (AT for short) is a widely used approach to enhance the robustness of DCNN models by data augmentation. In AT, the DCNN models are trained with clean examples and adversarial examples (AE for short) which are generated using a specific attack method, aiming to gain ability to defend themselves when facing the unseen AEs. However, in practice, the trained DCNN models are often fooled by the AEs generated by the novel attack methods. This naturally raises a question: can a DCNN model learn certain features which are insensitive to small perturbations, and further defend itself no matter what attack methods are presented. To answer this question, this paper makes a beginning effort by proposing a shallow binary feature module (SBFM for short), which can be integrated into any popular backbone. The SBFM includes two types of layers, i.e., Sobel layer and threshold layer. In Sobel layer, there are four parallel feature maps which represent horizontal, vertical, and diagonal edge features, respectively. And in threshold layer, it turns the edge features learnt by Sobel layer to the binary features, which then are feeded into the fully connected layers for classification with the features learnt by the backbone. We integrate SBFM into VGG16 and ResNet34, respectively, and conduct experiments on multiple datasets. Experimental results demonstrate, under FGSM attack with $\epsilon=8/255$, the SBFM integrated models can achieve averagely 35\% higher accuracy than the original ones, and in CIFAR-10 and TinyImageNet datasets, the SBFM integrated models can achieve averagely 75\% classification accuracy. The work in this paper shows it is promising to enhance the robustness of DCNN models through feature learning.

Improving the Robustness of Deep Convolutional Neural Networks Through Feature Learning

Image Sobel edge extraction algorithm accelerated by OpenCL

This paper discusses some digital image processing ways,which has standard format of the image files.A program database was built .In order to read digital images from exterior system correctly and compress them effectively.By this way,some hardware devices have been saved,e.g.the digitalization apparatus,and great convenience has been given during the process of pretreatment,analysis ,recognition of the following images and the retransfer of the treated ones through the net ware.

Han Xiao

Papers

Image Sobel edge extraction algorithm accelerated by OpenCL

The Digital Image Processing Ways with the Standard Format of the Image Files