A Multi-Condition Relighting with Optimal Feature Selection to Robust Face Recognition with Illumination Variation

In this paper, we proposed a new approach for face recognition with robust to illumination variation. The improved performance to various lights in recognition is obtained by a novel combination of multicondition relighting and optimal feature selection. Multi-condition relighting provides a "coarse" compensation for the variable illumination, and then the optimal feature selection further refines the compensation, and additionally offers the robustness to shadow and highlight, by deemphasizing the local mismatches caused by imprecise lighting compensation, shadow or highlight on recognition. For evaluation, two databases with various illumination mismatches have been used. The results have demonstrated the improved robustness of the new methods.

Tomato detection method using domain adaptive learning for dense planting environments

This study aimed to address the challenge of accurately and reliably detecting tomatoes in dense planting environments,a critical prerequisite for the automation implementation of robotic harvesting.However,the heavy reliance on extensive manually annotated datasets for training deep learning models still poses significant limitations to their application in real-world agricultural production environments.To overcome these limitations,we employed domain adaptive learning approach combined with the YOLOv5 model to develop a novel tomato detection model called as TDA-YOLO(tomato detection domain adaptation).We designated the normal illumination scenes in dense planting environments as the source domain and utilized various other illumination scenes as the target domain.To construct bridge mechanism between source and target domains,neural preset for color style transfer is introduced to generate a pseudo-dataset,which served to deal with domain discrepancy.Furthermore,this study combines the semi-supervised learning method to enable the model to extract domain-invariant features more fully,and uses knowledge distillation to improve the model's ability to adapt to the target domain.Additionally,for purpose of promoting inference speed and low computational demand,the lightweight FasterNet network was integrated into the YOLOv5's C3 module,creating a modified C3_Faster module.The experimental results demonstrated that the proposed TDA-YOLO model significantly outperformed original YOLOv5s model,achieving a mAP(mean average precision)of 96.80%for tomato detection across diverse scenarios in dense planting environments,increasing by 7.19 percentage points;Compared with the latest YOLOv8 and YOLOv9,it is also 2.17 and 1.19 percentage points higher,respectively.The model's average detection time per image was an impressive 15 milliseconds,with a FLOPs(floating point operations per second)count of 13.8 G.After acceleration processing,the detection accuracy of the TDA-YOLO model on the Jetson Xavier NX development board is 90.95%,the mAP value is 91.35%,and the detection time of each image is 21 ms,which can still meet the requirements of real-time detection of tomatoes in dense planting environment.The experimental results show that the proposed TDA-YOLO model can accurately and quickly detect tomatoes in dense planting environment,and at the same time avoid the use of a large number of annotated data,which provides technical support for the development of automatic harvesting systems for tomatoes and other fruits.

Centre symmetric quadruple pattern-based illumination invariant measure

A centre symmetric quadruple pattern-based illumination invariant measure(CSQPIM)is proposed to tackle severe illumination variation face recognition.First,the subtraction of the pixel pairs of the centre symmetric quadruple pattern(CSQP)is defined as the CSQPIM unit in the logarithm face local region,which may be positive or negative.The CSQPIM model is obtained by combining the positive and negative CSQPIM units.Then,the CSQPIM model can be used to generate several CSQPIM images by controlling the proportions of positive and negative CSQPIM units.The single CSQPIM image with the saturation function can be used to develop the CSQPIM-face.Multi CSQPIM images employ the extended sparse representation classification(ESRC)as the classifier,which can create the CSQPIM image-based classification(CSQPIMC).Furthermore,the CSQPIM model is integrated with the pre-trained deep learning(PDL)model to construct the CSQPIM-PDL model.Finally,the experimental results on the Extended Yale B,CMU PIE and Driver face databases indicate that the proposed methods are efficient for tackling severe illumination variations.

A Local Binary Pattern-Based Method for Color and Multicomponent Texture Analysis

Local Binary Patterns (LBPs) have been highly used in texture classification for their robustness, their ease of implementation and their low computational cost. Initially designed to deal with gray level images, several methods based on them in the literature have been proposed for images having more than one spectral band. To achieve it, whether assumption using color information or combining spectral band two by two was done. Those methods use micro structures as texture features. In this paper, our goal was to design texture features which are relevant to color and multicomponent texture analysis without any assumption. Based on methods designed for gray scale images, we find the combination of micro and macro structures efficient for multispectral texture analysis. The experimentations were carried out on color images from Outex databases and multicomponent images from red blood cells captured using a multispectral microscope equipped with 13 LEDs ranging from 375 nm to 940 nm. In all achieved experimentations, our proposal presents the best classification scores compared to common multicomponent LBP methods. 99.81%, 100.00%, 99.07% and 97.67% are maximum scores obtained with our strategy respectively applied to images subject to rotation, blur, illumination variation and the multicomponent ones.

Adaptive Illumination Normalization Approach Based on Denoising Technique for Face Recognition

A novel adaptive illumination normalization approach is proposed to eliminate the effects caused by illumination variations for face recognition. The proposed method divides an image into blocks and performs discrete cosine transform(DCT) in blocks independently in the logarithm domain. For each block-DCT coefficient except the direct current(DC) component, we take the illumination as main signal and take the reflectance as "noise". A data-driven and adaptive soft-thresholding denoising technique is employed in each block-DCT coefficient except the DC component. Illumination is estimated by applying the inverse DCT in the block-DCT coefficients, and the indirectly obtained reflectance can be used in further recognition task. Experimental results show that the proposed approach outperforms other existing methods. Moreover, the proposed method does not need any prior information, and none of the parameters can be determined by experience.

Feature extraction for classification of different weather conditions

Classification of different weather conditions provides a first step support for outdoor scene modeling,which is a core component in many different applications of outdoor video analysis and computer vision.Features derived from intrinsic properties of the visual effects of different weather conditions contribute to successful classification.In this paper,features representing both the autocorrelation of pixel-wise intensities over time and the max directional length of rain streaks or snowflakes are proposed.Based on the autocorrelation of each pixel’s intensities over time,two temporal features are used for coarse classification of weather conditions according to their visual effects.On the other hand,features are extracted for fine classification of video clips with rain and snow.The classification results on 249 video clips associated with different weather conditions indicate the effectiveness of the extracted features,by using C-SVM as the classifier.