arXiv:2212.00476v2 Announce Type: replace 
Abstract: Mobile stereo-matching systems have become an important part of many applications, such as automated-driving vehicles and autonomous robots. Accurate stereo-matching methods usually lead to high computational complexity; however, mobile platforms have only limited hardware resources to keep their power consumption low; this makes it difficult to maintain both an acceptable processing speed and accuracy on mobile platforms. To resolve this trade-off, we herein propose a novel acceleration approach for the well-known zero-means normalized cross correlation (ZNCC) matching cost calculation algorithm on a Jetson Tx2 embedded GPU. In our method for accelerating ZNCC, target images are scanned in a zigzag fashion to efficiently reuse one pixel's computation for its neighboring pixels; this reduces the amount of data transmission and increases the utilization of on-chip registers, thus increasing the processing speed. As a result, our method is 2X faster than the traditional image scanning method, and 26% faster than the latest NCC method. By combining this technique with the domain transformation (DT) algorithm, our system show real-time processing speed of 32 fps, on a Jetson Tx2 GPU for 1,280x384 pixel images with a maximum disparity of 128. Additionally, the evaluation results on the KITTI 2015 benchmark show that our combined system is more accurate than the same algorithm combined with census by 7.26%, while maintaining almost the same processing speed. Source Code: https://github.com/changqiong/Z2ZNCC.git

تم اقتراح نهج تسريع جديد لحساب تكلفة المطابقة باستخدام الارتباط المتقاطع العادي الصفري (ZNCC) في أنظمة المطابقة الاستريو المحمولة، وخاصة على وحدة معالجة الرسوميات المدمجة Jetson Tx2. تعمل هذه الطريقة على تحسين سرعة المعالجة من خلال مسح الصور المستهدفة بطريقة متعرجة، مما يسمح بإعادة استخدام حسابات بكسل بشكل فعال وتقليل نقل البيانات.

Se ha propuesto un nuevo enfoque de aceleración para el cálculo del costo de coincidencia mediante correlación cruzada normalizada a cero (ZNCC) en sistemas de coincidencia estereoscópica móviles, especialmente en la GPU embebida Jetson Tx2. Este método mejora la velocidad de procesamiento al escanear imágenes objetivo en un patrón en zigzag, permitiendo la reutilización eficiente del cálculo de píxeles y reduciendo la transmisión de datos.

Une nouvelle approche d'accélération pour le calcul du coût de correspondance par corrélation croisée normalisée à zéro (ZNCC) a été proposée pour les systèmes de correspondance stéréo mobiles, en particulier sur le GPU embarqué Jetson Tx2. Cette méthode améliore la vitesse de traitement en scannant les images cibles de manière en zigzag, permettant une réutilisation efficace des calculs de pixels et une réduction de la transmission de données.

A novel acceleration approach for zero-means normalized cross correlation (ZNCC) matching cost calculation has been proposed for mobile stereo-matching systems, particularly on the Jetson Tx2 embedded GPU. This method enhances processing speed by scanning target images in a zigzag manner, allowing for efficient pixel computation reuse and reduced data transmission.

Efficient stereo matching on embedded GPUs with zero-means cross correlation

arXiv:2512.06793v1 Announce Type: new 
Abstract: Real-time stereo matching methods primarily focus on enhancing in-domain performance but often overlook the critical importance of generalization in real-world applications. In contrast, recent stereo foundation models leverage monocular foundation models (MFMs) to improve generalization, but typically suffer from substantial inference latency. To address this trade-off, we propose Generalized Geometry Encoding Volume (GGEV), a novel real-time stereo matching network that achieves strong generalization. We first extract depth-aware features that encode domain-invariant structural priors as guidance for cost aggregation. Subsequently, we introduce a Depth-aware Dynamic Cost Aggregation (DDCA) module that adaptively incorporates these priors into each disparity hypothesis, effectively enhancing fragile matching relationships in unseen scenes. Both steps are lightweight and complementary, leading to the construction of a generalized geometry encoding volume with strong generalization capability. Experimental results demonstrate that our GGEV surpasses all existing real-time methods in zero-shot generalization capability, and achieves state-of-the-art performance on the KITTI 2012, KITTI 2015, and ETH3D benchmarks.

تم اقتراح شبكة جديدة لمطابقة الصور الاستريو في الوقت الحقيقي، وهي حجم ترميز الهندسة العامة (GGEV)، لتحسين التعميم في تطبيقات مطابقة الصور الاستريو، مما يعالج القيود المفروضة على الأساليب الحالية التي تركز بشكل أساسي على الأداء في المجال. يستخدم GGEV ميزات حساسة للعمق ووحدة تجميع تكلفة ديناميكية حساسة للعمق لتحسين دقة المطابقة في المشاهد غير المرئية.

Se ha propuesto una nueva red de coincidencia estereoscópica en tiempo real, el Volumen de Codificación Geométrica Generalizada (GGEV), para mejorar la generalización en aplicaciones de coincidencia estereoscópica, abordando las limitaciones de los métodos existentes que se centran principalmente en el rendimiento en dominio. El GGEV emplea características sensibles a la profundidad y un módulo de Agregación de Costos Dinámica Sensible a la Profundidad para mejorar la precisión de la coincidencia en escenas no vistas.

Un nouveau réseau de correspondance stéréo en temps réel, le Volume d'Encodage Géométrique Généralisé (GGEV), a été proposé pour améliorer la généralisation dans les applications de correspondance stéréo, répondant aux limites des méthodes existantes qui se concentrent principalement sur la performance en domaine. Le GGEV utilise des caractéristiques sensibles à la profondeur et un module d'Agrégation de Coût Dynamique Sensible à la Profondeur pour améliorer la précision de la correspondance dans des scènes non vues.

A novel real-time stereo matching network, Generalized Geometry Encoding Volume (GGEV), has been proposed to enhance generalization in stereo matching applications, addressing the limitations of existing methods that focus primarily on in-domain performance. The GGEV employs depth-aware features and a Depth-aware Dynamic Cost Aggregation module to improve matching accuracy in unseen scenes.

Efficient stereo matching on embedded GPUs with zero-means cross correlation

Was this article worth reading? Share it

Gyfted

X Headshot

MindStudio