Contact us

A new window will open A new window will open

Features : Visconti™ Family

Feature 1: Parallel execution of up to eight image recognition applications

This is an example of simultaneous, parallel execution of multiple applications.

The Media Processing Engine (MPE) is suitable for image recognition and other multimedia processing. It is a very long instruction word (VLIW) machine that can issue multiple instructions simultaneously. It consists of Toshiba's proprietary low-power 32-bit RISC core called a Media Embedded Processor (MeP) and either a coprocessor suitable for multimedia processing called Image Recognition VLIW Coprocessor 2 (IVC2) or its successor, IVC2 with FPU (IVC2F). IVC2 and IVC2F can execute one or two SIMD (single instruction stream, multiple data stream) instructions, which perform one or two operations on multiple data in parallel. The second-generation IVC2F can efficiently perform double-precision floating-point operations. The Visconti Family can execute multiple applications simultaneously* because of its multi-core configuration containing multiple MPEs.

* See the product specifications for details.

Visconti4 Block Diagram

TMPV760 Block Diagram

Visconti3 Block Diagram

TMPV752 Block Diagram

Visconti2 Block Diagram

TMPV750 Block Diagram

Feature 2: Image processing solutions that combine high speed and low power consumption

The Visconti Family contain image processing accelerators that perform the image processing required frequently in each step of image recognition at very high speed with low power consumption. The Visconti Family contain an affine transformation accelerator, which corrects any distortion of a fish-eye or wide-angle lens and performs point-of-view transformation. In addition, the Visconti Family contain filter accelerators, which perform noise reduction, edge detection, color space conversion and other image processing. The Visconti Family also include a histogram accelerator that creates histograms, and a matching accelerator that performs disparity calculation for a stereo camera system and motion detection. Furthermore, the Visconti2 Series (except the TMPV7504XBG) and the Visconti3 Series contain a HOG accelerator, which handles Histogram of Oriented Gradients (HOG) and/or Toshiba Co-occurrence Histograms of Oriented Gradients (CoHOG) features that are suitable for robust human recognition. These processors can detect both moving and stationary pedestrians in real time. The latest Visconti4 Series incorporates an Enhanced CoHOG accelerator that combines luminance-based CoHOG feature descriptors with color-based feature descriptors obtained using a newly developed technique. It delivers a significant improvement in nighttime image recognition.  

Image processing solutions that combine high speed and low power consumption

Feature 3: Toshiba’s original pattern recognition using multiple types of color features (Visconti4)

Toshiba’s original pattern recognition using multiple types of color features (TMPV760 Series)

In addition to the Toshiba-original luminance-based CoHOG, the image recognition accelerator contained in the Visconti4 processes four kinds of feature quantities: color-based CoHOG, Co-occurrence Histograms of Pairs of Edge Orientations and Color Differences (CoHED), Co-occurrence Histograms of Color Differences (CoHD), and Color Histograms. This accelerator provides high recognition performance even at nighttime and at scenes with less luminance differences between objects and the background. The Visconti4 realizes nighttime pedestrian recognition as reliable as daytime pedestrian recognition available with conventional vision systems.

Feature 4: General obstacle detection using 3D reconstruction technology (Visconti4)

The Structure from Motion (SfM) accelerator provides three-dimensional (3D) reconstruction, a process of estimating 3D structures (distance, height and width) from 2D image sequences using a monocular (i.e., single) camera. While the Visconti2 and Visconti3 Series support the detection of pedestrians, vehicles and other objects using only motion detection and pattern recognition, the SfM accelerator of the Visconti4 Series also makes it possible to detect general stationary obstacles such as fallen objects, fallen rocks and landslides without pre-learned object models.

Detection of Road Obstacles Using 3D Reconstruction Technique

This is an explanatory diagram of the general obstacle detection using 3D reconstruction technology.

* MIPI is a registered trademark owned by the MIPI Alliance.

* Microsoft and Windows are either registered trademarks or trademarks of Microsoft Corporation in the United States and/or other countries.

* Red Hat and Enterprise Linux are registered trademarks of Red Hat, Inc. in the United States and other countries.

* HDMI is a trademark or registered trademark of HDMI Licensing, LLC in the United States and/or other countries.

* MediaLB and MOST are registered trademarks of Microchip Technology Inc. in the United States and other countries.

* Arm and Cortex are registered trademarks of Arm Limited (or its subsidiaries) in the US and/or elsewhere.   

* All other company names, product names, and service names mentioned herein may be trademarks of their respective companies.

Contacts

If you have any questions, click one of these links:

Technical queries
Questions about purchasing, sampling and IC reliability
To Top
·Before creating and producing designs and using, customers must also refer to and comply with the latest versions of all relevant TOSHIBA information and the instructions for the application that Product will be used with or for.